Wikipedia talk:No original research: Difference between revisions

Proposed clarification: Combining indisputable facts is okay

I'd like to add the following clarification to "Synthesis of published material which advances a position":

“ Combining plain, indisputable facts in a way that results in another plain indisputable fact does not fall under the "advancing a position" part of the policy and is generally allowed. For example, is generally okay to combine the reliably-sourced and indisputably true statements "Bill Clinton was the 42nd President of the United States" and "Bill Clinton was one of two US Presidents to be impeached" to a new, non-sourced but indisputably true statement "The 42nd President of the United States was impeached." Care should be used to make sure that the supposed facts that are combined are not merely "not currently in dispute" but in fact beyond disputable. Likewise, care should be taken that the combining of the facts is clear and logical. When doing this type of combination, the editor must include the references to all source material and should explain the combination of facts in the edit log or article talk page. ”

This comes out of a discussion on Talk:Line of succession to the British throne#Request for Comment. I believe that User:Grover cleveland is mistaken in his interpretation but if he made the mistake, others will make the same mistake, so clarification is necessary. Comments? davidwr/(talk)/(contribs)/(e-mail) 01:24, 31 August 2008 (UTC)

I disagree. Like the much-argued WP:V point with regard to citing things like the sky being blue, any genuine case where this clause would apply would already be covered with WP:IAR and a "stop being silly" directed at whoever was trying to challenge the sentence. Therefore it's meaningless instruction creep at best, and at worst will give an obnoxious loophole for some wikilawyer or other. --erachima talk 01:35, 31 August 2008 (UTC)

I (not surprisingly) also disagree. First, the suggested change to the policy undermines WP:V. The touchstone for Wikipedia has always been verifiability, not truth. Wikipedia editors are not supposed to decide what are, and what are not, "plain indisputable facts". (Is "Bill Clinton was one of two US Presidents to be impeached" really "beyond disputable"? Suppose, for example, that new evidence turns out suggesting that Johnson wasn't really impeached). Second, as erachima suggests, the proposed change would completely vitiate WP:SYN: it would always be open for someone synthesizing information to claim that it is "indisputable", "obvious", etc. Would even the plagiarism example that is given on the main page still violate this rule? Third, if the resulting synthesized information is really germane and important to the subject of the article, some reliable source on the topic will include it, making the synthesis unnecessary. Grover cleveland (talk) 02:00, 31 August 2008 (UTC)

Given reliable sources that say "In the 1980 census, Smallville had a population of 1,234 people" and "In the 1990 census, Smallville had a population of 1,244 people" to say "between 1980 and 1990, official census figures show a population gain of 10"? This is the kind of "ignore all rules would seem to apply, but some people are dense, so we need to spell it out for them" situation I'm thinking of. To be more trivial, is it against Wikipedia no-synthesis rules to say 1,244-1,234=10 if that equation can't be reliably sourced? If so, WP:IAR applies.

In the case of "what if it turned out Johnson wasn't impeached," we'd be in the same situation as if someone did moot the whole point and there was a reliable source saying "The 42nd President of the United States was impeached." To paraphrase a late-20th-century movie referencing the Jewish Holocaust, some things are historical facts and beyond question. davidwr/(talk)/(contribs)/(e-mail) 04:10, 31 August 2008 (UTC)

The RFC referenced above is not clear about the sentence in dispute in Line of succession to the British throne. It's a pity, because it would help understand the issue, and possibly offer other avenues for resolution. Pcarbonn (talk) 08:33, 31 August 2008 (UTC)

Is there anything I can do to help clarify things? davidwr/(talk)/(contribs)/(e-mail) 13:33, 31 August 2008 (UTC)

The article Heights of United States Presidents and presidential candidates makes the claim that Richard Nixon was the nineteenth tallest president. There is no source which says that he was the nineteenth tallest president, but there are sources for the heights of each president, and when one puts these sources together in a table, it shows that Nixon was nineteenth tallest. Is that a violation of NOR? Noel S McFerran (talk) 16:51, 31 August 2008 (UTC)

Noel, I think that there are several problems with that article. First, the self-declared purpose of the article is "to compare the veracity of the folk wisdom about U.S. presidential politics that the taller of the two major-party candidates always wins or always wins since the advent of the televised presidential debate, since 1900, and so on" which not only seems a somewhat dubious basis on which to base a Wikipedia article, but is almost explicitly declaring the article to be an original research project. Second, it isn't clear where the information on each President comes from, so there seem to be WP:V issues. Did all the heights come from a single source, or were they cobbled together from a patchwork of sources? Third, it isn't clear that the height information on each President is comparable. For example, the article claims that Bill Clinton (6 ft 2.5 in) was taller than George Washington (6ft 2 in). Even assuming that we have a reliable source for George Washington's height, is there any evidence that it was accurate to within 1/2 an inch? Consider the possibility that Clinton's height is to the nearest half inch, and Washington's to the nearest inch (this seems quite likely: all the pre-Civil War presidents except for Jefferson have a height that is a whole number of inches, something which would be extremely unlikely by pure chance). In that case: it is incorrect to claim that Clinton was taller than Washington, since Clinton could have been 6 ft 2.3 in and Washington could have been 6 ft 2.4 in. Now we don't know that this was the case, but neither (without a cited source) do we know that it was not the case and that the heights are comparable. This kind of complication, even in an apparently straightforward list, is precisely the reason why we have the WP:SYN policy Grover cleveland (talk) 02:41, 1 September 2008 (UTC)

Pcarbonn, the dispute in Line of succession to the British throne is not over one single sentence, but rather over the nature of the entire article. The page is an attempt to identify and list, in order, every single person who is legally qualified to succeed to the British throne according to British law. This involves the following alleged violations of WP:SYN:

• Adding newly born children to the list directly below their parents, on the basis solely of a source saying that the child was born to his/her parents. There will usually be no reliable source saying that the newly born child is in the line of succession, or that they are at a particular number (e.g. 532nd) in the line of succession.
• Removing persons who become Roman Catholic, or who marry a Catholic (this is a feature of the archaic British succession laws). There will usually be no reliable source saying that "XXX has become Roman Catholic and thereby removed him/herself from the succession". In addition, the exact legal details of how and when a person becomes disqualified because of Catholicism are unclear and debated on the article's discussion page.
• Removing people who die from the list, solely on the basis of a report of the death. There will usually be no reliable source stating that "XXX has died and thereby removed him/herself from the succession".
• Adding/removing people from the list on the basis of their legitimacy (i.e. whether their parents were married at the time of their birth).
• Of course, every time a person is added to or removed from the list, this has the effect or decreasing or increasing the ranking of every person later in the list. Not surprisingly, these changes in the rankings are not directly supported by any reliable sources.

I hope this clears the nature of the dispute up, at least as regards WP:SYN (there are other issues as well). Grover cleveland (talk) 16:53, 31 August 2008 (UTC)

"Removing people who die from the list".

Now this is exactly the type of complaint I was referring to with the "stop being silly" comment above. Dead people cannot hold the throne, and seriously suggesting that as a case of WP:SYN violation makes your whole argument look more like a joke than anything else. You may have a point on some of the other parts though, I'll admit I'm not well versed in archaic British succession laws. --erachima talk 19:40, 31 August 2008 (UTC)

Grover, you oversimplify things. I assume your goal is the same as ours - to make a better Wikipedia. I also assume your goal is to make this article the best article it can be, and to make WP:SYN a useful tool to apply to all articles in Wikipedia, with the obvious exceptions where WP:AIR might apply. If I'm wrong, if you are trying to have the article deleted or restructured in a major way so that it is no longer what its title implies, the place for that discussion is the article talk page, not here.

I should probably do this on the article talk page, but in the interest of keeping the discussion in one place I'll address your items point by point here:

• "Adding newly born children to the list directly below their parents, on the basis solely of a source saying that the child was born to his/her parents. There will usually be no reliable source saying that the newly born child is in the line of succession, or that they are at a particular number (e.g. 532nd) in the line of succession."

• Rebuttal: I've already conceded that you have a case when there is the possibility that the parent is in the line but for whatever reason, say, unclear legitimacy or the possibility of Catholicism of the child, the child is not. If the parent is in the list at position 592, we should either assume that is true or place a "citation needed" tag on it. With the lack of such a tag, all that is needed to insert the child is 1) the fact the child is born, 2) the fact the child is Protestant or being raised Protestant, or more specifically, not Catholic, 3) the fact the child is legitimate, and 4) the names, gender, position, and exclusion-status of any older siblings. Newspaper reports typically give all information with the possible exception of religion. Since the Catholic church won't baptize infants when the parents won't raise him Catholic - something I could find a reliable source for at the nearest parish church - and de facto the won't Baptize a child who is being raised by non-Catholic parents, it would be absurd to argue that a child is Catholic if his parents are both known to be Protestant. Therefore, there is a much, much, greater chance of error and is much, much more likely to lead to an incorrect list if you exclude a newborn child who is known to be a child of someone in the list. This means either 1) the NOR/SYN policy is creating error and needs to be changed, 2) you are misinterpreting the NOR/SYN policy, or 3) the NOR/SYN policy is creating error and it is the will of Wikipedians that this is perfectly okay. I threw in #3 just to be mathematically complete, I doubt very much that is the case.

• 'Removing persons who become Roman Catholic, or who marry a Catholic (this is a feature of the archaic British succession laws). There will usually be no reliable source saying that "XXX has become Roman Catholic and thereby removed him/herself from the succession". In addition, the exact legal details of how and when a person becomes disqualified because of Catholicism are unclear and debated on the article's discussion page.'

• Rebuttal: You have a point in certain instances. However, there are certain instances when it is clear: If a person is baptized a Catholic and professes the Catholic faith, or he marries a practicing Catholic, it's clear he is out at the time of his profession of faith or his marriage. What is not clear is what happens to any children he had prior to that point. There may also be some ambiguity if there is no genuine profession of faith, such as a person whose spouse had been baptized Catholic but who had not practiced it as an adult, or who had converted to a Protestant or non-Protestant, non-Catholic denomination before marriage. My suggested change to WP:SYN would cover the unambiguous cases only.

• 'Removing people who die from the list, solely on the basis of a report of the death. There will usually be no reliable source stating that "XXX has died and thereby removed him/herself from the succession".'

• Rebuttal: Oh please, a reliable report that he is dead suffices. We can always revert if the report was incorrect. A reliable report that said "So and so died but he's planning on coming back to life so he is specifically not removing himself from the line of succession" would still result in his removal.

• "Adding/removing people from the list on the basis of their legitimacy (i.e. whether their parents were married at the time of their birth)."

• Rebuttal: The law assumes a child is legitimate and that he is the child of the mother's husband if the mother is married. I'm not sure if this is at the time of birth or the time of conception, and I'm not sure how this affects widows married at conception, but I'm sure British law handles this with clarity. There doesn't need to be a reliable source saying "so and so is legitimate" only that "so and so's mother was married at the time of the birth" or "... at conception" depending on British law. In cases where there is a claim of a secret marriage or a claim that the child is illegitimate despite appearances to the contrary, then specific reliable sources are needed. Otherwise, the marital status of the mother at the time of birth or conception is all that is needed. We can safely assume legitimacy or illegitimacy based on that fact and applicable British law.

• "Of course, every time a person is added to or removed from the list, this has the effect or decreasing or increasing the ranking of every person later in the list. Not surprisingly, these changes in the rankings are not directly supported by any reliable sources."

• Rebuttal: They are, however, supported by mathematics. While it is true that if someone is in the list and shouldn't be, or is not in the list and should be, everyone else's ranking is off by one, that is inherent in any list. If someone 40-odd down on the list died tonight, everyone below him would be one lower than they should be until that person was removed. If it took a couple of days for the news to be reported, then for 2 days there would be 1500+ entries that would be incorrect. So what? We do the best we have with the information we have.

The question remains open: Is my suggested change as worded what I intended: An explicit statement of what is already implied? Is making this explicit good or bad for Wikipedia? User:erachima has said that it is what I intended but adding it would open the door to Wikilawyering. I'm not sure what User:Pcarbonn is saying with respect to my suggested change, nor am I clear on what User:Grover cleveland is saying with respect to the proposed change. User:Noel S McFerran gives another list-order-synthesis-type example where WP:SYN either does not apply or if it does apply, Wikipedia will suffer as a result. I welcome further comments in the abstract - now that we have McFerran's presidential-height example the details of the British Throne example aren't as important.

davidwr/(talk)/(contribs)/(e-mail) 21:29, 31 August 2008 (UTC)

DavidWR: my initial attempt to answer Erichima's query has led into a full-scale debate which would belongs at Talk:Line of succession to the British throne. I'll respond to your points there, if I have time. I suggest that we keep detailed debates over that article out of this discussion page and stick to general principles (with specific examples where they illuminate the questions) here. Grover cleveland (talk) 23:29, 31 August 2008 (UTC)

Find an appropriate guideline to make any such suggestion. It is a common sense argument which should probably not be the subject of a full paragraph in policy. Strictly speaking, I'd suggest the two facts (1) Clinton was the 42nd president and (2) he was impeached should be simply stated. The process of comparing him to Johnson or counting how many presidents have been impeached are acts of interpretation. While this shouldn't require a source, certainly one could be found if some wikilawyering editor is making silly arguments against this "synthesis."--SaraNoon (talk) 22:30, 31 August 2008 (UTC)

I agree with the thinking of the original proposal. It might be obvious, but uncontentious original synthesis is very common on Wikipedia (indeed, every WP article is an original synthesis to some extent, otherwise it would almost certainly be a breach of copyright), and I don't think it would do any harm to mention this in the policy to stop the occasional wikilawyer in their tracks.--Kotniski (talk) 07:50, 1 September 2008 (UTC)

If an instance of synthesis is "uncontentious", then by definition nobody's complaining about it and it therefore doesn't need a rule about it. I think a lot of people miss the point that on Wikipedia, the policy pages are there to enumerate the very few things that aren't permissible, not the multitudinous ones that are. --erachima talk 09:16, 1 September 2008 (UTC)

Even if that's so, the policies should still be worded so that the definition of what's not permissible doesn't inadvertently include things that are actually permissible. I notice that there are people who raise a fuss about things that aren't really contentious, either for their own satisfaction at seeing the rules enforced at whatever cost, or to try to make a point about some other situation. I agree that the policies don't have to cover every possible eventuality, but the one raised here is far from rare and deserves mention IMO.--Kotniski (talk) 09:30, 1 September 2008 (UTC)

Kotniski, you say that every WP article is an original synthesis to some extent, otherwise it would almost certainly be a breach of copyright. With respect, I really don't think that's true. A typical Wikipedia article consists of a collection of claims from multiple different sources. However, each individual claim can (or ideally should) be backed by a single source. Simply bringing multiple different pieces of information together is not a synthesis if it doesn't result in any new claims. For example, consider the article about presidential heights that is referenced above. If I have source A that claims that Washington was 6 ft 2 in tall, and I have source B that claims that Clinton was 6 ft 2.5 in tall, then I can bring these claims together as

1. Washington was 6 ft 2 in tall. Clinton was 6 ft 2.5 in tall.

or even

2. Washington was 6 ft 2 in tall, and Clinton was 6 ft 2.5 in tall.

Neither of these formulations poses any WP:SYN problems. It is only if I bring these two claims together as

3. Clinton was taller than Washington

that there is a WP:SYN violation, since I have now made a new claim that can't be backed by any of my individual sources on their own (and it may not be a logically sound deduction from them -- see the discussion above). Now, the vast majority of Wikipedia articles do not do this. I invite you to point to specific examples if you disagree, but a typical well-sourced article merely rephrases and rearranges information that can be found in other sources. The worry that vast swathes of Wikipedia would violate WP:SYN is misguided: articles such as the Line of succession to the British throne or Heights of United States Presidents and presidential candidates are very atypical in this respect. Cheers. Grover cleveland (talk) 11:38, 1 September 2008 (UTC)

Yes, I agree that this kind of deduction is atypical (I had in mind the rephrasing and rearrangement, which a pedant might claim is original interpretation). I'm not saying that all logically valid deductions should be allowed - just that we should perhaps try to define (broadly) which are and which aren't. For example, trival things like imperial-metric unit conversions are presumably OK. --Kotniski (talk) 12:33, 1 September 2008 (UTC) PS In fact, I would basically like to make the essay WP:These are not original research into part of the policy (or at least a separate guideline).--Kotniski (talk) 12:48, 1 September 2008 (UTC)

If I can throw in my two cents, what you are talking about is syllogisms, equations, and other forms of logical deduction. deduction (done properly) is non-problematic, and simple deductions of the type you are talking about should never be a concern. the problem is that most people do not explicitly know how to deduce correctly. most people get the general idea, but often make formal errors that render the deduction invalid, or make inferences that they can't distinguish from deductions. both of those errors are accidental violations of wp:SYN. for instance, given reliable sources that say Washington was 6'2" tall and Clinton was 6'2.5" tall, it would be perfectly proper to say that Clinton was taller than Washington, but it would be improper to say that Clinton was larger than Washington (since 'larger' accounts for other things than height). if you're just looking for a guideline, anything that can be explicitly formulated as a valid deduction (one that doesn't violating the principles of logic) should be fine; anything that does violate those principles shouldn't be allowed. --Ludwigs2 23:31, 1 September 2008 (UTC)

As others have noted above, in non-contentious articles, some common sense synthesis is common. But since it is common sense and the facts are clear, no one cares. Policy exists more for the sake of contentious articles than non-contentious articles, however. Specifically, the no synthesis rule exists, and is often misinterpreted and abused when passionately held points of view come into conflict. In such articles, wikilawyers will try to block even statements similar to your example that Clinton was taller than Washington (if it goes against their POV) and demand that the contributing editor must find this conclusion in a reliable secondary source. Then if a source is found, it is quite possible that a strongly biase editor will still argue that the source you provide it is not sufficiently reliable, or that since they can't verify it (because they don't want to go to a real library) it should not be accepted. This is all disruptive nonsense, but it points to the difficulties involved in contentious articles. The practical rule which follows is that if someone is making your life difficult, take extra great care to state A and B and do not even hint at C. Let the reader draw from the bare facts, "Source A reports Clinton is this height." "Source B reports Washington is this height." And leave it at that. Unfortunately, your more contentious editors won't even want to allow that, but there you have it. Rather than fight for the right to a simple obvious comparison, if people complain focus on the appropriateness of stating the facts with no comparison statement.--SaraNoon (talk) 02:36, 2 September 2008 (UTC)

all sources have a certain degree of error , and we do not aim for complete scientific accuracy. A difference of fractions of an inch in people's height is only significant if the dispute over that particular fractional difference is in fact controversial (which is not all that rational in the first place, as a person's height changes slightly over time, even over the 4 or 8 years of a presidential term.) There has to be some common sense on what is controversial (obviously, showing that something is indeed controversial in the sources is sufficient for that--but then it provides sources). DGG (talk) 10:19, 2 September 2008 (UTC)

• Oppose the clarification proposed. Much of the discussion in support of the proposal offers examples of what appear to be original research to advance new claims. This encyclopedia doesn't do research analysis of the heighth of the US presidents, it doesn't do genealogy research to determine lines of succession. This kind of stuff is what the policy is intended to reign in. If other sources are found in which the individuals' placements in lines of succession are given, we cite them. If studies have been conducted ranking presidents by heighth, we cite them. But we don't originate those kinds of studies. The current wording of the policy is sufficient for areas which are basic common sense. But studies like those I just mentioned are not, which entail all kinds of complex factors which must be dealt with in terms of the data used and/or making justifiable analytic use of the information. Professor marginalia (talk) 15:21, 2 September 2008 (UTC)

• I respectfully disagree with your interpretation of NOR, at least with respect to lists. To use hypothetical example: Suppose in 1945 someone published a list of the past and present governors of each state listed by state and within each state alphabetically by last name. Suppose that as recently as last year reliable sources provided up-to-date alphabetic lists for some of the states. Suppose furthermore that there were already reliable sources that listed each and every state governor since 1948 with their last name spelled correctly. In my opinion, it is not OR to create an article or set of articles that contain and up-to-date list the state governors with their name sorted alphabetically. Such an article might be deemed unencyclopedic but it would not be OR. The "list of heights" and "list of people in line for the British Throne" articles are substantially similar to this case: They take existing, verified information and merely apply existing, verified rules to the data to create a list. This is no more original research than sorting all Wikipedia users by username given the same list sorted by date of registration (the verifiable data) and a rule for sorting alphabetically (the verifiable set of rules). Yes, there are a few entries in the line of succession to the British Throne where the underlying facts are in dispute, but those can be handled individually through tags such as "dubious" and notations in the list that the status of the individual is not clear. The clarification to WP:NOR is needed because at least one editor is claiming that "a implies b, b implies c, therefore a implies c" and similar easy-to-follow logical arguments constitute original research. My concern is other editors may make the same error. Elsewhere in this discussion, someone recommended that Wikipedia:These are not original research be changed from an essay into a guideline. That might solve the problem without requiring a change in the policy document. davidwr/(talk)/(contribs)/(e-mail) 15:55, 2 September 2008 (UTC)

As to the last suggestion, I say, No, no, no, no, perish the thought, NO!. There are other wiki's in the project where original research has a place. The encyclopedia is an assembled look-up reference, a compendium of general knowledge. If some wikipedian is the first to come up with this knowledge, good for him or her, but it doesn't qualify as the "general knowledge".

As to the former discussion, the article on presidential height has a topic sentence, "The table above was created to compare the veracity of the folk wisdom about U.S. presidential politics that the taller of the two major-party candidates always wins or always wins since the advent of the televised presidential debate, since 1900, and so on." I say, reign it in folks. Start a blog, publish in a journal, write a book, take it to another wiki. But this is way off target for wikipedians to be doing independent research writing articles that try to investigate or debunk folk wisdoms. Professor marginalia (talk) 16:18, 2 September 2008 (UTC)

Question: Does this proposal mean that it would be acceptable to combine indisputable facts that also respectively result from the combination of indisputable facts? For example, if "A", "B", "X", "Y" are indisputable facts, would it be acceptable to combine "A" with "B" to form "C", and "A" with "Y" to form "Z", and then to combine "C" with "Z" to form "W" (Of course, all the sources that verify the individual facts "A", "B", "X", "Y" would have to be used to verify the resultant fact "W")? This process of combining indisputable facts could go on further. I'm wandering if this could become an issue. --Phenylalanine (talk) 12:08, 3 September 2008 (UTC)

I hold the view that Line of succession to the British throne is engaging in original research unless the order of succession we give is verifiable from reliable sources. I do not support loosening the policy that we have. I do support shortening the disputed article to only include what is properly sourced. If there is no source for some successors, why should we care to document it ? Pcarbonn (talk) 15:49, 8 September 2008 (UTC)

As long as conclusions reached from legal tautologies are allowed and extremely likely "obvious" conclusions are allowed with a notation. For example, when Prince William has a baby, as soon as it's reported that he will be baptized Protestant, it becomes a fact that he meets the requirements to be in the list even if there is no reliable source saying he is now in the line of succession. The same logic applies all the way down the list: If the parent is in the list, the child is legitimate, the child is Protestant (or specifically, not Catholic), then the child is in the line. End of story. Likewise, if the parent is in the list, the child is legitimate, and both parents are known to be Protestants, it is an almost-certainty that the child will not be baptized a Catholic in infancy, and therefore, it is fair to insert him in the line with a notation that we only know with 99.999% certainty that the child is not Catholic. It's been claimed that the vast majority of the names are found in a particular book or a particular online source, and there is some effort to audit this and tag the names with ^R and ^W. As far as people who are excluded, the reasons for keeping them is to prevent them from being re-added by someone who finds them in a reliable source that pre-dates their exclusion, usually due to converting or marrying a Catholic. Where there is truly no source that the person is in an unbroken legitimate line of non-Catholics to someone who is in the list, AND there is no source that specifically says the person is in the line, then he should be removed or, if his mother or father is in line, marked "excluded." In other words, you need an reliably-sourced unbroken non-Catholic legitimate line or you need a specific source saying X is in the line and where. davidwr/(talk)/(contribs)/(e-mail) 18:57, 8 September 2008 (UTC)

I have reverted the insertion of this footnote n°2 as there were no consensus on this topic (e.g. it is opposed by User:Professor marginalia and User:erachima and myself).Pcarbonn (talk) 20:55, 23 September 2008 (UTC)

WP:SYN should be taken with a grain of salt (at least) for math articles

“ "A and B, therefore C" is acceptable only if a reliable source has published this argument in relation to the topic of the article. ”

For math articles editors engage all the time in the above type of argument when it is "trivial enough". Asking editors to provide a citation for every implication of that kind, no matter how trivial, is patently absurd. It's worth quoting from a Wikipedia critic in this case:

“ stringent rules, spotty enforcement. ”

The undersigned, VasileGaburici (talk) 11:34, 9 September 2008 (UTC)

May I suggest also reading WP:V. The community of mathematicians and math buffs has shown itself very capable of identifying the more serious cases of original synthesis of mathematical formulas and non-standard transpositions of formulas. In general they tend to know the difference between a useful illustration of a previously published formulaic approach on the one hand, and an original formula or interpretation of a formula on the other hand.
..... To the extent that your assertion is that numerous articles deviate from the core content policies, which involve an interactive triad of WP:NPOV, WP:V and WP:NOR, then the point is granted. The main reason for this, of course, is that anyone can edit the encyclopedia and Gödel's theorem kicks in big time. For an approach where more restraint is exercised in allowing content, today there increasingly are other wikis available, e.g., Citizendium, which try to keep enforcement more stringent by setting predetermined limits on contributions, along with a rapidly growing list of special interest wikis whose proliferation rate also vastly exceeds the world's ability to fully vet them (speaking of (Gödel). But on articles considered important enough by the math community to double-check each other's contributions, the content policies tend to work fairly well on the whole.
..... If the assertion above is, however, that "A and B, therefore C [requires a reliable source for the statement]" means that editors of math articles can't do standard calculations in a math article, then it's an overly strict interpretation of the statement in WP:NOR rather than an overly stringent editorial policy. For similar reasons, in non-mathematical articles the rule is not in general expected to mean "quote everything exactly" but rather that the point being made in the WP article will be recognized by those conversant in the language of the article as consistent with the point(s) made in the relevant reliable sources. ... Kenosis (talk) 14:31, 9 September 2008 (UTC)

Indeed, math editors routinely remove original research from math articles, especially novel theories. In the end, only editorial discretion can distinguish truly original research, which is not desired, from simple restatements of well-known facts - because it takes a human to survey the literature to determine whether something is novel. The scientific citation guidelines describe the goals we are aiming for. Of course many articles are not there yet. — Carl (CBM · talk) 14:49, 9 September 2008 (UTC)

I agree with facts stated by User:Kenosis and User:CBM, and in particular with the assertion that the math-interested community on Wikipedia has done good job of distinguishing between original presentation and significantly new claims/results. If you read the comments above this section however, you'll see that there are editors that desire a strict enforcement of the WP:SYN rule, and they consequently oppose any amendments to it, even when those amendments would do little more than codify existing practice and guidelines (like WP:SCG or more generally WP:NOTOR). Thankfully those editors seem to stay away from math articles... The WP:SYN supporters seem think that it's the last best hope from keeping triviality and fringe theories out of Wikipedia by applying some rule mechanically. As Kenosis and CBM pointed out above, this mechanical approach is not practical. So, keeping the WP:SYN rule on the pedestal but the clarifications in the shadows simply provides low hanging fruit for wikilawyers; this was already pointed out by User:SaraNoon and User:Kotniski above. Others argued above that any clarifications would increase the amount of wikilawyering. I find this argument disingenuous; a rule that needs to be often ignored (as impractical) has more potential for wikilayering than a rule that requires fewer overrides via WP:IAR or WP:DICK. VasileGaburici (talk) 15:59, 9 September 2008 (UTC)

Agreed that it's low hanging fruit for "wikilawyers". So is much of WP:NPOV and WP:V, along with many of the statements in WP:Consensus. This "low-hanging fruit" has arisen out of a half-decade of community experience that began with much more generalized statements of the core content policies. The more generalized statements turned out to be equal fodder for argumentation, except that the argumentation tended to be less focused on particular parameters of editorial policy. Lately, though, we see a lot of prominent examples of attempting to persuade other users that more specificity is needed with various aspects of these wiki-wide content/editorial policies. It has the interesting characteristic of advocating yet more low-hanging fruit under the reasoning that there's too much low-hanging fruit. (perhaps see also the essay on WP:CREEP, the essay and the phenomenon itself quite well known to experienced WP users.)
..... Many, perhaps even most, of the users advocating such changes under the auspices of "specificity", "clarity", etc.,, turn out upon close examination to be advocating a substantive change in the policy tensions among WP:NPOV, WP:V and WP:NOR, rather than merely trying to reduce conflicts and misunderstandings. In other words, simply put, many of those advocating change to WP:NOR in particular, as well as certain aspects of WP:NPOV and/or WP:V, are actually trying to change the policy to suit their preferences about how to edit this encyclopedia. Which is, of course, quite human.
..... Please remember that there is a tension among the core content policies that on a local article level must be resolved by consensus. Thus we see as a matter of course many of the various human approaches that commonly are brought to bear in a consensus process, including but not limited to: "lawyering", nitpicking, "nonspecific emotionalism", truths, partial truths, stretched truths, half-truths, truthiness, verisimilitude, the occasional outright falshood, ad hominem or personal attacks disguised as civility, ad hominem or personal attacks not even bothering to be disguised as civility, groupthink, excessive argumentation, peacemaking, peacemaking that turns out to be fruitless, feigned peacemaking, highly conditional attempts at peacemaking, countless logical fallacies both formal and informal, misplaced accusations of logical fallacies both formal and informal, excessive logical positivism, etc. etc. As to the last of these, It might be useful to keep in mind that these are editorial policies that must strike a balance between flexibility and clarity/consistency, as well as must attempt to maximize the neutrality and reliability of reporting on various topics across a very wide range of topic areas, while also being available to be edited, within certain very approximate and debatable limits, by anyone on earth. And, of course, what's being argued about here, fundamentally, is where that balance can best be "centered" so that it works reasonably well across what has turned out to be a very, very wide wiki.
..... As to the wikilawyering-reduction arguments, we see this argument being brought to bear fairly often on talk pages of the three core content policies. I wouldn't make too much of it, because there's also WP:Ignore all rules to balance that out. And that balance involves a consensus process too-- surely it's enough to drive a positivist to drink. ;-)... Kenosis (talk) 17:28, 9 September 2008 (UTC)

Clearly from the discussion above WP:SYN doesn't seem that useful for math articles. So, can anyone provide some clear cut examples where WP:SYN, in particular the "A and B, therefore C" is acceptable only if a reliable source has published this argument in relation to the topic of the article rule, was used to prevent some ORish content that could not be refuted otherwise? In other words, is the claimed necessity of this very specific rule supported by use cases? VasileGaburici (talk) 20:10, 9 September 2008 (UTC)

As I indicated and CBM also indicated just above, it's used all the time to knock out original formulas, original interpretations of formulas, and original syntheses of formulas. As you said at the beginning of this thread, enforcement is far from perfect. ... Kenosis (talk) 21:33, 9 September 2008 (UTC)

No offense Kenosis, but you have no contributions to math articles that I could find. So, how do you know WP:SYN been used the way you claim? VasileGaburici (talk) 23:04, 9 September 2008 (UTC)

No offense taken. I'm familiar enough with the general trends to know that what I've said is indeed a reasonable summary of what has in fact occurred across the wiki, including math articles. Among other things, I was involved in the article on entropy for a time, have had a number of reasons to briefly check in on articles using mathematical formulas in support of premises contained in a fair number of scientific and technical articles including those in my real-life work, which I intentionally do not edit on the whole, and the issue of WP:NOR's application in math articles and other highly technical articles came up repeatedly in extended conversations on this project talk page and related ones through the latter part of 2007, during which time I was able to ascertain that the same principles do indeed apply to math articles within their own language of expression. But, your point at present appears exactly the same as when you started this thread, so please don't feel obliged to take it from me.
..... May I inquire, what are you advocating here w.r.t. math articles? That they be explicitly exempted from WP:NOR? From some particular clause in WP:NOR? Or merely proof that WP:SYN has also worked w.r.t. math articles per se? (W.r.t. the last, experience has taught that not only can I not answer this to the satisfaction of all, but that no one can answer this to the satisfaction of all.)
..... Incidentally, I happen to agree, as should be obvious by now, with your opening statement: "WP:SYN should be taken with a grain of salt (at least) for math articles" In other words, it's generally not advisable to apply it too literally and strictly to every minute detail but is advisable instead to keep it in perspective with the rest of WP:NOR and with the other two core content policies and WP:Consensus. ... Kenosis (talk) 23:31, 9 September 2008 (UTC)

I'm advocating exactly what I've put in the heading of this section "WP:SYN should be taken with a grain of salt (at least) for math articles". I haven't made any statements about WP:NOR and math, other than to agree that common sense has prevailed in practice. Let's keep the discussion, which is getting pretty verbose already, focused on the topic at hand, which is WP:SYN. In particular, in this subsection we're debating whether WP:SYN has any merits for math articles. Based on the things said so far, I think that a blanket de facto exception has been used in practice. You seem to be the only one claiming that WP:SYN is useful for math, and gave a physics article as example. I looked at your edits, and at the history of that article back to 2006, but saw no mention of WP:SYN.
............ BTW, the WP:SYN text draws the wrong conclusion for the Jones example that "it expresses the editor's opinion". A syllogism like Modus ponens is not an opinion, certainly not in the dictionary sense of the word ("a personal belief or judgment that is not founded on proof or certainty") Looking at the footnote of that page, it seems that someone was way too eager to codify Jimmy Wales's disapproval of historical theories, so we ended up with this rule (and dubious example). In order to avoid Wikipedia being filled with P=NP proofs and the like, it would be far more reasonable to simply write that significant new claims/results should be first published elsewhere. But WP:OR is already stating that, so WP:SYN rule has dubious usefulness at best. I see no use for it in math articles. If you have some concrete evidence that it helped somewhere, please provide it, but please no more vague claims like "I've seen it used somewhere". VasileGaburici (talk) 00:27, 10 September 2008 (UTC)

VasileGaburici, leaving aside various interpretive issues such as, e.g., your last statement ("... but please no more vague claims like 'I've seen it used somewhere' ", which I trust was drawn from somewhere other than what I've written above), I think I now understand somewhat better what you're asserting w.r.t. this editorial policy. In response to your assertion that this conversation has gotten "verbose" and your request to stay "focused on the topic at hand, which is WP:SYN", I'll try to be as brief as I think is reasonably possible here.
,,,,, We both agree with the statement "WP:SYN should be taken with a grain of salt (at least) for math articles". I say "we agree" because you made this statement, and I said I agreed with this statement.
..... WP:SYN was not originally a part of WP:NOR, but is a clarification and extension of WP:NOR, integrated into WP:NOR by consensus and specifically designed to deal with situations where folks synthesize original ideas based upon multiple reliable sources in order to arrive at an original conclusion and then say, essentially, "but that's not OR". To which the WP community responded, essentially, "No, it is OR, a particular kind of OR which we will call "original synthesis". It thus is a necessary part of WP:NOR, based upon community experience with WP:NOR. I don't think we need an exception specifically for mathematics articles, because that's for the community of, as you've called it the "math-interested" to work through via WP:Consensus. And if the math-interested community has an adequately strong consensus that the premiss that, say, articles in Category:Mathematics are excluded from WP:SYN, I imagine it will become understood by the broader WP community that maths articles are sufficiently problematic to merit specific exclusion from WP:SYN.
..... My offhand impression, speaking as just one WP user, is that a case for such an exclusionary clause has yet to be effectively made with sufficient strength to persuade the WP community that the general expectation w.r.t. WP:SYN cannot be handled solely by a reasonable application of WP:Consensus and WP:Ignore all rules on an article-by-article basis, or subcategory-by-subcategory basis, or even as a "meta-category" such as "Mathematics". ... Kenosis (talk) 03:14, 10 September 2008 (UTC)

I agree that one should not stick to wiki rules if sticking to such rules leads to absurd situations. I explained here that for scientific subjects we need the following guidelines for editors:

• Editors of a wiki article on a scientific subject should be experts in that subject. If you make substantial edits to such a article, you should be absolutely sure that you have completely mastered the subject you are writing about.
• Wikipedians are encouraged to engage in technical discussions on the talk page. Such technical discussions may be used to improve the article.
• Detailed arguments/derivations don't have to be sourced if included to make statements verifiable. In that case they are not included as encyclopedic content whose veracity should be verifiable to non-experts who are not able to understand the derivation/argumentation.

Count Iblis (talk) 22:42, 9 September 2008 (UTC)

errr... without disagreeing with your intent, doesn't your first point totally defeat the purpose of an openly-editable encyclopedia? In fact, it almost seems like an incitement to using wikipedia for original research (in the best sense of the word, mind you, but still...) --Ludwigs2 01:21, 10 September 2008 (UTC)

The points are proposed as guidlines, similar to the COI guideline. Anyone can edit wikipedia, but when someone joins wikipedia we should ask them politely that when they make substantial edits to a technical article they are experts about that subject. They don't have to be professors, they can be self taught experts. It doesn't matter. We also point new editors to the COI guidlines for editing, so I don't see anything wrong with this.

As I explained here, we should look at actual problems in wiki articles and then make rules that address those problems. The problems I encountered occured in some thermodynamics articles. What I noted was that the usual wiki rules for NOR, and Verifiability failed. From my experience it would always have taken an expert to spot and correct the errors. The articles should have been edited by experts in the first place. Count Iblis (talk) 03:10, 10 September 2008 (UTC)

This proposal is obviously not going anywhere because the opponents (obvious from the discussion above who they are) seem to have an infinite commitment to filibuster with no concrete evidence to back up their claims. Here's the gauntlet: find one mathematician (verifiable as such, and there are plenty on Wikipedia) to support WP:SYN for math articles. Find one logician (verifiable as such) to agree that the syllogism about Jones from the WP:SYN text is indeed opinion.
.....The proponent(s) of WP:SYN think that mechanical application of a rule like WP:SYN is a substitute for editor's knowledge for what "new" and "significant" mean in a given field. This attitude comes close to the Tyranny of Ignorance incident.
.....There is however a larger problem: legislative conservatism. Some assume that just because a rule is there, it must be golden. There's no mechanism in Wikipedia to survey how a policy has been applied and whether it has been useful, so once !legislation passes it becomes very difficult to abrogate or amend. Some of the guidelines (WP:SCG and WP:NOTOR) seem to exist specifically to contradict silly rules like WP:SYN. It is obvious to me that under these circumstances the Cambrian explosion of rules and guidelines, observed in this peer-reviewed study is bound to continue. So I won't be taking part in this discussion or similar anymore; I can spend my time in much more productive ways. VasileGaburici (talk) 11:40, 10 September 2008 (UTC)

(←) The main role of WP:SCG (scientific citation guidelines) is to complement (not replace or supersede) WP:V. It doesn't directly speak to WP:SYN. The thing to keep in mind with WP:SYN is that it only forbids novel synthesis. Synthesis that reflects widespread understanding in a field is perfectly fine. The more common difficulty is with sourcing, but in practice things work out OK.

Remember that the original motivation for NOR was to have a concrete written policy to apply to crank theories in physics and other science areas. Especially in the case of fringe theories critical of the overall consensus in a field, WP:SYN is a useful rule for science and math articles. — Carl (CBM · talk) 13:36, 10 September 2008 (UTC)

RE "Synthesis that reflects widespread understanding in a field is perfectly fine." This is extremely consistent with my understanding of WP:NOR and WP:SYN. ... Kenosis (talk) 15:35, 10 September 2008 (UTC)

I think that WP:SYN should be rewritten such that it becomes clear that the statement "A and B therefore C" refers to statements A, B and C that are not easily reducible to elementary logic. WP:SYN was motivated not by problems in science articles but more by problems in politics articles. A and B could be statements made by some politician at different times and the conclusion that C is true could well be conditional on the precise context A and B were assumed to be valid, or the politician could have changed his opinion about A when he said B. So, of course, you need a source which says that that politician backs statement C.

But if the truth of "A and B implies C" is a matter of elementary logic, accessible for people who can understand the article in question, then you don't need a source. It is already verifiable, a source would add nothing to the verifiability of C. Count Iblis (talk) 16:13, 10 September 2008 (UTC)

I would support this kind of brief addition to the existing language--with emphasis on the word "elementary". Perhaps a statement such as that proposed by Count Iblis in his edit summary to the effect that "A and B therefore C" refers to statements A,B and C that are not easily reducible to elementary logic" or "A and B therefore C" does not refer to statements A,B and C that are uncontroversial and easily reducible to elementary logic" ... Kenosis (talk) 16:30, 10 September 2008 (UTC)

I took a bit of a risk, and inserted a footnote speaking to this point (here). I trust this is not too radical a statement, but fully expect I'll get corrected quickly enough if I'm in error here. ... Kenosis (talk) 17:33, 10 September 2008 (UTC)

I think that's an excellent addition. I'd even support taking it out of the footnote and making it part of main text (it's a good preventative for a lot of silly arguments...) --Ludwigs2 22:29, 10 September 2008 (UTC)

RE "I think that's an excellent addition": Thank you for saying that. I think direct credit is due to Count Iblis for stating it in such a way that it seemed to make immediate sense, though nothing against anybody else. Thanks to everybody for what I considered to be a fairly interesting debate about a set of important editorial-policy issues that comes up every now and again. ... Kenosis (talk) 17:48, 13 September 2008 (UTC)

It's been abut a week, and I haven't seen any objections to this brief footnoted clarification thus far. Hopefully Count Iblis' suggestion will come to be regarded as useful clarification of WP:SYN both for math articles as well as elsewhere around the wiki. ... Kenosis (talk) 16:06, 21 September 2008 (UTC)

Policy clarification proposal

Per discussion regarding my previous proposal, I believe the following text or information should be added to the NOR policy. For an explanation of my rationale and specific examples of how this proposal differs from the current NOR policy wording, see this discussion.

“ When adding content to a Wikipedia (WP) article, the Wikipedia:No original research policy states that you must cite reliable sources that (1) provide information directly related to the topic of the article, and that (2) directly support the information as it is presented.

As a general guideline, you should also ensure that (3) the reliable sources which you cite directly relate this information to the topic of the WP article, and in the same way that the information is presented in the WP article. For example, if the information added serves to criticize the subject of the WP article, then you are required to cite sources where the information is clearly presented as a criticism of this topic. Whether the information added serves to contrast certain aspects of or statements about the subject-matter with opposing or mainstream points of view, or to support such statements or aspects, or even to point out aspects related to the topic of the WP article which happen to cast the subject-matter in a more favorable or unfavorable light, the same principal applies.

When developing the context for a subject described in a WP article, if no sources can be found that comply with the above guideline, you may occasionally cite reliable sources that provide information relevant to an aspect of that subject, but which do not directly relate that information to the topic of the WP article. However, to avoid breaching the Neutral point of view policy, you should never do this when adding information that may misrepresent any aspect of the subject or cast it in a more favorable or unfavorable light, for example if the information supports, discredits, highlights or downplays any aspect of that subject.

”

--Phenylalanine (talk) 05:22, 6 September 2008 (UTC)

The "how it differs" link doesn't work.

I don't support this. Raw foodism should be able to list the well-documented dangers of food poisoning related to raw milk, even though the CDC's webpage on the dangers of unpasteurized milk doesn't specifically say, "By the way, this apply to people that don't ever eat cooked foods just as much as it applies to people eating cooked foods every day." If the connection is patently obvious, your personal interpretation of the author's original intent (original research if I ever saw it, BTW) is unimportant. WhatamIdoing (talk) 20:06, 6 September 2008 (UTC)

Thanks for weighing in WhatamIdoing, I fixed the link. If the connection is patently obvious, then we are likely to easily find sources that make that connection, which we can then use, and if no readily available sources can be found that make the connection, then the connection probably isn't that obvious after all, and should not be mentioned in the article. I agree that the food safety issues are obviously relevant to the healthfulness of Raw foodism, but I can imagine some cases where what appears obvious at first glance really isn't after further examination. Since we can avoid making this kind of mistake by relying on the sources to make the connections for us and since it should presumably be easy to find information on food safety issues in sources that make the connection, I think we really have nothing to lose by following this approach, and, IMO, a lot to gain by it.

Furthermore, there is, IMO, a strong case for avoiding sources that don't make these sorts of connections in cases where the said connections are as obvious: belief "A" is based on a number of arguments, including "B"; research pertaining to "B" is therefore relevant to "A". In such as cases, I believe it would be inappropriate to add information on "B" in an article on "A" if the source cited did not directly relate research regarding "B" to belief "A", and I am not the only one to believe this. The current policy text does not prohibit such cases. One reason to avoid this, is that, if we follow the approach you recommend: "If the connection is patently obvious, your personal interpretation of the author's original intent is unimportant.", then there would be no way of accurately gauging the balance of the total body of reliable sources on a topic, e.g. Raw foodism, and so, it would be easy to not present the information in roughly the same proportion as it appears in reliable sources on the topic, e.g. Raw foodism, as required by the NPOV policy (see this comment).

For a specific example of this situation, see Raw foodism#Criticism and controversies. In that section, the following statements would constitute original research by the OR definition I suggest above, but not necessarily according to the current policy definition: Most other anthropologists oppose Wrangham, contending that archeological evidence suggests that cooking fires began in earnest only 250,000 years ago, when ancient hearths, earth ovens, burnt animal bones, and flint appear across Europe and the middle East.[38] This stance of Wrangham's, re cooking leading to bigger human brains, [can be] contrasted with several studies showing that average human brain-size has actually decreased in the last 35,000 years by 11%.[39][40][41] If you click on the refs, you'll notice that none of the sources discuss/mention/describe the concept of Raw foodism. None of the references for these two statements directly relate the information to the concept of Raw foodism. The first statement is verified by a source which refers to (i.e. criticizes) Wrangham's arguments suggesting that cooking played a major role in human evolution, but without mentioning (i.e. directly criticizing) Wrangham's criticisms of the concept of Raw foodism which are based on a number of arguments including the arguments in question which suggest that cooking played a major role in human evolution. The second statement is verified by sources that don't even mention Wrangham. Nevertheless, the editor who included these two statements considered that they were both directly related/relevant to the topic of the article (presumably because these two statements serve to contextualize Wrangham's criticisms of Raw foodism).

For another example, in a WP article on physics, such as "General relativity", would it be acceptable to add information that contextualizes this theory but that was not intended to do so in the source(s) cited, for example, recent research in physics that you believe calls into question Einstein's theory and that does indeed "relate to the Einstein's claims", but which the source(s) cited to not relate to the Einstein's claims. Such information would quickly be considered OR in the "General relativity" article and speedily removed. The "Raw foodism" article is no different in this regard (since it refers to or is based on a point of view, just like Einstein's theory of General relativity). We should adhere to the same standard that would be applied in the "General relativity" article.

On a final note, to interpret the intent of an author is not original research when that intent is clearly spelled out. When the intent is not clear, the source should not be used for that purpose. I don't see a contradiction between OR rule number (2) "directly support the information as it is presented" and the guideline I am suggesting, rather I think they go hand in hand. Cheers. --Phenylalanine (talk) 23:09, 6 September 2008 (UTC)

Phen, I understand that you have some concerns, but I don't think that this is an appropriate way to address them. I think this proposal creates more problems than it solves.

Criticisms of Raw foodism is a useful example. The Wrangham piece is poorly written IMO, but it's not actually original research. Some pro-raw people believe that the human body is better adapted to a raw food diet than to a cooked food diet. This is "verifiable" -- in the sense that you and an internet search engine could verify that some people do believe this -- but simply not sourced in the article yet (last I looked; it will presumably appear at some point in the list of rawist beliefs).

So we have pro-raw people saying that humans have evolved to thrive best on a diet of primarily or exclusively raw foods, and a scientist that says that humans have evolved to eat primarily or exclusively cooked foods (and specifically that a diet of primarily or exclusively raw foods does not provide sufficient food energy). This is very basic point-counterpoint. I do not think that it constitutes original research. WhatamIdoing (talk) 16:28, 8 September 2008 (UTC)

But WhatamIdoing, I never said that Wrangham's work was original research in the context of the Rawism article, indeed his piece does refer to raw foodists and their beliefs. What I illustrated above was studies used out of context as counterpoints to Wrangham's piece. Could you take another look? What problems will this proposal raise? I appreciate your feedback. --Phenylalanine (talk) 22:49, 8 September 2008 (UTC)

Whether or not basic facts espoused by Wrangham are accurate/accepted is obviously relevant: How else will the reader be able to decide whether to accept Wrangham's view, if we don't provide any context at all? Context can be provided by writing "Wrangham's theory is rejected by research Joe Smith" or by providing concrete information like this author has chosen (noting that I'd want to look long and hard at the specific sources and wonder why 35,000 years was chosen as the relevant measuring point...)

Your proposal basically outlaws common sense and context. WhatamIdoing (talk) 05:30, 9 September 2008 (UTC)

Please see Vassyana's comment and Someguy1221's remarks about my previous proposal. Based on these comments, I conclude that my current proposal would be the only interpretation of "original research" that could be consistent with the WP:NPOV policy. --Phenylalanine (talk) 12:08, 11 September 2008 (UTC)

• User:Phenylalanine's proposal won't do because it requires that we are able to divine the intent of an author and this is often not stated. Also, deciding whether a statement is critical or not is subjective too. And often sources and their statements are too complex or subtle to be interpreted in a simple for/against way. Per WP:CREEP, we must judge each case on its merits rather than trying to construct elaborate rules. Colonel Warden (talk) 21:32, 6 September 2008 (UTC)

Hi Warden, a source shouldn't be used for the purposes of the guideline unless its intent is very clearly laid out per OR rule 2. This, I believe, is consistent with the proposal. I will reword to make this more clear. More stringent rules isn't always a bad thing. Thanks. --Phenylalanine (talk) 23:25, 6 September 2008 (UTC)

• I'm afraid I must oppose this addition as well. These sort of situations should be handled on a case-by-case basis using the existing policies, common sense, and debate when necessary. I just don't like the idea of trying to set hard rules about such tricky issues as this. Kaldari (talk) 22:16, 6 September 2008 (UTC)

Thanks for the response Kaldari. I'll maybe get back to you with a simpler proposal. --Phenylalanine (talk) 00:07, 7 September 2008 (UTC)

Add WikiInfo Link to Introduction

I'd like to add the following to the introduction:

If you have original research or commentary to contribute to a subject, the place to do so is at [WikiInfo], not here at Wikipedia. Original research and original ideas are welcome at Wikinfo and editors have the opportunity to edit more freely.

I think this would be a good way to encourage those who just can't curb themselves from original research to find a welcome venue for their ideas. Any support or objections to this proposal?--SaraNoon (talk) 22:55, 10 September 2008 (UTC)

We would welcome that at Wikinfo, although in certain cases it will result in export of troublesome issues. Fred Talk 23:08, 15 September 2008 (UTC)

Since not objections, will add. I am sure Fred at Wikinfo can help to head off any "exported" troublemakers.--SaraNoon (talk) 03:53, 17 September 2008 (UTC)

I think it is highly inappropriate to highlight a single external project in this fashion. Additionally, there are other WMF projects that permit (and encourage) original research, making the promotion of a specific external project unrelated to the Foundation even more of a concern. We could create a section listing wikis where original research is appropriate and welcome, listing the appropriate WMF projects and a selection of other external projects. Vassyana (talk) 06:40, 17 September 2008 (UTC)

If you want to include additional alternatives, please do. I'm not sure what they might be, but just add them right after the WikiInfo link, or if you have a WMF project in mind, put it in front. It is better to add than to delete. I'm replacing so other editors can comment. In my view it is very important to include something of this sort within the first section precisely because it does point "original researchers" to go somewhere else, which is good for Wikipedia.--SaraNoon (talk) 13:07, 19 September 2008 (UTC)

<undent>W.r.t. this edit, I must say I agree with Vassyana's basic take on it above, though not necessarily the "modest list" below. For the moment, I didn't want to unilaterally revert it, so I've pared it down to a simpler statement here, making clear that Wikinfo is only an example, as Vassyana has pointed out. I figure the next step, consistently with what Vassyana says above, is someone else cries "foul! we've got an OR website too!", then I imagine in short order it is likely to be a place where virtually every "OR friendly" website will want to post their names and URLs, perhaps next we'll need to collectively decide which of the numerous blogs meet WP:Notability, etc., etc. Perhaps the statement might be consensused to be appropriate without any specific mention of particular sites wherein it is permissible for users to publish OR? Either way, the proposed statement appears to me to require the scrutiny of a somewhat broader sampling of the WP community than has weighed in on it thus far. ... Kenosis (talk) 14:00, 19 September 2008 (UTC)

Places where OR is welcomed

Following SaraNoon's lead above, it would be appropriate to make a modest list of places where original research is welcomed (adding such a section before or after "See also"). It would be easy to make a short list such as:

Some Wikipedia sister projects welcome original research:

• Wikinews
• Wikiversity
• Wikibooks

Some external collaborative projects welcome original research:

• Wikinfo
• Everything2
• h2g2

Starting a new wiki accommodating original research can be done at wiki farms:

• wetpaint
• Wikia
• Wikidot

Thoughts? Vassyana (talk) 06:40, 17 September 2008 (UTC)

Vassyana, I made a comment above. I think it's an excellent set of examples of where OR is welcomed, but also think it will very quickly become a linkspam magnet. ... Kenosis (talk) 14:23, 19 September 2008 (UTC)

Kenosis's abbreviation is fine. I took the extra line from WikiInfo itself at a time when I was mistakenly under the impression that WikiInfo was one of the Foundation's projects and would therefore be a reasonable cross link between projects. Oops. But the concept is still correct. After all, the reason people violate NOR is because they are wanting to contribute "Truth" as they see it. So instead of having a policy designed only to frustrate OR, we should include in the policy at least one line and link to redirect that desire (and energy) somewhere where it will be more productively used. The sooner, and more often, OR contributors see this "lead" for finding places that welcome OR, the better.

I have no particular favorites to recommend and have no connection with Wikinfo. But I do think we should not just say, "Don't do original research here" but should instead actively encourage people to find a place that does welcome original research. And to make a reasonable recommendation we should include some link...perhaps to a Wikipedia article or a guideline about "Original Research Wikis"(?).--SaraNoon (talk) 19:58, 19 September 2008 (UTC)

Citing sources simply for what they say.

I have a question that arose from a recent edit I suggested: It seems to me a matter of common sense that the usual cautions about not citing "primary sources" do not apply when a source is being cited simply for its content, and not to support a claim that the source is making. For example, a Wikipedia article about a novel should draw its straightforward synopsis of the novel's plot from the novel itself, not from a published "secondary source" about the novel. (Literary criticism of the novel is another matter, of course.) Similarly, the plot of a movie should be cited to the movie itself, not to somebody else's synopsis (unless that other synopsis makes a distinct claim of its own). And, for example, direct quotations from, or even straightforward summaries of, a politician's speeches should be cited, when possible, to official sources, not to "secondary sources." Finally, if an article about a notable figure wants to quote (in a straightfoward fashion, apart from additional analysis) that figure's views, it should feel free to cite that figure's website, blog, books, Op-Ed pieces, etc., rather than rely on "secondary sources" that report the figure's views.

So, am I right about this major qualification to the "primary vs. secondary sources" rule? And is the issue directly addressed somewhere? (There's a bit in the verifiability article about citing self-published and untrustworthy sources about themselves, but that's really only a subset of the larger issue I'm raising.) Just asking. Thanks. P.D. (talk) 13:26, 17 September 2008 (UTC)

The policy does not say you can never cite a primary source... it just says you should use caution when doing so. The key is to avoid OR (either intentional or unintentional). It is fairly easy to avoid OR when summarizing the plot of a book or movie... it is much, much harder to do so when summarizing something like a political speech. Even simply quoting a speech can end up violating NOR, if the quote is taken out of context. In short, avoiding OR depends more on how you use (or misuse) the source than on what the source is. It is easier to mis-use a primary source, hence the caution.

As for using blogs, op-ed pieces and other "opinion" sources... see WP:V and WP:RS. Blueboar (talk) 14:51, 17 September 2008 (UTC)

Please read, comment, and contribute to the discussion of clarifying guidelines on this at Wikipedia:Primary_Secondary_and_Tertiary_Sources.--SaraNoon (talk) 13:18, 19 September 2008 (UTC)

Proposed addition to WP:SYN

There's a discussion going on over at WP:OR/N about 'Is citing research that contradicts the claim of a Fringe theory WP:SYN?' Consensus there is that, it is not synthesis to include a source if they directly confirm or contradict a claim made in an article, even if the source does not directly refer to the group making the claim. I.e: If a group claims that "AIDS is caused by magic", it is not original research to cite research on how AIDS is caused by the HIV virus.

Consequently, I would like to add this paragraph to the end of WP:SYN:

However, it is not synthesis to include a source that is directly related to a claim made in an article, even if the source does not directly comment on the specific situation. If for example, the article states that, "Smith claims that not consulting original sources before citing them is plagiarism." Then, it would be acceptable to note that, "The Chicago Manual of Style defines plagiarism as using a source's information, ideas, words, or structure without citing them."

lk (talk) 09:33, 20 September 2008 (UTC)

As an aside, it would definately not be appropriate to say "The Chicago Manual of Style defines plagiarism as using a source's information, ideas, words, or structure without citing them" because the Chicago manual of style says no such thing - it barely mentions plagiarism and certainly does not define it. The definition of plagiarism as "using a source's information, ideas, words or structure without citing them" is paraphrased from Harvard's "Writing with sources" manual. Enchanter (talk) 16:06, 20 September 2008 (UTC)

In that case, the paragraph needs to be rewritten. Perhaps as follows:

However, it is not synthesis to include a source that is directly related to a claim made in an article, even if the source does not directly comment on the specific situation. If for example, the article states that, "Smith claims that not consulting original sources before citing them is plagiarism." Then, it would be acceptable to note that, "While the Chicago Manual of Style requires that original sources be consulted before citing them, it does not label the practice of not doing so as plagiarism."

Or is this better?

However, it is not synthesis to include a source that is directly related to a claim made in an article, even if the source does not directly comment on the specific situation. If for example, the article states that, "Smith claims that not consulting original sources before citing them is plagiarism." Then, it would be acceptable to note that, "However, the American Heritage Dictionary defines plagiarism as using or passing off the ideas or writings of another as one's own."

Or should I go with the AIDS caused by magic example? Even if these examples need to be reworked, do you agree in principle? lk (talk) 17:04, 20 September 2008 (UTC)

I agree; that CMS example has caused more trouble on WP with people using it to wikilawyer out sources they don't agree with. It could use some counterexamples. Squidfryerchef (talk) 18:35, 20 September 2008 (UTC)

Provide examples please. Jayjg ^(talk) 06:06, 21 September 2008 (UTC)

I strongly disagree. This runs counter to long-standing practice and consensus regarding this policy. It also creates a massive loophole that opens the door to abuse. Additionally, the case that spurred this proposal is spurious at best. There's no need to cite sources unrelated to the article topic (which is not the same as the broader topic to which it belongs). This particular case is particularly spurious, since reliable sources contradicting the theory are easily available.[1][2] Wikipedia is not the place for a pet debunking project. Stick to the sources that discuss the topic. Vassyana (talk) 05:59, 21 September 2008 (UTC)

I strongly disagree for the same reasons given by Vassyana. You don't get a free pass to insert WP:NOR simply because you want to debunk a fringe theory. Also, it is false to claim that there a "consensus" at the Noticeboard that this is permitted. Jayjg ^(talk) 06:06, 21 September 2008 (UTC)

There is a broader issue involved than just debunking fringe theories. I believe that what I propose simply acknowledges the current practice of including sources that are directly related to an issue that appears in an article, even if the source does not directly refer to the particular person(s) or group(s) that the article is about. For example, in the page about Margaret Thatcher, the observation that during the early Thatcher years "(VAT) was raised sharply to 15%, with a resultant actual short-term rise in inflation", is sourced to a report about VAT that doesn't specifically make the connection between Margaret Thatcher and inflation. Or, in the article about the 9/11 attacks, reference is made to "bin Laden's declaration of a holy war against the United States", sourced to reports before the 9/11 attacks. Obviously this statement is relevant, even though the reports were not directly referring to the 9/11 attacks.

Anyway, since it's still controversial, I'll withdraw my proposal for now. Any further discussions should probably occur over at WP:OR/N, in the section 'Is citing research that contradicts the claim of a Fringe theory WP:SYN?' lk (talk) 09:52, 21 September 2008 (UTC)

The Margaret Thatcher example is clearly original research as it does not support the statement made in the article. It doesn't even mention the United Kingdom or the 15% figure provided, let alone Thatcher. The 9/11 example does use a source that explicitly refers to the article subject, supporting the statement that investigators referenced his early statements of enmity as evidence of motivation to commit such acts as the 9/11 attacks. Your examples are respectively good illustrations of what is and what is not original research. Vassyana (talk) 11:12, 21 September 2008 (UTC)

Your opinion that the Thatcher reference is OR is in line with your beliefs. It does not however disprove my contention that this type of citation is currently common throughout wikipedia. Obviously I didn't read enough on the 9/11 article reference. Let me come up with a few more examples. Also, once again, all further discussions over at WP:OR/N please. Let's try to keep it in one place. lk (talk) 11:42, 21 September 2008 (UTC)

"Directly related" vs. "Refers directly"

As per the discussion last April, the phrase "directly related" is too broad, as it allows anyone to essentially insert whatever they want into a discussion, simply by claiming it is "directly related". The subjectivity of the phrase is far too high. On the other hand, "refers directly" leaves no wiggle room. The claim that fringe theories won't have been directly addressed by reliable secondary sources is also not a real issue; if a fringe theory hasn't been addressed by reliable secondary sources then it's not notable enough to include in the first place. Jayjg ^(talk) 18:05, 21 September 2008 (UTC)

Also, as an example of how this happens, last year I ran into an editor who insisted on inserting the original research into the HonestReporting article that the group "could be considered a classic example of the Hostile media effect", cited to various apparently reliable sources.[3][4][5][6][7][8] As was quite clear, the only source claiming HonestReporting was a "classic example" of hostile media effect was the Wikipedia editor himself; none of the sources actually mentioned HonestReporting. When confronted with this on the article Talk: page, the editor insisted it was not original research.[9] Rather, I was apparently misreading "directly related" as "directly referenced". He apparently could use any sources he wanted, so long as he felt they were "directly related". And this, of course, is the giant loophole in the phrase "directly related". Who decides whether or not the material is "directly related"; Wikipedia editors, or reliable secondary sources? Jayjg ^(talk) 18:19, 21 September 2008 (UTC)

I read through the discussion in Archive 34, and I don't see any consensus reached.

3 questions:

• Where specifically in the archive is the consensus on this issue? Where is the paragraph that says: "Do we have consensus on this issue?" Followed by a bunch of "Aye" "Aye" Aye".
• If there was consensus why wasn't the change made in April?
• Even if there was consensus then, there obviously isn't now. So are you still justified in pushing the change?

lk (talk) 18:20, 21 September 2008 (UTC)

Please respond to the points raised above. Jayjg ^(talk) 18:21, 21 September 2008 (UTC)

I think the issue is too complex to answer in two words - we need lots more examples of cases and try to formulate a careful rule. Meanwhile I think we should remain with the long-standing version until there is consensus for change.--Kotniski (talk) 18:27, 21 September 2008 (UTC)

I've brought an example. As pointed out, the wording that was used in the past was effectively meaningless, since anyone could claim the material they added was "directly related"; as, in their minds, it was. There was consensus for changing the words to reflect the actual intent of the policy, and it has been implemented. If you disagree, please respond to the points raised above. What is your answer, for example, to the person who insists that "hostile media effect" is "directly related" to HonestReporting? Jayjg ^(talk) 18:36, 21 September 2008 (UTC)

Jayjg, I can see your point given the example that you gave. However I think it would fail the test of WP:NOR even without the change you want, as the lead already requires that the source directly support the information as it is presented; and his sources do not directly support his POV statement.

We all want to improve wikipedia here, we all have to deal with problem editors who are POV pushing, so lets try to work out a version that makes both sides happy. How about we tighten up the second sentence, the one about directly support, instead of insisting that all sources must directly refer to the title of the article. I suggest: "and the source must directly support the statement or information presented.

lk (talk) 18:44, 21 September 2008 (UTC)

"Directly support" doesn't help much, since in the article in question, the material could just be inserted without the pre-amble. The editor doesn't have to make the direct connection between the material, but can still lead the reader to draw those conclusions anyway. Jayjg ^(talk) 18:55, 21 September 2008 (UTC)

The second reference to the basic point at issue here, in WP:SYN, which uses the words "directly related", is adequate if it's supported by the prior use of the words "refers directly to" in the lead section of the policy. If, on the other hand, the initial use of the words "refers directly to" is going to be subject to repeated attempts at watering it down with the reasoning that it's somehow inconsistent with the use of the words "directly related", then (1) the second will need to be brought into line with the first, as Jayjg has just done (here) in response to such a watering down, or (2) the longstanding language (since April at least) will need to continue to stand until this current resurrection of the April discussion is brought to rest in some reasonable way. Kotniski's last edit is reflective of neither. The initial change a few days ago was made by lk here... Kenosis (talk) 18:48, 21 September 2008 (UTC)

Just to clarify, before I reverted Jayjg's change (which had stood for only one day), I checked and the text has been "provide information directly related to the topic" for at least 3 months back. The long standing text has been "directly related" I would not have made the change otherwise. lk (talk) 19:21, 21 September 2008 (UTC)

Jayjg, my review of the recent history going back to April confirms what lk is saying here. So lk and Kotniski, I stand corrected on the issue of which is the "longstanding version" w.r.t. the two clauses at issue here. Why was this change never implemented subsequent to the April discussion? ... Kenosis (talk) 19:49, 21 September 2008 (UTC)

If you look at the version from December last, the wording was clear that sources must have "published this argument in relation to the topic of the article" and that the "precise analysis must have been published by a reliable source in relation to the topic before it can be published in Wikipedia by a contributor." (In fact, these passages are still present in the current version.) Now, how can a source publish something in relation to the article topic if it doesn't even mention the topic of the article??

As far as I can see, the problem arose with the good-faith insertion of a new lede paragraph in the SYN section on December 29, 2007, which stated that in order to avoid OR, sources must be directly related to the topic of the article – which, rather than simply introducing the argument made by the Smith & Jones case, as was the intent expressed on the discussion page, inadvertently weakened it. The intent of the Smith & Jones case, which requires a source that specifically comments on the Smith and Jones dispute (and that means nothing but a direct reference to the Smith & Jones case), is clear and has always been the same.

Also, please note the text of footnote 2 that exempts common-sense, uncontroversial statements from WP:SYN.

We cannot allow the SYN lede to direct the reader's mind down a path that is directly opposite to what the Smith & Jones scenario is all about. First impressions last ... I will revert to Jayjg's version. Jayen466 19:06, 22 September 2008 (UTC)

• The discussion Jayjg is referring to is Wikipedia talk:No original research/Archive 34#The problem with "directly related", which demonstrates consensus for the "refers directly" wording. There's also an interesting discussion there about how to source necessary background information (e.g. once I've referenced that Subject X and Subject Y are related in the article on X, can I then use sources about Y that don't discuss X?), but it ends up concluding that that isn't the purview of this policy and has more to do with WP:COATRACK and WP:UNDUE. --erachima talk 18:49, 21 September 2008 (UTC)

I've just again read through the entire discussion on 'The problem with "directly related"', and again, I see no consensus. I see a long discussion with no agreement between the two sides. I really think we can work out a wording that satisfies both sides. We just have to be creative about it. lk (talk) 18:58, 21 September 2008 (UTC)

I would like to lay out a basic statement of the issue that we can agree on:

• On one side, we have concern that POV pushers will add OR claims about the topic of the article, and cite them using sources that do not directly refer to the article topic. So, we have a desire to limit the sources that can be cited.
• On the other side, we have concern that POV pushers will wiki-lawyer to exclude relevant sources because they don't directly refer to the title of the article, even though they directly support a relevant statement.

Is that basically correct? lk (talk) 19:29, 21 September 2008 (UTC)

We are always concerned about how POV pushers will interpret and misuse our policies and guidelines, but it seems as if this is one situation where there is potential for abuse no matter how we word things. So I think it might help if we should focus on what we want to allow, rather than what we want to prevent. I favor "directly related" because it is slightly more inclusive than "refers directly". It allows editors to better discuss both sides of topics where there are differing opinions... and thus better adhear to WP:NPOV. Blueboar (talk) 20:04, 21 September 2008 (UTC)

But WP:NPOV cannot trump WP:NOR! We don't really care what editors want to "discuss", we care what reliable sources want to discuss. If reliable sources don't make the connection, or bother showing a specific "side", then why should Wikipedia editors be allowed to? Jayjg ^(talk) 20:06, 21 September 2008 (UTC)

lk, you've left out the critical point - in the second case, how do we know that the sources are "relevant"? In the first case, we rely on the sources themselves to ensure that they are relevant. In the second case, Wikipedia editors get to decide. Jayjg ^(talk) 20:06, 21 September 2008 (UTC)

I agree that NPOV does not trump NOR... but neither does NOR trump NPOV... they have to work in concert with each other. That is my point. I think they best work in concert if we use the wording "directly related". that said, I do support a narrow interpretation of what we mean by "directly related". Blueboar (talk) 20:26, 21 September 2008 (UTC)

While you may support a narrow interpretation of "directly related", others do not, and habitually use it as a massive loophole. We can't rely on you to be around to tell people "no, you're taking it too far now". Instead, we need wording that makes it clear. In addition, this wording doesn't trump WP:NPOV, which insists that articles reflect "all significant views that have been published by reliable sources". You're advocating the inclusion of views that haven't been published by reliable sources. If a reliable source has published a rebuttal, explanation, background, whatever, then we know it's significant and relevant. Jayjg ^(talk) 20:38, 21 September 2008 (UTC)

Jayjg, since I was not much involved between January and August of this year, I have no idea why this change discussed in April was not immediately implemented subsequent to the April discussion. Do you have any recollection of this? ... Kenosis (talk) 20:52, 21 September 2008 (UTC)

Sorry, no. Perhaps people just got distracted. Jayjg ^(talk) 01:11, 22 September 2008 (UTC)

I agree with Jayjg. This loophole can be used for promoting views which are not discussed by reliable sources on the topic of the articles. Wikipedia should not create polemic if it does not exist (in reliable sources). -- Vision Thing -- 20:56, 21 September 2008 (UTC)

I've identified the source of my gut aversion to this change. The change to WP:SYN contradicts and second guesses WP:SOURCES. WP:Sources dictates what sources may be used in an article. WP:SYN is supposed to cover statements that may be made in articles. This change to WP:SYN will restrict the sources that can be cited, which is the wrong domain for WP:SYN. As surely, a source alone, without a statement attached, cannot be synthesis.

I want to go back to proposing that we strengthen the phrase to "the source must directly support the statement made". This will bring the domain that WP:SYN covers back to the statements made. If a statement is really OR, surely a source cannot be found that will directly support it. lk (talk) 20:48, 21 September 2008 (UTC)

lk, I think I got what you're saying here. The clause in WP:SYN that says "... or if the sources cited do not refer directly [are not directly related] to the subject of the article," is a restatement of the concept that the source must be directly related to the topic. I'm not sure when it was put in there, but it appears to me to be a restatement of the principle of NOR specifically intended to prevent instances where a user presents an "A+B=C" scenario derived from a different topic. Blueboar and I have referenced this issue below, and perhaps others may wish to weigh in on it further. At this point, I consider the issue to be secondary to the discussion about what words best convey the concept in the lead section of this project page. ... Kenosis (talk) 02:29, 22 September 2008 (UTC)

"Relate" is a problematic wording because it's too vague, and essentially allows any source to be included on the grounds of "because I think so". This is of course most problematic in the case of POV pushers maliciously abusing it on controversial articles, but it can also cause confusion in good faith editors who fail to understand what it means. I recently saw a case where an editor had used a source about the general state of the comics industry to support a claim about why a specific series was canceled. The claim was original with respect to that series, but they thought it was acceptable usage because "the state of the comics industry at the time series X was canceled" intuitively seems like it's related to the series. "Directly refer" eliminates that problem handily by requiring all the cited sources talk about the specific subject, and though I can see it being used in a lawyering fashion as well, it is problematic in less situations, so until we find something more satisfactory we should probably use it.
As for finding a new, better wording, we probably need to rephrase the entire sentence rather than just trying to find a different two words that everyone can agree on the connotations of. Perhaps a phrasing along the lines of "article text must be based on reliable published sources that directly support the information in the article as it is presented, and these sources must apply directly to the topic of the article, as verified by either direct reference to the subject or in other reliable sources."? A little wordier than I would like, but the length may be warranted if it clears up the concern people appear to be having with the "directly refer" phrasing excluding sources for basic background information. (e.g. if we have reliable sources to verify that comic series X was indeed canceled due to problems in the industry at the time, but which do not discuss those problems in detail, it would be acceptable to cite a more detailed source that was about the industry but does not specifically mention the series.)--erachima talk 20:59, 21 September 2008 (UTC)

I'm inclined to agree with the assertion that at least one use of the words "refers directly to [topic X]" is needed here. Even the words "refers directly to" can be stretched to mean roughly "covers the same territory [as topic X or statement Y]", as in the comic-book industry example just above. But it's more of a stretch to assert that the broad claim about the state of the comics industry refers directly to the state of the specific series than it is to say they're "directly related", unless the putative reliable source about the state of the comic-book industry specifically mentions that series. It seems to me that in the rare cases where a source dealing with a particular topic doesn't mention the topic or a widely acknowledged synonym of the topic specifically, if yet another reliable source can't be found for the proposition that "Source A refers to Topic X", it shouldn't be used to support a statement made in an article about Topic X. The words "directly related to", on the other hand, have the problem illustrated by both Jayjg and erachima above, wherein the criterion used to assess the statement in the WP article is capable of being construed extremely broadly and a whole additional set of analyses comes into play w.r.t. what is meant by a "direct relationship". ... Kenosis (talk) 21:44, 21 September 2008 (UTC)

OK... you have convinced me that we need to find another way to discribe what we mean without using either phrase. Getting to the heart of the matter... Let us take the situation where source A talks about topic Y, and source B also talks about Y. I think it is highly appropriate to discuss what both source A and source B have to say about Y in the article on Y (doing so conforms the article with NPOV).

I think the problem comes when people start talking about what sources A and B say about Y, in an article that is really about Z. Blueboar (talk) 23:30, 21 September 2008 (UTC)

RE "I think the problem comes when people start talking about what sources A and B say about Y, in an article that is really about Z." : Yes indeed. I believe this is essentially the problem described by both Jayjg and erachima. ... Kenosis (talk) 23:57, 21 September 2008 (UTC)

Yes, that's it exactly, at least to an extent. But see also the example below. Jayjg ^(talk) 00:13, 22 September 2008 (UTC)

Let's use another example. Suppose there's an article about organization A, and person X is a member of said organization. Now suppose person X is accused of some crime in the newspapers; would that material be appropriate for an article on organization A? From a policy point of view, the person's relationship to organization A would only be relevant in this context if reliable sources had mentioned it, and even then, only to the extent that reliable sources had mentioned it. Jayjg ^(talk) 00:19, 22 September 2008 (UTC)

To add a bit to Blueboar's statement above, perhaps a slightly more accurate way of analyzing this might be to start the analysis around something like: "A problem w.r.t. WP:NOR occurs when people start talking about what source A says about Y, in an article or statement that is really about Z." . Whether there's a Source B is not necessary to this analysis, at least not on the examples that are considered to be problematic to the words "refers to" or "directly related to". Thus is seems to me the main bone of contention here ought be whether the words "refers to" should be placed in the lead section. Secondary to that, it seems to me, is whether to repeat the words "refers directly to" in WP:SYN, where the synthesis of A+B=C comes into play (where either "=" or "C", or both, is represented in one or more statements made in the article). This latter point is directly related, if you'll pardon my choice of words, to lk's statement above (the one posted at 20:48, 21 September). .... Kenosis (talk) 01:24, 22 September 2008 (UTC)

Then perhaps the solution to this isn't to worry about the language we use ("directly related" vs. "refers directly")... but to add a better explanations of what we mean by including an A,B,C... X,Y,Z type example the way we do for the basic definition of SYNT. Blueboar (talk) 03:18, 22 September 2008 (UTC)

I strongly support this policy clarification effort. In fact, I outlined a very similar proposal above (see Wikipedia talk:No original research#Policy clarification proposal), which may be helpful in rewording the policy. I wish I had seen this discussion before. Cheers! --Phenylalanine (talk) 03:57, 22 September 2008 (UTC)

Phenylalanine, If I read the above discussion properly, you are in favor of restricting when a statement is considered properly source, however you are not in favor of restricting the sources that can be used. Am I correct? lk (talk) 04:20, 22 September 2008 (UTC)

Yes, exactly. It's the way sources are used to verify statements that must be restricted. --Phenylalanine (talk) 04:52, 22 September 2008 (UTC)

I think we have to be very careful about restricting acceptable sources. It's fine to restrict acceptable content – tightening up when a statement is considered properly sourced, is I think, the way to go. Whatever statement we come up with, it shouldn't be incompatible with our current de facto community standards – that if a statement properly appears in an article, it is properly sourced as long as the source directly supports the statement made.

If a specific claim of fact is made in an article (assume that the claim should properly be there); it should be acceptable to cite articles that support or refute that particular claim of fact, even if the article does not refer to the topic of the article by name. Whatever we agree on should be able to accommodate these hypothetical cases:

• An article about Wild Civets in Asia should be able to cite articles that describe wild civets in general, and in the article section on 'Wild Civets in Japan', it should be able to cite articles on Japanese Wild Civets.
• In the article about the second Iraq war, if it is appropriate that a statement of fact appears about the first Iraq war, this can be sourced to an article written about the first Iraq war before the second Iraq war started.

lk (talk) 04:11, 22 September 2008 (UTC)

Lawrencekhoo, here's a summary of my proposal. I believe it would be consistent with both examples you provided above: "As a general guideline, when adding information to a Wikipedia (WP) article, you should ensure that the reliable sources which you cite directly relate this information to the topic of the WP article, and in the same way that the information is presented in the WP article. For example, if the information added serves to criticize the subject of the WP article, then you are required to cite sources where the information is clearly presented as a criticism of this topic. When developing the context for a subject described in a WP article, if no sources can be found that comply with the above guideline, you may occasionally cite reliable sources that provide information relevant to an aspect of that subject, but which do not directly relate that information to the topic of the WP article. However, to avoid breaching the Neutral point of view policy, you should never do this when adding information that may misrepresent any aspect of the subject or cast it in a more favorable or unfavorable light, for example if the information supports, discredits, highlights or downplays any aspect of that subject." Cheers! --Phenylalanine (talk) 04:50, 22 September 2008 (UTC)

I don't understand all of this. We don't add information to "serve to criticize" anything; we add it because it's relevant to readers' understanding of the topic. And the bit at the end: it is just as likely that providing such information will uphold NPOV as breach it, so we can't be categorical about this. In general, though, I think we're going in the right direction. We must make it clear, though, that the "topic" is not necessarily the title of the WP article - a good article will often contain facts about many subtopics or synonymous or related topics, serving to increase readers' understanding of the subject at hand. We shouldn't impose artificial restrictions on the building of good articles for the purpose of allegedly upholding NPOV, when NPOV works perfectly well on its own.--Kotniski (talk) 06:35, 22 September 2008 (UTC)

"NPOV is about accurately presenting the balance of the total body of reliable sources on a topic. "Neutrality" in NPOV is about editors being neutral in the handling and presenting of material, not about balancing the article by some subjective standard of "neutrality". Our goal is not to manufacture some pseudo-objective balance, but rather to present the information in roughly the same proportion it appears in reliable sources. If the vast majority of material about a topic is critical and negative, an article heavily weighted towards presenting critical information is not a problem. Ignoring the no original research rule would lead to a violation of NPOV (representing a view not present in the sources), not a correction towards it."[10] "It is not the job of editors to decide what conflicting facts may be relevant to mention in an article on a fringe topic. As with all topics, that job is left to reliable sources. And there is actually no risk "lean[ing]...towards the views of said fringe theorists." A fringe theory necessarily has insufficient mainstream support to give any of its claims as facts. And since wording something as a claim (John says X is true) is kind of inherently neutral (except when someone is trying to weaken what should be stated as a fact), a properly written fringe article can't give the biased impression that you fear. But back to the previous point, it is part of Wikipedia's fundamental being that all content be easily verifiable; anything more complicated than pointing to a source is heading in the wrong direction. Whenever we find ourselves saying, "sure, no source makes this point about this topic, but trust us," we've done a bad thing."[11]. --Phenylalanine (talk) 10:41, 22 September 2008 (UTC)

I think I would generally agree with you as long as we take a flexible view of what is "the topic". It shouldn't have to be the title of the WP article or even something exactly synonymous with it, but could be another related or subsidiary topic being discussed in the article, as in the civet cats and Iraq war examples given above. If this flexibility is apparently abused to make original allegations and so on, there are almost certainly other policies that can be used to deal with that.--Kotniski (talk) 11:41, 22 September 2008 (UTC)

Yes, I completely agree with you. --Phenylalanine (talk) 11:47, 22 September 2008 (UTC)

Jayjg, we have to hear from you about this. Suppose the lead of WP:NOR makes it clear and insists that 'a statement is only considered properly sourced if the source used directly supports the statement made'. This would address the problem example you gave, as the statement he made, "HonestReporting could be considered a classic example of the Hostile media effect" cannot be properly backed up with the source he used, as the source only defines what "Hostile media effect" is, it does not make the connection to HonestReporting.

You may argue that this allows him to add a sentence explaining what "Hostile media effect" is to the article, without making direct reference to HonestReporting. However, you could then argue that such a statement does not belong in the article, using WP:NPV, WP:NOTE and WP:SYN (edit: also WP:TOPIC). My point is that WP:SYN should concern itself with restrictions about content (the statements made), it shouldn't restrict the sources used – that should be the domain of WP:SOURCES.

lk (talk) 09:10, 22 September 2008 (UTC)

I offer another example of when the requirement that a source 'refer directly' to the article title would be troublesome:

Suppose that newspapers mention the formation of a small group 'Jihadists for peace'. This makes them notable and an article in Wikipedia appears. A few months later, they print a book, Jihad and Peace. However, being a small group, no newspaper comments on it. Since self published sources are acceptable sources about the group itself and its activities, a paragraph appears on their Wikipedia page describing their book, "A book on how Jihad is always a peaceful act ...." Their chairman is quoted as saying, "This book documents the history of the concept of Jihad, and shows how it has never been used to justify any violent acts." There is no more newspaper coverage of the group.

lk (talk) 09:54, 22 September 2008 (UTC)

If the statement is properly attributed, I think it's ok: "The chairman of 'Jihadists for peace' states/believes that "This book documents the history of the concept of Jihad, and shows how it has never been used to justify any violent acts." --Phenylalanine (talk) 10:58, 22 September 2008 (UTC)

I don't see a problem here. If the article title is "Jihadists for Peace", and the group publishes a book, then the book will directly refer to "Jihadists for Peace" – it will state that they are the publishers and that it describes their views. That fulfils the requirement in Jayjg's version of a "direct reference".

Conversely, if the book contains no such direct reference to "Jihadists for Peace", then it may simply present the private views of the Chairman rather than those of the group. In other words, if he does not identify himself as the Chairman of "Jihadists for Peace" in the book, and does not anywhere state that he is writing in that capacity, then the book should only be cited in the article on the Chairman, and not in the article on "Jihadists for Peace". Linking his book to the article on "Jihadists for Peace" requires a source that makes that connection. Jayen466 13:39, 22 September 2008 (UTC)

• Strongly support Jayjg's version. It is absolutely vital that the sources cited make a connection between the content added and the subject of the article, and that this connection is not just present in the mind of the editor adding this material.
• Of course, using common sense, "directly refer" does not mean that the reference must use the exact string of letters present in the article title. If one newspaper article refers to the "Republican party's 2008 presidential campaign", another to the "2008 Republican presidential campaign", and a third to "McCain's campaign", they still very clearly refer to the same thing.
• An editor gave two examples above:

*An article about Wild Civets in Asia should be able to cite articles that describe wild civets in general, and in the article section on 'Wild Civets in Japan', it should be able to cite articles on Japanese Wild Civets.

• Now, assuming the hypothetical "article that describes wild civets in general" does not directly refer to Asiatic civets, then I strongly disagree with the first half of the editor's statement. However, I agree with the second.

• Re the first half of the statement: If we assume that "wild civets in Asia" are notable enough, i.e. have enough of a literature devoted to them to deserve their own Wikipedia article, then the sources we should use for this article are those specifically discussing Asiatic civets. If these refer to civets elsewhere, for example to contrast Asiatic civets with African civets, then such statements may of course be included. What I do not support is an editor inserting something like this: African tropical rainforests are marked by greater vegetation density than Asian ones (citing a paper comparing African and Asian rainforests), and it can be found that African civet species have shorter tails than Asiatic ones (citing a statement from a paper describing tail length as a percentage of body length in two African civet species, and another paper describing tail length of two Asiatic species). This editor could argue that the references cited directly support the statements they have inserted into the article. However, this is nothing but original research, and indeed may be completely misleading. For example, analysis of tail length in all extant species, rather than the four covered in the cited papers, may show that African civets on average have longer tails than Asiatic ones and that the species selected are not representative of either African or Asiatic species in general. Which means that this editor's clever (to their own mind) observation and resulting good-faith insertion cause Wikipedia to state what is, essentially, superficially plausible but in actual fact completely misleading nonsense.
• The situation changes immediately if a general paper on civets points out that tail length is shorter in African species than Asiatic species. This general paper can be cited, because it directly refers to "Asiatic civets" (or civet species found in Asia, or any other such synonym) and makes a citable statement about morphological differences between Asiatic and African types in general. Otherwise, general articles on civets that do not directly refer to Asiatic civets should be used in our generic article on civets only (note that if we have a separate one for Asiatic civet species, we are bound to have a generic one on civets, and we should not unnecessarily duplicate information).

• Re the second half of the statement: If we have an article on "Asiatic civets", with a subsection for "Japanese civet species", then it is fine to cite, in that subsection, a paper describing any of the Japanese species, even if the paper has a narrow focus and does not refer to "Asiatic civets" in general. By referring to a Japanese species, it is referring to an Asiatic species, because there is widespread agreement that Japan is part of Asia. Again, this is just common sense.

• The other example given was:

*In the article about the second Iraq war, if it is appropriate that a statement of fact appears about the first Iraq war, this can be sourced to an article written about the first Iraq war before the second Iraq war started.

• The appropriateness of making such a statement about the first Iraq war in the article on the second depends entirely on whether such a statement on the first war is made in the literature on the second. If it is not, then Wikipedia should not include such a statement either. Editors can come up with all manner of good-faith ideas and mental cross-connections that may be based on insufficient background knowledge, or erroneous assumptions, and if we allow editors to insert such ideas without a source directly referring to the first and second Iraq wars, then Wikipedia would become a primary source for making this connection – and then we are deeply in OR territory and far from what Wikipedia should be: a mirror of the existing literature.

• The change proposed by Jayjg is of key importance to a correct understanding and interpretation of WP:SYN. Jayen466 12:57, 22 September 2008 (UTC)

We get the point, but the question is how categorical we want to be. We must in any case tighten up the wording, so that the kind of reasoning you give about the Japanese civets is explicitly supported (to avoid future pointless debates). In fact I don't agree with your "depends entirely". The appopriateness of including a (reliably sourced) statement in a particular article depends essentially on whether it's going to help readers understand the topic. Not every statement from sources which refer to that topic is going to be appropriate (note: "sources that refer to" isn't the same as your "literature on") - a consensus of editors decides what does go in. Similarly, sometimes an appropriate - usually totally uncontroversial - statement may happen to be referenced to a source that doesn't refer to our topic - the same consensus decision-making can apply in this case too. We must not tolerate individuals' using this latitude to advance positions, but that is already dealt with elsewhere in our policies. --Kotniski (talk) 13:43, 22 September 2008 (UTC)

If mention of a point helps readers understand the topic, then surely some reliable source, somewhere out there, that directly refers to the second Iraq war (I take your point about sources referring to != literature on) will have made that mention. If no such source can be found, then the editor's assertion that "this will help people understand" looks decidedly iffy! Jayen466 13:54, 22 September 2008 (UTC)

I agree with Kotniski on this, the issues are already dealt with elsewhere. Jayen, the trouble with your viewpoint is that you're trying to push onto WP:NOR and WP:SYN, issues that should be handled by policies like WP:V, WP:NOTE and WP:TOPIC. We should not artificially restrict potential sources, this would contradict WP:PCR (Provide Context to the Reader). Sometimes, we may need to explain what wild civet cats are, and there may not be any articles about Asian wild civet cats that provides the necesary explanation.

The example you gave is synthesis, but can already be identified as such using the current guidelines. For example, the statement: "It can be found that African civet species have shorter tails than Asiatic ones" cannot be directly supported by citing any number of papers about tail lengths. It needs an article that specifically compares the average tail lengths of African and Asian species. Also, a phrase like "Civet tail lengths are affected by the density of vegetation" must necessarily follow or precede the statement you gave. Such a sentence cannot be directly supported unless the link between tail lengths and vegetation has already been made. If a sentence like that is left out, then the statement left cannot stand on its own. "African tropical rainforests are marked by greater vegetation density than Asian ones" can be challenged as being irrelevant to the article (using WP:TOPIC).

lk (talk) 14:27, 22 September 2008 (UTC)

I disagree that in the article on Asiatic civets we have to explain what civets are. That's what the article on civets is for, and it's what wikilinks are for, as WP:PCR states. Similarly, in the article on the Iraq war we do not have to (and in fact do not) explain that "Iraq is a middle-eastern country located ..." or that "war is an armed conflict ..." and so on.

Forgive me any shortcomings in the example I made up, but note that if you say It needs an article that specifically compares the average tail lengths of African and Asian species what you are describing is a source that directly refers to Asian civet species, so we are in practical agreement on that point.

WP:NOTE is about whether WP should have an article on a topic, not about what to include in an article. The editor citing extraneous sources unrelated to the topic of the article will argue that they comply with WP:V, because their info is sourced. And surely, the description of civets' habitat is on-WP:TOPIC.

WP:SYN really is an important point that is not made elsewhere, and we have to get this right. --Jayen466 14:51, 22 September 2008 (UTC)

WP:NOTE and WP:TOPIC are classified as WP guidelines. WP:NOR is a WP content policy. ... Kenosis (talk) 17:09, 22 September 2008 (UTC)

My two cents. This appears to be another fight over a word choice around which editors can wikilawyer and argue about policy instead of content. If a source has material relevant to an article, editors should be able to use it without being pulled into wikilawyering attempts to measure how much the article "refers directly" to the subject as opposed it's being merely "directly related" to the subject. I tend to agree with lk (Lawrencekhoo) that we should avoid wording that may be misused to block reliable sources based on an argument that, while no one disputes the facts in a source, they question whether it is sufficiently refers directly to the subject.

Directly related is not too broad. If one wants a compromise, I could accept "Refers directly to facts, material or opinions related to the subject matter."

I also agree with Jayn that SYN is the core issue dealt with in NOR. So it is that section that concerns about misusing "weakly" related articles should be addressed with examples. We should not encourage two words ("refers directly") to become the center of argument, when the real subject of this policy is not about sources but about using sources in a way that reflects their content without adding original research. --SaraNoon (talk) 15:56, 22 September 2008 (UTC)

My concern is that what is "directly related" may vary depending on the POV of the editor.

• For example, are the cultural and religious traditions of a particular ethnic group "directly related" to a public figure who is a representative of that group? I would say, no; others might say, yes, arguing that it's obvious to everyone that his views reflect his cultural or religious background. But discussing an ethnic group's beliefs or traditions in the article on a public figure who is a member is OR, unless there are sources explicitly naming that public figure as a representative of that group and discussing his or her views in the context of their background.
• If some crime or financial statistic rose during the tenure of a minister, is that "directly related" to the minister? In my view, it isn't, for WP purposes, unless there are sources explicitly linking the change in the statistic in question to the tenure of that minister. Citing such correlations without sources establishing the link is OR. But another editor may say, of course it is directly related – he was a member of the government responsible.
• If a source says, "Quality management is bunk", should that be cited in the article on Six Sigma? Unless Six Sigma is directly referred to by the source, it should not; the article on Six Sigma should have criticism that's been specifically levelled at Six Sigma. Bringing forth a source that describes Six Sigma as a quality management tool does not help; but an editor may argue that it proves that the comment on quality management is directly related to Six Sigma, because Six Sigma is a quality management method. Etc.

"Directly related" is simply too elastic a term, and that elasticity was the exact defence used by the editor in Jayjg's real-life example given above. It is not asking too much that the source should "directly refer" to the topic of the article. Jayen466 16:27, 22 September 2008 (UTC)

"Directly related" vs. "Refers directly", cont'd

<Unindent>I'm of the impression that this discussion should be extended, and the implications discussed somewhat more thoroughly. Take, as an example, the article on intelligent design, which was a featured article, and which at present is essentially the same as when it was a featured article. A significant number of the sources used in that article do not refer directly to the topic of the article, but instead refer directly to the particular material in the article supported by those citations. For instance, some sources are used to describe ID as a teleological argument, but other sources used in support of the description of the teleological argument itself refer to specific aspects of the teleological argument rather than to ID per se. So, a hypothetical question: Under the language that sources must "refer directly to the topic of the article", do the passages in the article on ID that are supported by sources that do not refer specifically to intelligent design constitute original research? ... Kenosis (talk) 20:59, 22 September 2008 (UTC)

You are referring to the sources in Intelligent_design#Origins_of_the_concept, right? Such as Intelligent_design#cite_note-36 and following? If so, I would prefer there being a source that describes Intelligent Design as continuing a tradition begun by Plato, Aristotle, Cicero and Thomas Aquinas. I believe the editor who included this information is quite probably correct, and perhaps internalised a source or sources that related ID to these historical precursors to such an extent that they forgot to cite these sources, simply treating it as "their" expert knowledge (note that there is no such thing, all such knowledge is accumulated from stuff one has read, or heard, somewhere). On the other hand, if they just had a good knowledge of the history of ideas, and wrote that section "off the cuff", establishing the linkage themselves, then it is Original Research, pure and simple. Look at it from the opposite way: Assume there is no reliable source linking ID, philosophically and historically, to these figures (i.e. Plato, Aristotle, Cicero and Thomas Aquinas). Would it be correct for Wikipedia to describe such linkage? Clearly not. In my view, the ideal citations for this section would be from a work examining the historical context of ID, and quoting the passages quoted in our article as representative precursors of ID. (Also relevant in this context, perhaps, is Wikipedia:No_original_research#cite_note-0.) Jayen466 23:10, 22 September 2008 (UTC)

I think, Jayen, that a key point w.r.t. this example I just gave is that the new language is "to demonstrate that you are not presenting original research, you must cite reliable sources that refer directly to the topic of the article". It does not say "cite solely to reliable sources that refer directly to the topic of the article." This statement is clarified and amplified by the clause immediately following, which says "... and that directly support the information as it is presented." That all cited sources must be "reliable sources" is of course covered by WP:RS. But I'm certain we haven't considered all the basic slants on this yet, so I wanted to give one or more examples of relatively complex articles that are controversial so this issue gets more fully analyzed across a broader range of specific cases in WP than the examples already given by Jayjg and erachima above.

In the case of the article on intelligent design, it is uncontroversial among philosophers that the argument from design, or teleological argument, as a theological apologetic, goes at least as far back as Aquinas, and it's also uncontroversial that assertions of a coherently designed cosmos go back at least as far as Plato. As we can readily see, the analysis of WP:V and WP:NOR can get fairly complex in some topic areas-- even more so when negotiating and consensusing NPOV in a controversial topic such as that one. Similarly it has been the case with other complex and controversial articles, including global warming, the creation-evolution controversy and numerous others. As I said earlier, I'm generally inclined to support at least one use of the words "must cite reliable sources that refer directly to the topic" as a threshold question in a WP:NOR analysis, wiki-wide. ... Kenosis (talk) 01:44, 23 September 2008 (UTC)

Just as a brief note, it is indeed easy to find sources that discuss Plato, Aristotle, Cicero and Aquinas in the context of Intelligent Design; the chapter starting here is an example, and took 2 minutes to find. Here is another. So if this passage in the Intelligent Design FA were challenged by another editor based on WP:SYN, it could easily be defended as reflecting an argument put forward in published sources that explicitly refer to Intelligent Design, and the sourcing updated accordingly. And you know what, that is a good thing. Editors should be able to challenge such passages, and if they don't stand up to scrutiny, i.e. if the same arguments cannot be found in published sources, expressed in relation to the article topic, they should be removed.

I take your point about the "not solely". A supplementary citation, such as a convenience link to an online dictionary entry explaining the etymology of a foreign word, may add value, and if the editor can demonstrate that there are reliable sources that have bothered to explain this etymology in relation to the article topic, it will also be uncontroversial. So by all means, let us look at the complexities involved. But I am confident that in the end it will always boil down to the same thing: if reliable sources don't present an argument in direct and explicit relation to the article topic – and that means mentioning the article topic or a paraphrase of it – then the argument has no business being in our article on that topic. What we should strive for is exhaustive research and summarisation of such sources as do refer directly to our article topic, and we will have plenty of work doing just that, without bringing in sources that are talking about a different topic altogether. If a linkage to some other topic really is relevant, then some reliable sources will have made it. Jayen466 11:48, 23 September 2008 (UTC)

W.r.t. your assertion that you immediately found a source that meets the criterion "refers directly to": The relationship of intelligent design -> teleological argument -> Cicero, Aristotle and Plato, is, at minimum, a two-step analysis, arguably of the kind described in WP:Synthesis. Many sources have termed ID not a synonym of the argument from design, or teleological argument, but a new slant on a theological argument going at least back to Aquinas, wherein ID is asserted by its proponents to be scientific theory; while yet other sources show that it's uncontroversial that arguments from design/teleological arguments as a whole go back at least to Plato. AFAIK, no reliable sources state specifically that intelligent design goes back to Plato. Please remember this article was quite controversial, and virtually every clause and phrase in it was hotly contested over a lengthy period of time. ... Kenosis (talk) 13:09, 23 September 2008 (UTC)

Okay ... I'm wary of discussing the intricacies of Intelligent Design on this page, because we might be here a while, and to boot I am really not an expert. :-) This source, which I cited above, at any rate equates ID to the design argument, and, referring directly to ID, describes it as part of a historical movement of ideas beginning with Plato. So there are at least some published sources that describe this progression in relation to ID. How to weight such sources, assess their reliability etc. we'll have to leave to the editors of the Intelligent Design page. Jayen466 15:36, 23 September 2008 (UTC)\

RE your statement that "This source, which I cited above, at any rate equates ID to the design argument, and, referring directly to ID, describes it as part of a historical movement of ideas beginning with Plato." : Exactly. The connection between Plato and ID is not one that most reasonable persons, upon analyzing it closely, would I think say the source "refers directly to intelligent design [the topic of the article]" as going back to Plato, but rather is one that seems more likely to be said to be "directly related", two examples of the same class of things that are presented in the chapter you linked to that discusses the design argument in general. This would appear to me to move the debate into the WP:SYN analysis. Which is why I personally don't support a repetition of the words "refers directly to the topic of the article" in the section on WP:SYN, at least not as Jayjg proposed it. In the source you just provided, the topic that is being directly referred to is the design argument. ID proponents have asserted it's not a philosophical or theological argument, but rather a scientific theory, and that the premises for ID as a scientific theory are different from philosophical and theological arguments from design. Other reliable analysts have stated that ID is an argument from design that doesn't specify "who" or "what" the designer might be, except to refer to an "intelligent designer". As straightforward as the connection is going back to Plato, no reliable sources explicitly stated that ID goes back at least to Plato. Which is a "direct relationship to the topic of the article", but not a "direct reference to the topic of the article" in explicit support of the premise that ID itself goes back to Plato. Rather, it is the teleological argument from design that goes back at least to Plato, and ID belongs in this class of things.

I feel certain there is a solution to this little dilemma, but it seems to me the language will need to be looked at a bit more carefully so that relevant sections on a subtopic such as, e.g., the discussion of the teleological argument from design in the article on intelligent design, are covered w.r.t. WP:SYN. In other words, for a WP:SYN analysis, I don't think there necessarily need be an additional use of the words "refers directly to the topic of the article", because, assuming the change from "directly related" to "refers directly" in the lead of WP:NOR achieves consensus, it will already have been stated in this policy that there must be included some reliable source(s) that "refer(s) directly to the topic of the article" in order to show, as a basic threshold question, that original research isn't being done. When we get to the WP:SYN analysis, there's no need, IMO, to repeat the words "refers directly to the topic of the article". In WP:SYN, the words "directly related to the topic of the article" appear to me to be adequate to demand the needed connection to the topic. ... Kenosis (talk) 16:07, 23 September 2008 (UTC)

By "directly referring to ID", I simply meant that both the sources provided mention Intelligent Design by name. (The first one – while not an ideal example – has two copious chapters on Intelligent Design preceding the linked appendix, which again explicitly mentions the concept of an "intelligent designer", and the second one mentions Intelligent Design by name, and presents its information about Plato et al. in direct relation to Intelligent Design.) This is in contrast to Jayjg's example, where the relevant organisation that the article was about was not mentioned or alluded to anywhere in the sources that the editor tried to use. In other words, I am less concerned with how direct or indirect the linkage is that the source establishes, I am happy if the topic even appears in the source, with any kind of linkage! First things first. Jayen466 21:33, 23 September 2008 (UTC)

I'm inclined to support the strict view, that is "directly refer" rather than "directly related". Other editors such as Jayen and Jayjg have done a good job of explaining why the looser language causes problems. Also, I've seen it plenty of times myself. However the question raised by Kenosis should be addressed. It comes down to what is being directly referred to. Let's say that we have an article on Joe Smith, and we have a reliable source that says Smith is a proponent of "Nowism". Since WP doens't have an article on Nowism an editor adds a sentence explaining the concept and uses a source that directly refers to Nowism but which doesn't mention Smith. In that case, I think the source would be valid as it directly refers to the assertion being made, and there is another source that directly refers to the subject in connection with that assertion. Where we'd get into trouble is if we then go back and say specifically that Smith believes in XYZ because a source says that Nowism holds those views and because Smith is a Nowist. ·:· Will Beback ·:· 21:27, 22 September 2008 (UTC)

I believe the reservation expressed in footnote 2 does give enough wiggle room for such uncontroversial cases. Jayen466 23:10, 22 September 2008 (UTC)

Are footnotes part of policy? We shouldn't hide parts of policy in a footnote. I continue to believe this proposed change will cause more conflict, not less.--SaraNoon (talk) 03:08, 23 September 2008 (UTC)

A reasonable question, IMO. Footnotes are, of course, in addition to being a standard method of citing sources, also a standard method of specifying clarifications to reasonably anticipated questions and/or other clarifications that are percieved by the writer(s) to be necessary, but not sufficiently central to the points made in the body text such that they should add unnecessary confusion to those readers who are only interested in the basics. I trust that I needn't provide a full list of writing styles in which this is widely accepted common practice. If there's any doubt about this, please let me know, and I'll attempt to clarify further. ... Kenosis (talk) 03:29, 23 September 2008 (UTC)

Yes, that footnote does appear to cover the scenario I outlined. With that included I strongly support the "directly refer" langugage. ·:· Will Beback ·:· 23:27, 22 September 2008 (UTC)

Note that the footnote has today been deleted by a participant in the discussion above. :0 Jayen466 21:36, 23 September 2008 (UTC)

While I strongly support this effort to clarify the policy, I have some reservations regarding the specifics of this proposal, which requires that the article title or exact synonyms thereof appear in the sources. For example, take an article such as the "Environmental effects of meat production", I don't think it would constitute OR, under any reading of the OR policy, to add information to the article that is verified by sources referring to the environmental effects of "intensive livestock farming", even if the sources don't use the exact term "meat production" or synonyms thereof. I also see that focusing on terms is not the right approach since there will be cases where the sources will not explicitly mention the relevant terms but will still refer to the concept described by those terms by indirect means. The key issue here is the specific concept described by the article's title. There are articles, where it seems to me obligatory to use sources that refer to the concept described by the article's title. Such would, IMO, be the case for articles like "Sustainable food system" and "Raw foodism". Perhaps the reason is that in the case of "Sustainable food system", we are dealing with a term defined within a very specific discipline or field of study. While, "Raw foodism" is not defined within a specialized field, it refers to a concept that is based on a point of view. "Raw foodism" is based on the belief that uncooked foods are better than their cooked counterparts. In this article, it would, IMO, be unacceptable to provide information intended to contextualize (i.e. contrast with opposing research/evidence or with mainstream views) the statements supporting/contradicting/criticizing the Raw foodism concept if the contextual information were not verified by sources that refer to the concept designated by the article's title. It would be sufficient that the sources refer to the general concept of Raw foodism, that means terms or phrases such as "raw foodist", "raw veganism" or "the dietary approach advocated by Aajonus Vonderplanitz" would be acceptable. Terms such as "Raw meat diet", "Raw food diet", "Raw milk", "raw veggies" would also be adequate in contexts such as "Some people believe that a Raw food diet is superior to a diet composed mainly of cooked foods, however studies have shown..." or "Some food faddists believe that raw milk is healthier than the milk you find in the supermarket..." or "A popular dietary regimen these days is the consumption of a Raw meat diet, needless to say many nutritionist have expressed worries regarding such practices", etc. So, as long as the sources refer to the general idea of raw foodism (a dietary approach based on the premise that some or all raw foods are healthier than their cooked counterparts) as illustrated by the above examples, I believe that they may be cited in the "Raw foodism" article. I believe that this approach should be followed for all articles about concepts based on points of view or specialized terms such as "Sustainable food system. On the other hand, very broad and category-like articles such as the "Environmental effects of meat production" could hardly be required to meet the requirement that the article title or exact synonyms thereof appear in the sources. Cheers! --Phenylalanine (talk) 23:35, 22 September 2008 (UTC)

FWIW, I would consider a source referring to "the environmental effects of intensive livestock farming" to be close enough to the topic to qualify as a source for Environmental effects of meat production, as long as there is some evidence in the source that the livestock farming in question supports meat production (rather than dairy production only). Jayen466 23:44, 22 September 2008 (UTC)

I think another productive route is to define clearly what is synthesis, and agree about what is not synthesis. We can then craft the lead statement better.

• What synthesis is: "[Synthesis] occurs when an editor comes to a conclusion by putting together different sources, [when] sources cited do not explicitly reach the same conclusion" (from WP:SYN)
• What synthesis is not: Clarifying statements and concepts already presented in the article. Providing background and context to the reader WP:PCR. Providing information directly related to the article (according to WP:TOPIC).

Whatever we come up with should restrict the first, but shouldn't restrict the second. lk (talk) 06:14, 23 September 2008 (UTC)

I think the second part opens the gate to the addition of all sorts of "background and context" that may not be present in reliable sources discussing the topic. Jayen466 12:14, 23 September 2008 (UTC)

Yes, but that is a question for WP:TOPIC and WP:UNDUE. It is not Synthesis, which is A and B therefore C (where C is not directly supported by any source). lk (talk) 13:02, 23 September 2008 (UTC)

All content policy goes back to the neutral point of view, which is the principle that editors should neutrally present topics as they are presented in reliable sources. Following the principle of NPOV, we must be able to confirm that claims in articles arise from reputable sources (verifiability). That goes hand in hand with the principal that we must avoid making claims or connections not made by reliable references (no original research). There's bound to be some overlap, but each of the three policies elaborates on the overall principle (relying on reputable sources) in a different fashion.

TOPIC is part of a style guideline, an inappropriate place to establish content policy. UNDUE relates that information only put forward by an extreme minority of reliable sources (in relation to the article topic) should be excluded. This policy is about information not put forward by reliable sources, which makes this policy the appropriate place to address the point raised by Jayen466.

Addressing the two points you raise above, we certainly allow context and background to be provided to topics as they appear in the body of reliable sources. NPOV tells us we should present specific topics as they are treated in reliable sources. Verifiability bothers to repeat the point. This policy tells us it's inappropriate to use sources not addressing the topic and also reinforces the position of NPOV. What Wikipedia is not further reinforces the point. We don't use editors' opinions or deductions about what is accurate, relevant to a topic, or so on. We rely on the body of reliable sources to make these determinations. Presenting rebuttals, claiming information is salient and important to the topic, stating that certain topics are related and so on when reliable sources do not make those claims are clear examples of original research (and completely contrary to the underlying principles of our core content policies). Vassyana (talk) 13:59, 23 September 2008 (UTC)

You make it sound as if there are numerous other places in the Wikipedia policies that recommend that 'only sources that refer directly to the article topic by name may be used'. However, I've just gone through the policies you cite WP:V, WP:NPV, WP:NOR, WP:NOT, and I can find no such recommendation. (Perhaps you can point to specific paragraphs?) IMO, this is a proposal that will make new policy; policy that will have far reaching effects. It shouldn't be undertaken lightly. Perhaps we should even pull WP:TOPIC into the policies. lk (talk) 14:59, 23 September 2008 (UTC)

My assertion is nothing more or less than the core content principles direct us to present topics as they are presented by the body of reliable sources. A given article's name defines the topic that it covers. To assert otherwise would be plainly ludicrous and against all reason. (If an article's name does not properly frame the topic, it should be renamed.) No one is advocating that the sources must exactly match the article name. Using sources that address synonyms, subtopics, and other similar cases where it is clear the references address the article's subject are uncontroversial. No new policy is necessary. Vassyana (talk) 16:51, 23 September 2008 (UTC)

And who will decide what are "synonyms, subtopics, and other similar cases where it is clear the references address the article's subject"? I don't see that it as solving anything. lk (talk) 18:23, 23 September 2008 (UTC)

If a common understanding of English and reliable sources are not enough to define a synonym, it is likely an inappropriate relation. If reliable sources do not explicitly present a subject as a subtopic of the broader article topic, it is likely an inappropriate relation. Jayen466 presents some good examples below. Vassyana (talk) 04:37, 24 September 2008 (UTC)

(outdent)I think a subtopic is a subtopic.

• Japanese civets are a subtopic of Asiatic civets.
• Sheep keeping is a subtopic of livestock farming.
• A public figure's family members (children, notable parents) are conceivably a subtopic.

The use of sources that directly refer only to such subtopics is, we might agree, defensible. However:

• The viewpoints of orthodox Islam or Turks are not a subtopic of any individual notable muslim, or Turkish person.
• London stock market trends are not a subtopic of a given chancellor of the exchequer.
• Perceptions of reporting bias are not a subtopic of an individual press monitoring group.

Information relating to such broader topics should only be introduced in an article if reliable sources have presented them in direct relation to the article topic. Jayen466 21:17, 23 September 2008 (UTC)

_________________

After having had a chance to review this set of examples and analyses provided above, I'm in support of the use of the words "refer directly" in the relevant passage in the lead only, which would then read: "... ; to demonstrate you are not doing original research, you must cite reliable sources that refer directly to the topic of the article, and that directly support the information as it is presented." This usage plainly helps strengthen the threshold test for original research.
..... I am not in support of using the words "must refer directly to the topic of the article" a second time in WP:SYN, because it appears to me to unnecessarily confuse the WP:SYN analysis. The existing language in that section, among other things, already makes clear that:"If the sources cited do not explicitly reach the same conclusion, or if the sources cited are not directly related to the subject of the article, then the editor is engaged in original research" . This is presently buttressed by the statement: "A and B, therefore C" is acceptable only if a reliable source has published this argument in relation to the topic of the article" . In my estimation, the existing language is quite adequate in WP:SYN. ... Kenosis (talk) 22:30, 23 September 2008 (UTC)

In theory you're right, but I see a potential concern in this wording:

If the sources cited do not explicitly reach the same conclusion, or if the sources cited are not directly related to the subject of the article, then the editor is engaged in original research.

We have to bear in mind that WP:SYN is often linked on talk pages, and that not everyone will bother to read the whole section. The above phrase occurs in the lede of the section. My fear is that what will lodge in editors' minds is if the sources cited are not directly related to the subject of the article, then the editor is engaged in original research. So they will interpret this to mean, "If they are directly related, then I'm okay." And the next thought will be something like this, "But of course this is directly related! The guy is an Arab, so what the Qur'an says about women is directly related to his repulsive views about marriage!" Or, "That guy was the minister of the interior. So of course the fact that car theft doubled throughout his time in office is directly related to him!" Or, as per Jayjg, "These guys see bias everywhere. That's obviously directly related to this effect which that paper here describes, which is of crucial importance to understanding where these guys are coming from!" That's all fine and dandy if reliable sources make these points, but it isn't if they don't. Do you see what I mean? Or do you see another way of addressing this? Jayen466 23:09, 23 September 2008 (UTC)

Another way of addressing it is to point to the sentence that says "... is acceptable only if a reliable source has published this argument in relation to the topic of the article." Perhaps this sentence can be modified to read "... in specific relation to the topic of the article" . (Remember too that this was a significant part of lk's aversion to the change, though he may wish to clarify if I'm incorrect about that.) Or, the language in WP:SYN could be modified to read "... has published this argument in direct connection to the topic of the article or relevant section in the article." Etc. There are other ways to accomplish this goal of helping ensure that the connection to the material in the article is direct. ... Kenosis (talk) 23:32, 23 September 2008 (UTC)

_________________

Jayen466, that's a very nice way of putting it! If I may expand your list. In the context of this policy:

• The environmental effects of intensive livestock farming is a subtopic of the environmental effects of meat production.
• Raw veganism is a subtopic of Raw foodism
• "Some people believe that a Raw food diet is superior to a diet composed mainly of cooked foods." is part of the topic of Raw foodism
• "Some food faddists believe that raw milk is healthier than the milk you find in the supermarket." is part of the topic of Raw foodism
• "A popular dietary regimen these days is the consumption of a Raw meat diet." is part of the topic of Raw foodism
• The Large Hadron Collider is a subtopic of the Safety of the Large Hadron Collider

But (again in the context of this policy):

• Raw food and raw milk are NOT a subtopic of the Raw foodism (a dietary regimen advocating the restriction of raw foods)
• Criticism of Richard Wrangram's (critic of Raw foodism) views on human evolution and the history of cooking (which form the basis for his criticism of Raw foodism) are NOT part of the topic of Raw foodism
• Cereal grains are NOT a subtopic of the Paleolithic diet (a dietary regimen advocating the restriction of cereal grains, etc.)
• Post-Paleolithic "dietary" adaptations are NOT a subtopic of the Paleolithic diet (a dietary regimen advocating the reversion to the diet of our Paleolithic ancestors on the ground that humans are not adequately adapted to Post-Paleolithic foods.)

--Phenylalanine (talk) 22:57, 23 September 2008 (UTC)

Perhaps the participants in this discussion can find agreement that the words "refers directly to" are reasonable in the language of the lead of this policy as a basic NOR threshold question? While continuing discussion as to it's suggested use in WP:SYN? ... Kenosis (talk) 23:44, 23 September 2008 (UTC)

An alternative to "refer"

---

What does "refer" mean as it is used in the policy right now?

• "1. allude to, mention, cite, speak of, bring up, invoke, hint at, touch on, make reference to, make mention of" or
• "2. relate to, concern, apply to, pertain to, be relevant to" [12]

Clearly, it does not mean (A) "requires that the article title or synonyms thereof specifically appear in the sources", considering the "Raw foodism" and "Environmental effects of meat production" examples provided above. Nor is "refer" used in the second sense of the word (B) "2. relate to..." (since the proposal is intended to avoid "directly relates to"). But editors could nevertheless justly interpret "refer" as meaning (A) or (B) based on the two senses of that word, which is not what we are trying to convey by using "refer".

Furthermore, it is not sufficient to "cite reliable sources that refer directly to the topic of the article". One must cite sources that refer to the topic of the article in connection with the information (in the wikipedia article) that these sources are used to verify. Indeed, it is possible that a source may refer to the topic of a WP article with regards to some subject-matter that is totally unrelated to the information added to the WP article that is nevertheless verified by that same source.

Therefore, in order to avoid any misunderstandings regarding the meaning of "refer" and the intent of the policy, I suggest rewording the text as follows: "when adding information to a Wikipedia article, cite reliable sources that directly relate this information to the topic of the WP article, and in the same way that the information is presented in the WP article." --Phenylalanine (talk) 02:45, 23 September 2008 (UTC)

I think that wording could work. It would have resolved the argument with the editor in Jayjg's example at the beginning of this section, wouldn't it? Jayen466 12:07, 23 September 2008 (UTC)

---

I have indicated above why I think "refer to" is not an ideal choice of words. --Phenylalanine (talk) 00:00, 24 September 2008 (UTC)

Not to necessarily argue with you, Phenylananine, but are there any illustrations you can find where a topic that is not original research cannot include at least one reliable source that makes direct reference to the title of the article? ... Kenosis (talk) 00:51, 24 September 2008 (UTC)

No, but I don't understand the relevance of the question... --Phenylalanine (talk) 01:02, 24 September 2008 (UTC)

The proposed wording for the lead reads: "...; to demonstrate that you are not presenting original research, you must cite reliable sources that refer directly to the topic of the article, and that directly support the information as it is presented." This, at the extreme, demands at least one source that refers directly to the topic (as it doesn't specify that all sources must refer directly to the topic), and at the most lenient appears to be a bit stricter than requiring at least one source to be "directly related to the topic." Thus, the question was:"are there any illustrations you can find where a topic that is not original research cannot include at least one reliable source that makes direct reference to the title of the article?" ... Kenosis (talk) 01:23, 24 September 2008 (UTC)

That is not at all how I understood Jayen466's proposal. If this is how the policy should be interpreted, then I strongly disagree with it. Every single source used in a WP article to verify information should present that information in relation to the topic of the WP article, whether the sources present this information in relation to the main topic of the WP article (the specific concept described by the WP article title) or to a subtopic of the WP article (see examples of appropriate subtopics given by Jayen and I). --Phenylalanine (talk) 01:43, 24 September 2008 (UTC)

It seems to me that if taken in the strictest literal sense and applied to all citations, the words "refer to" can create the unreasonable expectation that each and every citation in an article will explicitly refer to the topic of the article. If taken in the more general sense of meaning "in direct relation to" the topic, then of course it makes excellent sense as to WP:NOR. ... Kenosis (talk) 02:24, 24 September 2008 (UTC)

To follow up on that, it seems unreasonable to me to demand that each and every source used must refer explicitly to the topic of the article as defined by it's title. It's more reasonable to demand that at least one source relates the information presented to the topic or subtopic of an article (note that 'How the Large Hadron Collider Works' is a subtopic of 'Safety of Large Hadron Collider'). lk (talk) 05:01, 24 September 2008 (UTC)

LK, "topic" is used here in a very general sense, such that it encompasses the specific concept described by the WP article's title as well as subtopics of the WP article, for which you provide an example. I think it's necessary that every single source used in a WP article to verify information should present that information in relation to the topic of the WP article. This requirement is important to uphold the WP:NPOV policy: to quote Vassyana, "NPOV is about accurately presenting the balance of the total body of reliable sources on a topic. "Neutrality" in NPOV is about editors being neutral in the handling and presenting of material, not about balancing the article by some subjective standard of "neutrality". Our goal is not to manufacture some pseudo-objective balance, but rather to present the information in roughly the same proportion it appears in reliable sources. If the vast majority of material about a topic is critical and negative, an article heavily weighted towards presenting critical information is not a problem. Ignoring the no original research rule would lead to a violation of NPOV (representing a view not present in the sources), not a correction towards it." --Phenylalanine (talk) 06:53, 24 September 2008 (UTC)

I would like it spelled out explicitly that topic of the article is to be understood in the broad sense, including related subtopics. (e.g. 'How the Large Hadron Collider Works' is a subtopic of 'Safety of Large Hadron Collider') Otherwise I can see people who want to see some particular infomation censored demanding that a source be removed, because it does not refer to the title by name, even if there is no synthesis (ie. there is no A & B, therefore C). lk (talk) 07:48, 24 September 2008 (UTC)

As an aside, will the people who keep on imposing the wording change please stop claiming we have consensus, until we actually have consensus. Consensus exists when someone asks, 'Do we have consensus for this?' followed by agreement from the editors involved. lk (talk) 07:54, 24 September 2008 (UTC)

Remove the disputed fragment altogether

Sources shouldn't really be the domain of this policy. Supporting statements by sources is covered by WP:V. Trying to include or exclude statements on the basis of what the sources actually refer to would (if it were implemented literally everywhere) damage the encyclopedia by restricting the information which can legitimately be presented to readers. "Directly refer" is unreasonably strong (as has been shown in numerous examples above); even "directly related" is undesirable as I see it. I propose we drop the phrase altogether and try to deal with the problems in detail under the section on synthesis. Indeed we must make a better effort to delineate the boundary between acceptable and unacceptable original synthesis (something like WP:These are not original research attempts to do). Doing it on the basis of what a particular source relates or refers to might give the right answer in many cases, but it isn't the right algorithm.--Kotniski (talk) 11:08, 24 September 2008 (UTC)

I disagree. I think it's necessary that every single source used in a WP article to verify information should present that information in relation to the topic of the WP article. This requirement is important to uphold the WP:NPOV policy: to quote Vassyana, "NPOV is about accurately presenting the balance of the total body of reliable sources on a topic. "Neutrality" in NPOV is about editors being neutral in the handling and presenting of material, not about balancing the article by some subjective standard of "neutrality". Our goal is not to manufacture some pseudo-objective balance, but rather to present the information in roughly the same proportion it appears in reliable sources. If the vast majority of material about a topic is critical and negative, an article heavily weighted towards presenting critical information is not a problem. Ignoring the no original research rule would lead to a violation of NPOV (representing a view not present in the sources), not a correction towards it." --Phenylalanine (talk) 11:19, 24 September 2008 (UTC)

I don't disagree with your viewpoint, but we must be careful how we word the policy to achieve that without doing damage elsewhere. Here's a made-up example of why even "directly related" might be harmful:

Notable weirdo A publishes a claim that hydrogen atoms have two protons. Reliable source S reports this claim (without mentioning how many protons hydrogen atoms really have). Biased editor X inserts this claim into the article on A, referenced to S. Sensible editor Y clarifies the significance of this information by inserting the statement that mainstream science gives hydrogen atoms only one proton. Editor X challenges this "claim" with a fact tag. Editor Y produces any number of sources that say how many protons hydrogen has. But naturally none of them mention or relate in any way to A or his ideas. Editor X cites this policy and removes Y's sources, and possibly her clarification as well. Y might try to invoke WP:IAR, but in any case there's going to be a lot of needless fuss, and at worst the reader is going to be deprived of this important contextual information. Note that it is not true to say "if A is notable then there must be reliable sources which mention this fact in connection with A". There may well be many sources about A in general, but only one (and one is enough) about this specific claim about hydrogen, and that particular source might not happen to mention the true number of protons.--Kotniski (talk) 11:27, 24 September 2008 (UTC)