Wikipedia talk:Primary Secondary and Tertiary Sources/Archive 4

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

Sources proposal

I have worked on a draft fleshing out the idea of what we're trying to address in this policy. There is a variety of sometimes conflicting definitions for primary, secondary and tertiary sources. Our concern should be what sources articles should rely upon, and which sources should be used with caution. This line of thought was spurred by BirgitteSB's comments about such a distinction.[1] I've taken some time to read over the talk page archives here and review what everyone has been saying, to try and take many of the opinions expressed into account. I think I have encapsulated the intent in a more direct fashion, with less confusion between primary, secondary and tertiary sources. I also believe the proposal draft better addresses the circumstances that are most appropriate for the use of historical, primary and similar sources, at least from what people have seemed to express. The draft proposal is intended to replace WP:NOR#Sources. Please take a look over the draft and express your opinion. Please tell me if this approach is workable, in your view, and why or why not. If it is workable, but the draft is flawed (as drafts tend to be), please share your criticism and concerns. I figured it was a thought worth fleshing out and discussing. Vassyana 16:41, 25 August 2007 (UTC)

I think it is too biased against primary sources. I do like the concept of "auxiliary" sources, but to me that would be for something like a blog which you wouldn't solely rely on, but might link to for informational purposes. Squidfryerchef 04:06, 26 August 2007 (UTC)

Why is it too biased against primary sources? What specific problems do you think it would run into? Vassyana 10:26, 26 August 2007 (UTC)

Again, to cite the aviation example, it would prohibit basing an article on manufacturer data on an aircraft, such as gross weight, size, range, etc. which are critical to understanding these types of subjects. —Preceding unsigned comment added by Dhaluza (talk • contribs) 12:36, August 26, 2007 (UTC)

Actually, I think this kind of concern is already addressed. I worked on language specifically to acknowledge such exceptions. (from the draft) "Some of these sources may be particularly reliable and helpful in presenting a complete encyclopedic article, such as census data. Some auxiliary references may also be useful for providing supporting facts, figures or limited quotations to accompany claims and analysis from reliable secondary sources." Does that address your concerns? If not, what needs to be added/changed to address those concerns? Vassyana 18:41, 26 August 2007 (UTC)

On the whole, I like your draft a lot better than the current policy page. It is more concise and contains far better explanations and examples. However one point I saw on my quick "look-over" is the last section on auxillary sources, with the sentence "Article claims that rely on an auxiliary source should (1) only report what the source states, the accuracy of which is easily verifiable by any reasonable, educated person without specialist knowledge, and (2) make no analytic, synthetic, interpretive, explanatory, or evaluative claims". It's probably the very rare occasion, but I think that there would be times when something like an eyewitness claim could enhance an article with an explanation that is missing in other published or secondary sources, since those are usually "space limited" (especially articles), and the additional explanation may not have been worthy enough to keep in the article with the wealth of information on that subject and space restrictions. I do agree that these "additions" should somehow be appropriately referenced, perhaps with a special "reference template" (for lack of a better term) that helps distinguish these rare occasions from the more acceptable secondary sources, but not to inhibit (or prohibit) them in their entirety. I think that the draft should elaborate a tad more on this so that some people don't read it literally as "the law", but instead will understand that it should be taken figuratively and that there are exceptions to every rule. I guess, maybe something as well that these are general guidelines for most articles, and that exceptions can and do occur. I doubt if this could be "categorized" into something along the lines of "these types of articles (science, physics, etc.) should never use auxillary sources, but these types of articles (modern history, current events, etc.) may use auxillary sources on rare occasions when a further explanation only available from the auxillary source would further enhance the article without adding conclusions or other inferences". Just my opion on one point, I'll have to re-read the entire several times and think a bit more on it. Thanks for the effort so far though, it's very good. wbfergus 13:37, 26 August 2007 (UTC)

"Using auxiliary sources" specifically refers to WP:IAR, WP:COMMONSENSE and WP:CONSENSUS in an attempt to emphasize there are exceptions. Is there something I could do to help make that more clear? Vassyana 18:41, 26 August 2007 (UTC)

(Why is it too biased against primary sources?) It implies that primary sources can only be used in very, very limited circumstances. In certain types of articles, such as the aforementioned example of aircraft, the primary sources usually have the best information, and you might have only one secondary source to establish notability. I'm of the opinion that, once notability is established, an article should cite as many primary sources as can be found, and we should encourage that. Squidfryerchef 17:10, 26 August 2007 (UTC)

This should not be a rule with exceptions, it should not be a rule at all. As has been discussed ad-nauseum above, there is no consensus for limiting the use of primary vs. secondary sources. All sources can be misused regardless of arbitrary classifications. Singling out a particular type of source is just inviting a crusade against articles using these types of sources by a misguided cabal of editors with an authoritarian personality preclusion. This is why we make policy descriptive, rather than prescriptive. To "make it more clear", address the real issue with OR, rather than trying to promulgate simplistic prescriptive guidelines that fit most but not all cases. The real issue with OR is editors confusing and conflating articles and sources by trying to make WP a source of new published information, rather than a repository of information already published in WP:RS, regardless of whether those sources can be classified as primary, secondary, tertiary, quaternary or what-have-you, including auxiliary. Dhaluza 21:33, 26 August 2007 (UTC)

See my comments above. Generally speaking, a single secondary source is not enough to establish notability. Also, as emphasized in the draft, this is specifically focusing on concerns related to original research. The current policy version is more restrictive on primary sources and arose from the tendency of primary sources to be abused for original research. I do not think a policy that's wildly permissive, and especially encouraging, of primary sources is at all in the interests of the project, which focuses on reflecting current knowledge. Permitting a heavy reliance on primary sources is contrary to that focus and an invitation to a wide variety of content abuse, notably including original research. Most people who object to the current formulation are concerned that the policy will prohibit people from using primary sources in some select circumstances. I tried incorporating that (apparently) consensus concern. Two such "exceptions" that seem to be brought up frequently (which I took into account in the draft) are using particularly reliable data, such as aviation figures and census reports, and using primary sources to complement/illustrate secondary sources. Additionally, it explicitly encourages people to employ WP:IAR, WP:COMMONSENSE and WP:CONSENSUS to identify specific exceptions. Vassyana 18:41, 26 August 2007 (UTC)

If by "generally speaking" you mean in most cases, then yes, multiple secondary sources are usually used to establish notability. However there are many articles where one or no secondary sources are cited, needed, or even available. To take something that is mostly true and make it a requirement is not only unnecessary, it is contrary to the whole idea of what a policy on WP is supposed to be. Regardless of how sound an argument may seem, or how convincing a case you may make, WP policy is supposed to represent actual good practice, not what people wish good practice would be. To some extent WP:NOR is an exception as a core policy, as it expresses a basic tenant that all articles must meet. But taking this and shaping the definition of OR so narrowly that it only allows secondary sourcing except in the most extreme cases would create a bias against producing good articles in a host of subject areas where hard facts are more important than opinion. Dhaluza 21:17, 26 August 2007 (UTC)

If there are no reliable secondary sources available, it is impossible to make an encyclopedic article without original research and therefore is an inappropriate topic for Wikipedia. I have yet to hear any convincing case where primary sources are needed, only convenient. Even census statistics, aviation data and the other examples the draft would explicitly permit are not needed since there are copious secondary sources that include such information. However, it is good practice (and convenient) to source the figures directly. The proposal is nowhere near as restrictive as you'd try to make it sound. The draft explicitly allows auxiliary sources, including primary sources, to be used in a broader context than the current formulation would allow or imply, in line with "actual" good practice and a review of this talk page's archives. I'd also direct you towards some of BirgetteSB's insightful and eloquent comments, which may help further clarify the issue.[2][3]][4] Vassyana 23:03, 26 August 2007 (UTC)

True, if there are no secondary sources, the subject is possibly not notable. However, once the threshold of notability is passed, there's no good reason for a general rule requiring verifiability only to secondary sources except in "rare cases". There are pros and cons for using both primary and secondary sources: primary sources are generally more accurate and less filtered, but sometimes lack context. Secondary sources are generally less accurate and more biased, but sometimes provide context, although that context is sometimes biased. It's all case-dependent, and a general rule that primary sources should be "rare" and secondary sources "primarily relied upon" is neither good practice, nor a good description of current Wikipedia practice. COGDEN 19:56, 30 August 2007 (UTC)

(Answering points raised since) A single secondary source can be enough to establish notability. Secondary sources aren't required for articles that are inherently notable, such as many geographic topics. It is very easy to use primary sources without slipping into OR. I don't understand where all this is coming from. We're not here to rewrite WP:RS. Squidfryerchef 23:18, 26 August 2007 (UTC)

Nothing I've said here is contradictory to WP:RS or WP:V#Sources (on the contrary, I think the proposal would bring the policies into better harmony), so I'm not sure why you get the impression we're revisiting reliable sources. Regardless, it's nearly impossible to build an encyclopedic article based entirely on primary sources without liberally engaging in original research. We absolute need secondary (including tertiary) sources to provide the context, analysis and interpretation of those source necessary, so a proper article can be built. Otherwise, the article either violates this policy or it's just a collection of facts. Either way, sole reliance on primary sources in highly inappropriate. Vassyana 23:31, 26 August 2007 (UTC)

I disagree that you can't write a good encyclopedic article based entirely on primary sources. It is true that you can't usually write an article without at least some secondary sources, but the reason you include the secondary sources is for the primary source material found within them. For example, if you're writing an article about the Mountain Meadows massacre, you will certainly cite secondary sources such as 20th century books by historians such as Juanita Brooks. However, the main reason you cite Brooks is to present her personal views, analyses, and conclusions concerning the massacre, which is primary source material—her original research. Sure, you can also cite Brooks when she acts as a pure secondary source by describing the contents of people's diaries, etc., but you could just as easily quote the diary itself, if it's published. And if Brooks' reading or paraphrasing of the diary is controversial, you almost have to quote the unfiltered language of the original published diary. COGDEN 19:56, 30 August 2007 (UTC)

WP:RS is what should define the various classes of sources, not WP:NOR. We should give editors a wide berth to include the kinds of sources they feel appropriate. Please remember that an article about a mass transit system can use primary sources to a greater extent than an article about religion. Squidfryerchef 23:46, 26 August 2007 (UTC)

RS does indeed address sourcing, along with V#Sources, but sourcing in relation to original research is indeed relevant here. RS is even built using WP:PSTS, imported from this policy. You left the remainder, and main thrust, of my reply unanswered however. Vassyana 00:04, 27 August 2007 (UTC)

I agree with Squidfryerchef: we don't need to define the different types of sources here. All we need to say is, whatever kind of source you're using, don't add your own interpretation. --Coppertwig 00:06, 27 August 2007 (UTC)

There is an operative difference between the types of sources, particularly in relation to this policy. If the use of a certain class of sources tends toward original research, or that such abuse reached such an acute level that it required a policy change (both true), there's obviously a need for the policy to address those distinctions. Vassyana 00:13, 27 August 2007 (UTC)

You keep beating the primary sources inevitably lead to OR drum, but that simply is not supported by a broad consensus, nor is it true as you assert. It is possible to write a completely factual article using only primary source facts. There are thousands of examples created by a bot using census data. In some subject areas, editors may use primary sources to push their POV, but editors can use secondary sources to push POV too. The problem is POV pushing, not sourcing. Focus on the problem. Dhaluza 00:38, 27 August 2007 (UTC)

(outdent) Please explain how my assertions here are incorrect.[5][6] Also, those bot created "articles" are not encyclopedic articles and if left untouched are suited for Wikisource, but not for Wikipedia. Vassyana 01:03, 27 August 2007 (UTC)

Vassyana, your proposed changes are very appropriate for articles about ideas ( religion, humanities, scientific theories ), but I would not like to see them applied to articles about things ( including organizations ). Suppose I begin a stub about Yoyodyne, Incorporated. I cite the company's web page that says it was founded in Pittsburgh in 1947. That's a primary source, but there is no O.R. Squidfryerchef 01:23, 27 August 2007 (UTC)

Supporting facts are explicitly acknowledged and permitted in the draft proposal. ("Some auxiliary references may also be useful for providing supporting facts, figures or limited quotations to accompany claims and analysis from reliable secondary sources.") However, such an article could not principally rely on such sources without conflicting with NPOV and potentially Wikipedia's fundamental nature. Is there some adjustment that could be made to better accommodate your concerns, or more clearly express the permissibility of supporting facts from auxiliary sources? Vassyana 01:30, 27 August 2007 (UTC)

I should also note that the proposal specifically takes into account WP:SELFPUB. Vassyana 01:34, 27 August 2007 (UTC)

Vassyana, you seem to assume that the date the company is founded is not a principal fact and that an article consisting of that sort of fact is not "encyclopedic". I disagree. --Coppertwig 13:16, 27 August 2007 (UTC)

This is a fairly acute distortion of what I'm saying. Of course, it is a basic fact (though probably not a principal one) and such facts should be in Wikipedia articles. However, Wikipedia articles are not merely collections of such facts. Third-party sources are required to provide context and analysis. Vassyana 17:09, 27 August 2007 (UTC)

Would the company's own website necessarily be a primary source for the company's founding date? The primary source for the actual date of founding might be the court papers filed when registering the company with the government. The company's website might be a secondary source for that particular information.

This is besides the fact that some information from primary sources is allowed, that's not even a question. The question is how many articles rely solely on primary sources. I'd say very, very rarely. The examples provided so far on airports, runway sizes, etc, don't rely solely on primary sources - but they can use primary sources for that information if necessary. It's all covered by policy. We want to make it very clear that articles should not rely solely on primary sources. Dreadstar † 17:59, 27 August 2007 (UTC)

I think you just described almost all the "episode pages" for a TV show. Corpx 18:04, 27 August 2007 (UTC)

Sadly true. Even worse, there's no need for it. TV Guide and similar publications, as well as the huge number of industry periodicals, pop culture magazines and related references, render any "need" for the watch-and-write approach null and void. Vassyana 19:37, 27 August 2007 (UTC)

I think some people need to be reminded that the policy in no way forbids the use of primary sources, it only forbids the use of primary sources to make original interpretive, explanatory, or synthetic claims or generalizations. For example, if you use the IBM website to say that IBM was incorporated in 1911 - well, this in no way violates the policy. But if you say that it was because IBM was incorporated in 1911 that Sun Yat Sen was elected president of China .... well, you better have a secondary source for that!! Slrubenstein | Talk 18:13, 27 August 2007 (UTC)

I think the proposal is really does nothing substantial other than re-label primary source as auxiliary source. I also disagree with some of these statements:

• "secondary sources are preferable because they generally provide analysis, offer a more independent view and provide a broader context for the subject". I don't think this is a valid statement across all sub-communities of Wikipedia. First, in political, historical, and religious contexts, secondary sources are more likely to be biased than primary sources, because the writer is citing the primary source to prove a point, and will interpret the primary source in a way that favors the point. In such cases it is imperative, if one is to follow WP:NPOV, to cite the unvarnished language of the primary source. Second, although some secondary sources provide context to primary sources, they also shape the context in a way that supports their point of view.
• The definition of reliable secondary source is not compatible with general usage outside of Wikipedia. Particularly, journal articles containing original research are generally considered primary sources, unless they are an overview of prior research by people other than the author.
• "Reliable secondary sources should be the principle reference material for Wikipedia articles." While this may be justifiable for some Wikipedia articles, this statement is not generally applicable Wikipedia-wide. For many articles (perhaps even most, for some fields such as celebrity articles and articles about cutting edge science) articles, the principle reference material should be primary (aka "auxiliary") sources. For example, the following excellent articles, among countless others, rely primarily on primary sources, and are much the better for it: Sunshine (2007 film), Deinonychus, Baby Gender Mentor, Same-sex marriage in Spain, Mendip Hills, and Ehime Maru and USS Greeneville collision. Featured articles are actually more likely than non-featured articles to rely primarily on primary (auxiliary) sources, which has to tell you something.
• Classification of "Historia ecclesiastica gentis Anglorum" as an "auxiliary" (aka primary) source doesn't make sense. It's actually, for most purposes, a secondary source. Like many secondary sources, it's not in most experts' opinions a particularly accurate or objective one, but it's still a secondary source, and it's a "reliable source" according to the definitions of WP:Reliable sources found in the guideline until very recently.
• The "rule" for using "auxiliary" (aka primary) sources is applicable to secondary sources as well.

COGDEN 20:41, 27 August 2007 (UTC)

Actually, the differences are much more significant and plainly articulated. The distinction between secondary and tertiary sources is rejected. The category of auxiliary sources is significantly more broad than primary sources.

• This is a gross misrepresentation of what the draft says and what NPOV is about. The proposal simply asserts they are more independent, not necessarily less biased. NPOV requires a presentation of all notable points of view, not some mythical objective neutrality.
• The proposal specifically addresses that differing fields have differing approaches to source classification and clearly provides a definition for usage.
• Your examples do nothing to contradict the position of the proposal draft. Sunshine (2007 film)#References and Deinonychus#References clearly show a dependence on reliable secondary sources as defined by the proposal, and generally defined by the Wikipedia community.
• This shows a deep misunderstanding of how primary sources are defined and used in real world academia. Bede is a primary source, much like Tacitus (who I demonstrated above is treated as a primary source). Historical sources are generally considered primary sources by historians. Also, claiming that experts say Bede's history is not accurate or objective while stating that it is a reliable source (and therefore suited for direct referencing) is an utter contradiction.

Vassyana 23:40, 27 August 2007 (UTC)

It's not a misunderstanding. It's true that sources like Bede are cited mainly as a primary source, but that doesn't change the fact that it's also a secondary source. Nobody cites it as a secondary source anymore, because it isn't accurate. But it does give insights into the medieval mind and culture, and in that sense it's a primary source. It's clearly a reliabile primary source, as well, because it's cited as a primary source by peer-reviewed scholars. Basically, it's a somewhat unreliable secondary source, but a reliable primary source. COGDEN 19:56, 30 August 2007 (UTC)

There's clearly some misunderstanding going on, as evidenced by the points you did not address. Part of your misunderstanding seems to derive from a divergence from the practical definition of sources, while obsessed with the "dictionary definitions" taken out of context. Bede is not a secondary source except in a strictly semantic sense. Also, your comments have solidified my opinion that you have a grossly distorted understanding of what "reliable" means, both for Wikipedia's purposes and as a plain definition. Source material, even in peer-reviewed articles, is not necessarily reliable. There are entire books and articles written about the reliability, or lack thereof, of individual historical sources. Again, even Caesar's Gallic War, seen as a pinnacle of ancient reporting, is also acknowledged as a masterpiece of propaganda and noted as containing inaccurate hearsay. Simply drawing upon the source material makes no judgment of its reliability or accuracy, and to assert otherwise is either dishonest or ignorant. Vassyana 20:19, 30 August 2007 (UTC)

The meaning of "secondary source" doesn't have to be so complicated. Bede is a secondary source to the extent he reviews and incorporates other pre-existing material. His editorializations and his fictional quotations, of course, are primary material. Bede was essentially peer reviewed during his time, but he's not viewed by modern historians as an entirely reliable secondary source. But he's a great primary source, and a completely reliable source as to his own opinions and insights. He can be cited as a reliable primary source, but a questionably-reliable secondary source. COGDEN 21:22, 30 August 2007 (UTC)

There is no way to distinguish between the reliability of various points in his material in the absence of reliable third-party references without engaging in original research. Vassyana 21:32, 30 August 2007 (UTC)

But there is no absence of reliable third-party references on Bede's credibility. If you said that Bede is not entirely reliable, that would not be original research, because it's not an original unpublished idea. It's not even really a controversial issue, and since Bede's looseness with historical facts isn't likely to be disputed, a primary source saying "My conclusion is that Bede was loose with history" or a secondary source saying "Scholarly consensus is that Bede was loose with history" is probably optional under WP:V. COGDEN 22:45, 30 August 2007 (UTC)

You just expressed exactly why articles need to rely on reliable third-party publications. Vassyana 22:56, 30 August 2007 (UTC)

If by "rely on" you mean "cite", that's not true. If "rely on" means "they exist", then maybe you have a point, but that's not a standard meaning of "rely on". COGDEN 19:03, 31 August 2007 (UTC)

OK, here is another example of an encyclopedic article written almost entirely from a single primary source, an NTSB factual report: Schempp-Hirth Nimbus-4. With the exception of one sentence on probable cause for one accident, the whole article quotes primary source data transcluded from a reliable public domain source (the probable cause represents the only interpretation of facts, and is therefore secondary source). This source is hardly "auxilary" it is the basis for establishing the detailed facts on the design and construction of the aircraft. Dhaluza 10:03, 28 August 2007 (UTC)

That report is not a primary source. It compiles numerous eyewitness accounts and compares and analyzes them. It also compares the accident with other accidents involving the same sort of aircraft. One entire section is labeled "Analyis" and there is a probable cause finding. This is not raw data. It is a secondary source.--BirgitteSB 14:39, 28 August 2007 (UTC)

Like many sources it contains both primary and secondary source material as you point out. But as I said in my comment above, all but one sentence came from the primary source material, and that one sentence is not the basis of the article by any stretch. The NTSB report does compile a list of other accidents, but I would not say it compares them in the sense of a critical review that would qualify as a secondary source. It simply lists the other accidents on record with a summary in the part of the report used in the article. So this article is based on a single primary source as I said. Dhaluza 23:14, 28 August 2007 (UTC)

Beyond what was said about the mislabeling of the source, this article is a perfect example of why multiple reliable secondary sources are needed. Such an article is contradictory to numerous policies and guidelines. The first half of the article is a bombardment of facts without context or explanation. It then sharply and immediately shifts gears to report the accident history. However, it does nothing to explain why the information is notable or of interest. It gives no explanation as to why the investigation was undertaken in the first place. (It mentions the accident. But, was the investigation launched because there were suspicions about the manufacturing of the craft? Was it launched because there was a public outcry? Was it launched because the engineering of the craft was believed to be potentially flawed? Why was the investigation started? What spurred the start of the investigation?) The article additionally makes no assertion of notability, which is a central requirement. The lack of context and lack of notability assertion qualify such articles under two speedy deletion criteria. (WP:CSD#A1 and WP:CSD#A7). As a collection of blunt facts without context, the article also is an example of what articles should not be. How is NPOV being served here, since it requires an overview of all notable points of view? Does the bulk of reliable secondary sources support such an emphasis on the accident report, or is this undue weight? In the absence of such sources, that article matches negative criteria under the criteria for speedy deletion and "what Wikipedia is not", and NPOV cannot be fulfilled. Vassyana 15:44, 28 August 2007 (UTC)

The source material used in the article meets the definition of primary source (but for one sentence) as explained above. As for your assertion that since the article does not anticipate all of your questions, and provide ready answers for them, it should be deleted, may I refer you to WP:IMPERFECT, and of course WP:NOR. We are not supposed to go beyond the sources, so explanatory claims, or speculating on the reason the NTSB took on this investigation to answer your inquiries would be OR (but, if you want my opinion, the unusually detailed investigation was done in this case because a former FAA Chairman was killed, but that's not relevant to an aircraft type article anyway). The facts are indeed blunt, but without context? Are you serious? They describe the design and construction of the aircraft, including how different materials are used differently on different parts. This is relevant in the context of what the aircraft is. It is of interest not only to someone interested in this type of aircraft, it is also relevant to composite aircraft in general, and the application of composites in industry. As for the rest of your laundry list of complaints, what points of view are relevant to an aircraft? It is what it is. The emphasis is not on the accident report--that document is much longer and discusses many topics. Just the topics relevant to the subject of the aircraft type have been transcluded, and yes the accident history is relevant. As for notability, first the article clearly states that this aircraft was certified to JAR 22 by the LBA, entered serial production, and was produced in significant numbers. Besides that, there is the obvious fact that it was involved in a notable accident. Also if you dig a little deeper, in the what links here, you will find that the world distance record for gliding of 3,008.8 km was set in this aircraft type. So while your suggestions for improvement are valid as far as that goes, you take it much too far by suggesting that since the article does not meet your high standards, it should not exist. And your assertion that it qualifies for speedy deletion is not even close to being supportable. By taking this example to these extremes, you undermine your repeated arguments favoring devaluing primary sources. Dhaluza 23:14, 28 August 2007 (UTC)

I did not say those questions had to be answered for the article to be sufficient. They were examples of what kinds of questions would provide context. Of course, trying to answer questions that sources do not address is original research, and contrary to your implication, I did not even imply that OR should be engaged in. Quite the opposite, I explicitly said more reliable secondary sources were needed. The case was unusually detailed because a former FAA chairman died and there's no reliable source that covers that? That seems more than a little unlikely. Yep, the facts are completely without context. Technical details (raw facts) do not provide context, no matter how detailed. You claim it's relevant to related topics, but the article presents no such explanation. Why is it important to know this information? How does it compare with crafts of its type? In other words, what makes this information suited for a Wikipedia article, instead of Wikisource? You claim it's relevant to related topics, but the article makes no such explanation and presents no evidence to that point. Plenty of points of view are relevant to aircraft. What audience uses them and why? What are seen as the positive points of the design and production? The negative points? Just a few questions to start fleshing out the views required by NPOV. The questions regarding context are also relevant, since they present points of view. You can claim the emphasis is not on the accident report, but it accounts for half the article. I realize a certain amount of space is necessary to discuss that subtopic, but is it that much a part of the coverage that it should occupy roughly half the article? Being simply certified as a worthy craft and produced does not equal notability. Is it a notable accident? According to what reliable sources? Where's the substantial coverage? If it set a world record for gliding distance, why is this not in the article and cited? It certainly would be an assertion of notability. If even a small bit of context were provided, along with that notability assertion, it would be well outside the bounds of WP:CSD (A1 and A7), but as the article currently stands, it clearly qualifies. Also, without context and commentary (that is, just a shotgun blast of raw facts), the article clearly violates WP:NOT#INFO. Vassyana 23:58, 28 August 2007 (UTC)

So what is your point? That the article could be improved with additional secondary sources? No argument there. That these sources are necessary? No, the article is fine as far as it goes, and removing it would not improve the encyclopedia as a whole. Dhaluza 00:17, 29 August 2007 (UTC)

The point is you held it up as an encyclopedic article, which it's plainly not. It's not compliant with Wikipedia policies and principles, as elaborated above. It's a terribly poor example if you intended to defend a primary source friendly position, since it is so problematic in its current form that it qualifies for speedy deletion. Is deletion the answer? Probably not, in this case. If your claims about the record and the relevance of the details to related topics are true, certainly not. Though, if you are aware of this information and sources for those facts, you should add that information with references. If you're aware of the information, but not the sources, it would probably be wise to tag the article as needing expert attention or (better yet) leave a message on the aviation project's talk page asking for help in finding sources. Regardless, this will be my last reply on this, as we're getting far off-topic here. If you so disagree with Wikipedia policy (such as WP:V#Sources, WP:NOT, WP:CSD and the requirements of WP:NPOV), as seems to be the case here, I would suggest you address those issues at the appropriate talk pages. Vassyana 01:59, 29 August 2007 (UTC)

I do not disagree with those policies, but I do object to your dismissal of my arguments by assuming I do. And your continued assertion that it qualifies for speedy is clearly not consistent with that policy as written and applied. Dhaluza 10:48, 29 August 2007 (UTC)

(outdent)It's only rated as a start-class article, quite appropriately. It lacks context for the facts, contrary to WP:NOT, WP:NPOV and WP:CSD. It lacks an assertion of notability, contrary to WP:N and WP:CSD. It's sourcing is quite inadequate, contrary to WP:V#Sources and WP:NPOV. By insisting such an article is "encyclopedic" and an example of how a good article can rely on primary sources, while shrugging off those failures, it certainly gives the impression you disagree with general policy, or would like them to say something quite different than they do currently. Vassyana 18:19, 29 August 2007 (UTC)

"in political, historical, and religious contexts, secondary sources are more likely to be biased than primary sources" So? You miss the point. Let's say alll primary sources are unbiased - the problem that this policy addresses is that if an editor uses a primary source to make a novel interpretive, explanatory or synthetic statement, it is the editor's bias that enters the article which is strictly forbidden by NPOV. NPOV does not insist that we use unbiased sources, indeed, NPOV is based on the principle of "verifiability, not truth." Of course our sources are biased. So what? All the better! Let them be biased! NPOV means representing multiple, diverse, even opposing views. We just need to identify the view and put it in its context, but all this is in our NPOV policy. NOR is about keeping our views out, and to do that we have to prohibit editors from using primary sources to forward their (editors) own arguments. period. Slrubenstein | Talk 21:41, 27 August 2007 (UTC)

I agree, but my point was that source bias is not an argument in favor of officially discouraging the use of primary sources. This is one of the main reasons put forward for the proposed decree against primary sources: that some editors have been using primary sources to promote an agenda or bias. COGDEN 19:56, 30 August 2007 (UTC)

As for scientific and historical sources, I think it would help to encourage the use of review articles (not book reviews, but general overviews of the recent research) and similar sections in the more specialized reference books, but not require it. Jacob Haller 22:26, 27 August 2007 (UTC)

That would be an appropriate subject for an essay, or eventually a guideline, but not really a main Wikipedia official policy. COGDEN 19:56, 30 August 2007 (UTC)

I'm not convinced that primary sources are generally/always/usually more liable to have Wikipedian interpretations added to them than secondary sources, especially if peer-reviewed scientific articles are considered to be primary sources, but also in any case. I think Vassyana is using a definition of "encyclopedic" which doesn't agree with either my dictionary or my idea of what Wikipedia (or any other encyclopedia) is or should be. The Schempp-Hirth Nimbus-4 article seems to me (at a glance) to be a good, encyclopedic article. Maybe we need to create a new category for ordinary peer-reviewed scientific articles, or get consensus on whether they're primary, secondary or a combination of the two, before carrying on much discussion using the terms "primary" and "secondary" when it's not clear what's covered under each. Similarly, clarification is needed as to whether a journalist's report of a traffic accident, for example, is a secondary source as stated on this page. --Coppertwig 16:58, 29 August 2007 (UTC)

In the proposal I drafted, I treat peer-reviewed/academic articles as secondary sources, based on WP:V and WP:RS, which emphasize the reliability of such sources and treats them as "third party". The draft also notes that some auxiliary sources may be considered especially reliable and useful. It additionally reminds people to use common sense and do what's necessary for the best of the encyclopedia. WP:V and WP:RS would indicate that reputable newspapers would be considered a secondary, or "third party", reliable source. Also, the dictionary definition of encyclopedic is "broad" (as in encyclopedic knowledge), or "of encyclopedia quality" (or similar self-referential variation). Outside of very technical or specialist encyclopedias (and even rarely then), articles are usually presented in context with narrative flow, not as a raw blast of facts lacking those features. Vassyana 19:58, 29 August 2007 (UTC)

My long held belief has been that we should dispose of the terms "primary" and "secondary" completely because of the lack of clarity you mention. An attempt at doing just that was short-lived due to objections, and do not believe the idea is anywhere near achieving consensus now. I watch for this topic to come up and try to contribute productively to making the issue clearer when it does. I don't know the ultimate answer, or if we will find this round or in the round six months from now. But please, please do not let yourself believe there not real reason for this discussion. That so many different people have for so long put effort into describing this problem, should be evidence enough that there is an underlying issue with certain types of sources being more liable o have Wikipedian interpretations added to them than other types. I make a real effort to explain these issues to the new people who show up around this topic instead of insisting they read the archives themselves. But it ridiculous at this stage of the recent discussion to still be "debating" whether the problem even exists. Anyone who is still "not convinced that primary sources are generally/always/usually more liable to have Wikipedian interpretations added to them than secondary sources" needs to read the archives and see for yourselves the variety of problems that have been outlined on this issue. Here is a good start and work forward in time. Also when done with that time period there is more in the regular archive of this page and in archive of the talk page at WP:ATT--BirgitteSB 20:33, 29 August 2007 (UTC)

I'm gradually beginning to agree with you on just disposing of the terms "primary" and "secondary" sources completely. I don't think it adds much here, and everybody has a different view on what the terms mean. Really, original research is original research, regardless of the type of source, and that's the main point of this policy. COGDEN 19:56, 30 August 2007 (UTC)

Well, it seems clear that there has always been a belief that secondary sources, however defined, should be the basic building blocks of articles. Let us assume that consensus on this issue existed in the past, even if there were disagreements on the definitions of these sources and the degree of their suitability, and move on. Hornplease 21:16, 29 August 2007 (UTC)

I don't think there has ever been such a consensus. If "secondary sources" does not include things like peer-reviewed journal articles, interviews, novels, movies, autobiographies, press releases, or government reports, then your statement above would require revolutionary changes to the entire Wikipedia culture, and a dramatic revision of a vast number of Wikipedia articles, including most featured articles. COGDEN 19:56, 30 August 2007 (UTC)

I would make some, admittedly major, changes:

• I think there is general agreement that peer-reviewed scientific research papers are reliable sources, but that some are often considered primary sources.
  • I personally think there should be a special note including peer-reviewed review articles, i.e. overviews of recent research in the field, as reliable secondary sources.
  • I think most other peer-reviewed research papers can be considered "reliable primary sources"
• I would add another category describing limited sets of "reliable primary sources," not regarding them as either secondary or auxiliary sources:
  • Most peer-reviewed research papers (in both science and the humanities).
  • In biography articles, an author's own works are suitable sources for his/her beliefs.
  • In political articles, a movement's foundational works and its internal commentaries are suitable sources for its beliefs.
  • In religious articles, a movement's internal commentaries are suitable sources for its own beliefs, but its sacred texts are not (because sacred texts often have multiple interpretations).
  • (In historical events, there is ongoing debate over how often to use eyewitness accounts.)
• I would add a third category discussing misuse of otherwise-reliable sources, e.g.:
  • Reading too much into passing comments
  • Reading too much into statements which go against the grain of the argument
  • Drawing analytic conclusions from definitions

I would also consider treating this as a guideline, not as policy, as you had suggested for the current text. Jacob Haller 22:34, 29 August 2007 (UTC)

Very sensibly laid out: I disagree only with one aspect: most peer-reviewed research papers in the social sciences - political science, in particular, economics to a degree, development economics most certainly - are in effect peer-reviewed survey articles, in that they frequently serve to synthesise other sources; thus should they be considered reliable secondary sources. I'm not sure I would be comfortable with any categorisation that did not recognise that. Hornplease 22:48, 29 August 2007 (UTC)

These articles are indeed secondary sources to the extent they review prior source material. However, as to their original conclusions and ideas based on these prior sources, they are primary sources. Anyone who is an original source of some new idea is the primary source as to that idea. COGDEN 19:56, 30 August 2007 (UTC)

I think Jacob Haller's comments are a good starting-point for an essay, which might eventually become a guideline. COGDEN 19:56, 30 August 2007 (UTC)

Advice

The policy page is protected - the talk page is a mess: there are many different suggestions/proposals, and also different argumnts, that are spread through a variety of sections and discussion is still going on in many different sections. I respectfully suggest to those people most active in these arguments - people on all sides - to pause and sort out (1) the major points of contention and (2) the major proposals, and then to archive the curretn talk, and start the talk anew with specific sections fot the major points of contention and proposals. I am not asking for a straw poll or any decisions, I am just suggesting a way to clean up the talk page. It will make it much easier for new people entering the discussion to follow it, and it might actually make it easier for people here to work through one argument, one proposal, at a time and make progress. This suggestion is not about deprecating anyone's position, just laying out the conflicts more clearly. I see people making the same points in three or four different sections on the same day. That is not good. The points may be fine, but scattering them in several sections doesn't do anyone any good. At this point people who have been active for the past week should not find it hard to refactor the debate in a way that includes all sides but in a clearer, more orderly fashion. Slrubenstein | Talk 10:47, 30 August 2007 (UTC)

Someone asked me to comment on these issues, but I can't work out who is saying what either. For my own part, I preferred the wording of the previous version to the current one: Dreadstar's version in this diff. [7] That is, as a rule, articles must rely predominantly on secondary sources. This reflects best practice. SlimVirgin ^{(talk)(contribs)} 11:33, 30 August 2007 (UTC)

Is this edit change the only issue under debate? If so, let's have just one section to debate it. But if there are other issues, let's sort them out and line them up neatly so we can keep different arguments straight, before continuing the debates. Slrubenstein | Talk 11:55, 30 August 2007 (UTC)

I concur - this page is so convoluted it is difficult to tell who is arguing for what. Dreadstar's version looks good to me, if nothing else for a starting point. What objection(s) might there be to that? KillerChihuahua^?!? 12:19, 30 August 2007 (UTC)

To me, it seems like the three main points of contention are the following:

1. Should the sentence be Wikipedia articles should rely on reliable, published secondary sources. or should it be Wikipedia articles should rely on reliable, published primary and secondary sources.?

2. Should the (second following) paragraph start with Although most articles should rely predominantly on secondary sources, there are rare occasions when they may rely on primary sources.? (I think this would depend on #1 above)

3. In the same paragraph as #2 above, which version of the sentence should be included, and (2) make no analytic, synthetic, interpretive, explanatory, or evaluative claims. or and (2) make no analytic, synthetic, interpretive, or evaluative claims. (explnatory is missing from the second version).

wbfergus 13:06, 30 August 2007 (UTC)

My opinion: on 1 and 2, I think it is better just to delete from the policy any statement of preference. Instead of urging people to favor one kind of source over the other, it should be sufficient to say that primary sources can be used as long as they are verifiable, and are not being used to ..... On 3, I think it is crucial to keep the word "explanatory" in.Slrubenstein | Talk 13:20, 30 August 2007 (UTC)

(ec)I have been mostly follow all this and believe we really have only half a point of contention. Those that wish to change the policy, believe that the policy needs to be changed to sopport the current practice of using primary in Feature Articles etc.[8] The opposing parties support the current practice of using primary in Feature Articles etc., but believe it is already supported by pre-dispute policy[9]. So there is not disagreement of any real merit regarding practice of this policy, but simply a disagreement over what the wording in the policy means.--BirgitteSB 13:13, 30 August 2007 (UTC)

True, but the half-point is a source of contention. As Hornplease indicates above there are those who believe the polices are and should be restrictive. That may be so but the restrictions should be uniform and not imposed and enunciated just for some types of source. Those relying on secondary sources surely should "make no analytic, synthetic, interpretive, or evaluative claims" (of their own.) Why single out those who use primary sources as the recipients of this injunction and exempt those who use secondary sources? I think you do believe that the injunction is uniform for users of all sources. Why not have the policy reflect that uniformity? The existing policy may permit use of primary sources but it puts such use under a cloud by targeting only users of such sources with specific restrictive injunctions. The restrictions are for all editors using all sources. Why not have a policy that says so? --Minasbeede 14:32, 30 August 2007 (UTC)

I agree with you, Minasbeede, that the restriction applies to secondary sources as well. The last time I worked on this page was a year ago. At that time, the section that introduced "secondary sources" ended with this statement: "Wikipedia articles should rely on reliable, verifiable, published secondary sources wherever possible. This means that we present accounts of views and arguments of reliable, verifiable scholars, and not the opinions of Wikipedians who have read the primary source material for themselves." In other words, there was a time, not too long ago, when the policy page had precisely the uniformity you see lacking. I personally would not object to restoring the passage that had been removed, but I wonder if this is all that is holding up unprotecting the page? I have to defer to those of you who have been active on this page from the time it was protected, for an account of the issues holding things up. I know what a few of you think, now, and I appreciate that. I wish you, Birgitte, wbfergus, and whoever else has another view, could summarize the issues concisely in one place - and I wonder if you could all agree as to which is the core problem, or prioritize the issues, or do something to put these points in some logical order Slrubenstein | Talk 14:53, 30 August 2007 (UTC)

Thanks, but I still don't see the need to single out primary sources and those who use them. Personal opinions and interpretations are forbidden, period. The prohibition only hits those who violate it. Whether that's 1% of the users of primary sources or 99% of them they're hit if they do it. Why doesn't that suffice? It's the behavior (personal opinions, interpretations) that's at fault, not the type of source.

There could be a long, exhaustive (or an attempt at such) essay on all the possible misuses that could occur for the users of different types of source. There is a prohibition on OR. Isn't that prohibition fully adequate without singling out particular types of source or trying to list all the possible misuses?

Looks like I have but one active point: don't single out one kind of source in the policy for special prohibitions (particularly when those prohibitions are blanket prohibitions anyway.) I have an inactive point, but since it's inactive it needn't be mentioned. Keep it simple. --Minasbeede 15:26, 30 August 2007 (UTC)

Well, there are other people who have been very active here. I urge you all to narrow it down to no more than four points of contention, create new sections to discuss each one and consider proposals, and archive the current talk which is all over the place and confusing. As I said, my aim is not to stifle any view, but to bring some order to the page so that every view can be clearly grasped. Slrubenstein | Talk 13:20, 30 August 2007 (UTC)

I do not understand what you are attempting. I outlined the issue that caused the protection. Whatever side-threads have coincidentally taken place at the same time are unimportant. It is silly to seriously discuss whether or not we should allow original research because some one suggested that while other parties involved themselves in an edit war. I don't have four points of contention. I think the version supported by SV above has consensus and while not all the edits to the policy the past week have been harmful, none have been a significant improvement over that version. My only contention is that I have not seen any significant improvement in the policy edits. If I had I would have supported the change. As I have not I have not supported the changes. It is that simple.--BirgitteSB 13:50, 30 August 2007 (UTC)

I am attempting to move the discussion towards a resolution of the conflicts. I have watched this talk page for the past several days and do not see that happening. I agree with you that it is silly to discuss whether or not we should allow original research, which is why I never suggested such a thing. But it appears that you have taken my comment personally when I was not addressing it to you personally but to all active editors involved in debates on this page. I understand that you believe there is only one issue. I never said you personally have four points of contention. Indeed, I never said there were four points of contention. But reading over this talk page I see many more than four threads and conflicts, and I was simply suggesting that editors on this page try not to tackle more than four at a time. That is a general point and I think it reasonable. Now, if most people here agree that there is only one issue, then lets just focus on that one issue. But another editor listed three issues. If many editors here disagree with you, Brigitte, over how many major points of contention there are, then unprotecting the page right now would be imprudent. I'd like to see it unprotected, and that means sorting out disputes on this page. You ask what i am attempting and I repeat for the third time: I am suggesting that editors agree on the major point or points of contention and deal with them systematically and seek resolution, rather than have thirty threads on the page going in different directions. Why is this a bad idea? Slrubenstein | Talk 14:03, 30 August 2007 (UTC)

Sorry I thought your comment was addressed directly to me.--BirgitteSB 14:08, 30 August 2007 (UTC)

I am glad to clear this up. I know you are one of the most active and important participants on this page, but it seems to me that there is a lot of contention here that involves many people other than you. Generally, in my recent posts on this page, when I say "you" I mean youse guys" i.e. second person plural (above I said "you all" but am now reverting to my native dialect) - I am assuming that whatever the impasse is - whether it is simply what you described, or what wbfergus described, or other issues - I assume that before the page can be unprotected a significant majority have to be in agreement, so I was addressing however many editors on this page could consitute that significant majority. SlimVirgin and KillerChihuahua agree with me that if you look at the sections above, it is very confusing. I though that my suggestion to sort out the confusion and streamline the discussion as a way of moving closer to a resolution would be uncontroversial. Slrubenstein | Talk 14:15, 30 August 2007 (UTC)

How about somebody archiving this entire page, then on the newer "clean page", state at the top that this "Discussion page" will be kept to only these 4 subjects (an arbitrary number, I'm only using it as an example). Then, create sub-pages for each of the additional subjects being bantered about, and have them hard-coded into a TOC at the top as well. So, when somebody comes to this page, it's easier to see and tell what this page is for and about, and for the other points of discussion or questions, there's a specific place associated for those. I could almost do all that, but it would take me a while to do. Also, I really don't feel "comfortable" enough doing it, being a relative newbie to this "article". wbfergus 15:11, 30 August 2007 (UTC)

If Birgitta and others who have been most active in the past ten days agree, then go for it. But enough people have to agree, in principle and concerning the threads or the approach you will take to sort things out, otherwise it won't work. Slrubenstein | Talk 15:29, 30 August 2007 (UTC)

Archiving and starting fresh would be fine. For my part I will commit to conciseness (I've already deleted several lines of what I was going to say here) and sticking to the listed topics on the new page. After resolution I'd assume the page could again develop/perform in the normal manner. --Minasbeede 15:40, 30 August 2007 (UTC)

I disagree with archiving active discussions. I don't want to cut off anyone's lines of thought or discussions prematurely. This talk page is no more convoluted than any Yahoo group, and all of us in the 21st century are used to this. Let's not cut short discussion. COGDEN 19:10, 31 August 2007 (UTC)

Comments on 9/20 draft (minor update) proposal

I "think" these minor changes were agreed upon. If not, feel free to delete or strikeout. wbfergus ^Talk 16:11, 20 September 2007 (UTC)

They were agreed on... good catch... on reading them in print, I don't think we need to "suppose" any thing ... we may not know what was the exact illness (polio or something else), but we do know he had an illness. BTW... I think very minor changes such as this can be made without creating a new draft. I only meant to say that if anyone wanted to change things significantly they should post a new version. Blueboar 16:22, 20 September 2007 (UTC)

This new formulation is more concise, and thus more likely to be read, that the versions that mentioned primary, secondary, and tertiary sources. Also, since the sources are not classified, there is no need to mention that facts may be found in both primary and secondary sources, which saves words.--Gerry Ashton 17:01, 20 September 2007 (UTC)

We can split the old PSTS into its own page and/or merge it woth other sections as needed and improve it there. As for concerns about where to use "close" sources and "distant" sources, does that belong here, on RS, or somewhere else? Jacob Haller 17:15, 20 September 2007 (UTC)

Wikipedia:Classification of sources is an attempt to address PSTS separately from this policy. Vassyana 20:48, 20 September 2007 (UTC)

Thanks. I couldn't recall the place. I suggest redirecting PSTS to CoS. Jacob Haller 20:55, 20 September 2007 (UTC)

Well, WP:SCLASS needs fixing up a bit first, I think. Last I checked, it still claimed that secondary sources couldn't contain 'new' analytical material without them being a primary source for that, which rather misses the point. SamBC(talk) 10:20, 21 September 2007 (UTC)

I prefer using "suppose" anyway (e.g. in case other details arise) but I don't feel strongly about it, so go ahead. I'm quite pleased with how this is progressing. --Coppertwig 17:05, 20 September 2007 (UTC)

The point of the example is that we have two fact statements that are obvious and uncontrovercial... if the FDR example does not fit, let's find another.

I have no problem spliting the old PSTS (and similar issues) to some other page... but I think we do need to keep the focus of this page on type of statement and avoid any discussion of type of source. It isn't the type of source used that makes something OR... it is the way you use it. Or to put it more succincltly: Sources are not OR... statements can be. Blueboar 17:28, 20 September 2007 (UTC)

I think the "born in Hyde park" example is better. That works well. I also like "Sticking to the Sources" as a section title. It sums up the issue pretty well. Do we want to have more than one example? There are a few different ways in which sources can be used for original research, so maybe we should have an example for each, such as where:

• an editor makes novel analytical conclusions based on two verifiable facts (like the FDR example above);
• an editor makes a novel factual inference based on two verifiable facts (e.g., source A says John did not eat meat, source B says John disliked leather, editor writes that John was a vegan).
• an editor elaborates on an author's published analytical conclusions in an original way (source says that Joe liked to drink and swear, editor writes that Joe wasn't religious);

COGDEN 21:02, 20 September 2007 (UTC)

As I've been noting, the primary / secondary material clarification of how sources are used covers issues that this proposal doesn't mention. My opinion is that it should be possible to subordinate this classification to the sort of fact / interpretation split that this proposal emphasises, but the primary and secondary idea has a simplicity and clarity that has to be fully matched for this proposed change to be acceptable. .. dave souza, talk 21:14, 20 September 2007 (UTC)

The problem was that primary and secondary did not have simplicity and clarity. Quite the contrary, there was a clear amount of complexity and confusion. A look through the archives at the discussion that occurred over the past month will make the confusion and (more importantly) the lack of consensus clear. Vassyana 21:36, 20 September 2007 (UTC)

It has a misleading simplicity that allows people to stop actually thinking about whether there's OR, and just say "ah, primary source, OR!!". SamBC(talk) 21:30, 20 September 2007 (UTC)

I've been slow to join in here because having looked over the archives there seems to be a lot of incoherence and no clear statement of just what the problems are with the primary / secondary descriptions. Could you please point to a summary of this, providing diffs, or add a concise statement here. After all, I don't want to do OR in interpreting all those archives ;) ... dave souza, talk 09:55, 21 September 2007 (UTC)

I sympathise. I'll attempt a summary below. Spenny 10:02, 21 September 2007 (UTC)

Since the example was changed, the new sentence didn't make sense, so I appended "its effect" into the sentence. wbfergus ^Talk 12:08, 21 September 2007 (UTC)