Archive 1Archive 2Archive 3Archive 5

Proposed new section on Primary and secondary sources

Here is a proposed new section on primary and secondary sources. I think this is a better explanation of the difference between primary and secondary sources, and sets a clear rule for editors to follow, which we need, since this is a policy, not a guideline. If we want to make further suggestions or guidelines regarding the primacy of primary or secondary sources, we can do that in a separate guideline, or at Wikipedia:Classification of sources.

Primary and secondary sources

Research that consists of collecting and organizing material from existing sources within the provisions of this and other content policies is encouraged: this is "source-based research," and it is fundamental to writing an encyclopedia. However, care should be taken not to "go beyond" the sources or use them in novel ways. Sources may be divided into two categories:

  • A primary source is a manuscript, record, or document providing original research or documentation. A primary source is where original data, information, theories, or conclusions first appear, and all original research begins its life in a primary source.
  • A secondary source is a second-hand report or review of a primary source. It includes reviews or interpretations of original research written by someone other than the original researcher. A source may be both a primary source and a secondary source; for example, if a scientist reviews the experimental results of another scientist, her review is a secondary source. But if the scientist makes further novel conclusions based on the prior scientist's data, her writing is also a primary source with respect to those novel conclusions. Secondary sources that review many primary and other secondary sources are sometimes called tertiary sources. This sub-category includes textbooks, treatises, dictionaries, and encyclopedias such as Wikipedia itself. Tertiary sources are not usually sources of original research, though they may report and review original research found in other publications.

Wikipedia can never be a primary source. Before being cited by a Wikipedia article, all original research, data, or information must first appear in a primary source, or be commented upon in a secondary source.

COGDEN 18:30, 23 August 2007 (UTC)

The use of "original research" here does not fit in with my understanding of it. Original research does not begin its life in a primary source. It can be born in my own imagination. I have done so before when composing examples on this talk page. This proposal is very far off-base in my opinion.--BirgitteSB 18:45, 23 August 2007 (UTC)
If it's in your own imagination, it isn't a "source". You can't cite "my mind": it has to be something written. COGDEN 22:33, 23 August 2007 (UTC)
Exactly "original research" is the opposite of "source-based research" it does not depends on sources. You cannot say "all original research begins its life in a primary source"--BirgitteSB 13:25, 24 August 2007 (UTC)

I disagree with this edit. It differs from WP:RS. It's important that Wikipedia articles be able to cite scientific research articles. In a scientific research article, the actual data (such as filled-out surveys) is a primary source; the published report is a secondary source and normally contains interpretation of the results by the scientific authors. Peer-reviewed articles are normally considered reliable sources. Disallowing them would result in major cutting of science pages -- a lot of useful information would have to be removed. I doubt this change could achieve broad consensus. --Coppertwig 19:14, 23 August 2007 (UTC)

I agree it's important to cite scientific research articles. This edit would not disallow that, and in fact encourages it. Just because a new theory in a scientific article is a primary source doesn't make it unciteable. Primary sources aren't bad, they're just original. COGDEN 22:33, 23 August 2007 (UTC)

Response to suggestion

Just as a quick note, it has been pointed out that, in certain unusual cases, wikipedia can be a primary source - about itself.

In terms of Wikipedia:Classification of sources, if there is a perceived need (as there seems to be sometimes) then that can be worked into a guideline, although it would need to change in focus somewhat. SamBC(talk) 18:40, 23 August 2007 (UTC)

Wikipedia can be a primary source, and not only about itself. For example, if someone makes a controversial statement on a Wikipedia talk page, and that statement is later quoted in newspaper articles, and then there's a Wikipedia page about the controversy, the original talk page statement would be a primary source about the statement -- which might have nothing to do with Wikipedia. --Coppertwig 19:03, 23 August 2007 (UTC)
The Wikipedia talk page can be a primary source, but a Wikipedia article should never be a primary source. That's the essence of the original research policy. COGDEN 22:35, 23 August 2007 (UTC)
An edit to a Wikipedia article can be a primary source, for the same reason as the talk page. Wait, even a Wikipedia article itself (or a particular version of it) can be a primary source. Suppose there are a lot of newspaper articles about an incident centring around the fact that a certain Wikipedia article contained a certain controversial (perhaps false or even outlandish) statement at a certain time. The original version of the Wikipedia article would be a primary source to establish a basic fact re the incident -- although the newspaper articles would be required to establish the significance of the whole incident. But, the Wikipedia article would not be being used as a source in the ordinary way, i.e. to gather facts about the subject matter of the article by taking it at face value. --Coppertwig 22:55, 23 August 2007 (UTC)
I agree that a Wikipedia article can be a primary source. I'm just saying that it shouldn't. If a Wikipedia article every becomes a primary source, something in the process has gone wrong, or somebody hasn't followed Wikipedia policy, and it usually becomes an embarrassment for Wikipedia. COGDEN 19:27, 24 August 2007 (UTC)
I'm saying that a Wikipedia article should be able to be used as a primary source in a case such as I described. Please explain why you think it should not. To clarify, here's an example: Suppose a particular version of a Wikipedia article says something controversial about a country, and a leader of another country publicly quotes it as "Wikipedia says that ...", the first country takes offense, and a war begins as a result. Suppose it's all documented in newspaper articles and other reliable sources. Later, a Wikipedia article about the causes of the war should be able to use the (appropriate version of the) original Wikipedia article as a primary source to establish a basic fact, i.e. the fact that the Wikipedia article did (or did not) say a certain thing, in the same way that it would use, say, the original newspaper article as a primary source if it was a remark in a newspaper instead that had started the whole thing. --Coppertwig 19:57, 24 August 2007 (UTC)


A different explanation in hopes of clarity

I don't know if this will be useful but it likely won't hurt. So let me share an interesting quote from The Free On-line Dictionary of Computing: "Knowledge differs from data or information in that new knowledge may be created from existing knowledge using logical inference. If information is data plus meaning then knowledge is information plus processing."

I would argue that primary sources, for our purposes here, are "unprocessed" sources. They can only offer us "data" and/or "information", while an encyclopedia must contain "knowledge". Primary sources are certainly acceptable sources for "data" or "information". However if an article does not rely on sources which actually offer "knowledge", it will fail to be encyclopedic. You can write beautiful source-based descriptions and spreadsheets with primary sources, but you cannot put them into context without either relying on a)your personal interpretations or b)secondary sources. A) would be original research while b) would be source-based research. Of course this is not the only way to produce original research but is the way the concerns the disputed section. Hopefully this explanation can clarify a few points.--BirgitteSB 21:42, 24 August 2007 (UTC)

That's a reasonable distinction between "knowledge" and "information", but not between primary and secondary sources. If a primary source is so narrowly defined as to only include "unprocessed information", that would include novels, journal articles, interviews, and poems, all of which are treated by non-Wikipedians as primary sources, even though they usually contain conclusions, interpretations, and other original "research". For example, a transcript of a reporter's interview with Abraham Lincoln might contain Lincoln's novel conclusions and insights into the Civil War. It's interpretive and analytical, yet it's also a primary source by any definition in common use outside of Wikipedia.
I think the kernel of the primary/secondary source idea has to do not with the rather metaphysical distinction between knowledge and information, but rather with the concept of precedence in time: A "source" is an idea, information, or research that can be cited, and it is "primary" if what you're citing is the original source of that information or conclusion. The information or conclusion being cited is "secondary" if it comes to the reader second-hand, not directly from the idea's original source. COGDEN 23:45, 24 August 2007 (UTC)
That brings to mind another example: Churchill's six volumes on the 2nd World War. He was there when many decisions were made, he made many decisions himself. With anything Churchill wrote and with anything anyone else wrote there's a need to be aware and sensitive in choosing what to use in writing about the war but to claim that secondary sources on war-cabinet-level decisions, actions, discussions, and conflicts are superior to Churchill's own writings seems absurd. Granted, Churchill may often or always have a personal agenda, but there is no proof that writers of secondary material do not also have agendas. Nor is having an agenda inimical to the truth.
In some ways this entire discussion is reminiscent of a comic sketch in which one person says something and another immediately claims that what the first person meant to say was something else. Here the primary source speaks (or writes) and a multitude of Wiki editors scurries around looking for someone, anyone, who removes all that nasty "primary" character from the statement (or writing) and turns it into good, solid, secondary material - necessarily altering it in character to get rid of the stain of primariness. (There's a possible religious example to be used here, too: a religious leader or prophet says something, hoards of later proponents of the religion based on that prophet's words expound a hugely different and distorted message. But they're secondary, so by the attitude so often exposed here they're correct and the original great figure was misguided - or something.) There's good and bad primary material, good and bad secondary material. Use the good, don't use the bad. Be willing to have someone else show why what seems good may not be, be willing to correct that someone if that someone is in error, be willing to work to resolve the issue if it's not all that clear (and in the case where it's not all that clear nothing other than useless simplicity is served by relying on some strict but unrealistic rule.) --Minasbeede 03:52, 25 August 2007 (UTC)
Total in agreement, to my mind. — Preceding unsigned comment added by Stefanomencarelli (talkcontribs) 19:10, 25 August 2007
Maybe another religious example (thanks for reminding me of it) is this. My dad majored in dead and ancient languages when he went to seminary, so that he could read all of the original biblical texts in their native language instead of through somebody elses interpretation of them. One example he always used was the 6th Commandment, "Thou shalt not kill". According to my dad, in the original language (Hebrew I think), it actually tranlates closer to "Thou shalt not commit murder". So, the translation itself was flawed, by using the most common verbiage of the time, while the actual word more accurately translates into a different meaning. While murder is killing, killing is different than murder in that killing does neccessarily involve premeditation or forthought, hence a completely different meaning and context. So in this case, the secondary source is almost always cited as being correct, when in fact it is incorrect and replied solely on original research, while the primary source (the original scrolls in their native language) by Wikipedia standards, can't (or rarely, should) be used. wbfergus 17:34, 25 August 2007 (UTC)
Right, that's i also try to explain. No dogma can solve this issues, nor burocracy can do it as well. But it happears to be a dialogue between deafs.— Preceding unsigned comment added by Stefanomencarelli (talkcontribs) 19:10, 25 August 2007
There is a mountain of reliable modern secondary sources available for the meaning of Biblical phrases and words, including such discussions in context of cultural period, etc. If current sources do not document the claim, it's quite likely a fringe or extreme minority claim, if not purely novel. Also, Bible translations are not a secondary source. A translation of a primary source is still a primary source. Vassyana 23:55, 25 August 2007 (UTC)
Correct, and the modern Hebrew language (which hasn't changed that much from the ancient Hebrew), translate the key symbols or word on the scroll as "to kill, murder or slay". The problem with translations is (especially with ancient translations) that along with literally translating the words or symbols, the resulting translation must also take into account common usage and context, making most translations "interpretations" or secondary sources. This is something that wasn't done hundreds of years ago during original transcriptions of the Bible, which is why King James allowed a new version to be re-translated, though some of their "guidelines" still specified various occasions to not translate correctly, like keeping "church" instead of correctly translating it as "congregation". Again, this is an incorrect (and acknowledged) translation, showing how the orinal source is incorrectly "interpreted", but the resulting incorrect translation can be the one perpetuated down through the years, as it is more common in daily use. There are also numerous examples of updates from the King James or older versions to more common modern language, so the inaccuaricies get perpetuated once again, possibly giving even more of a different meaning to various phrases or words from the original source. Sometimes during the translations of ancient documents, the original language used words that have no counterpart in modern languages, and various translators over the years struggle (and argue) with which word or group of words, would most accuratelt convey the same meaning, again lending the resulting translation to "interpretation and synthesis", both of which move something such as this from the "primary source" into the "secondary source" category. Also, if as you claim, even the (incorrect) translations of the Bible are considered "primary sources", then by the arguments of some on here, there should never be any quotations from the Bible on Wikipedia, as they come from a primary source. wbfergus 12:35, 26 August 2007 (UTC)
This is a gross misunderstanding of what the policy prohibits. The policy certainly doesn't say the primary sources should never be used. It states that relying on primary sources should be rare and done with great caution. (Whether or not it creates a de facto prohibition due to the general interpretation of the policy is another question, and a possibility I'm willing to accept.) Additionally, on a ridiculously well-covered topic like the Bible and accompanying religions, it's quite sufficient to limit the citation of the Bible to what reliable references say about it. Certainly, in the case of the Bible, it's incredibly rare that you would be able to cite a Bible verse independently of a secondary source without explicitly engaging in original research. Far and away, the vast majority of cases where people want to cite Bible verses, it is to support a particular interpretation of Christianity, Jesus, the Bible or the verses themselves. It's a wonderful example of why primary sources came to be discouraged, in relation to original research. Vassyana 23:12, 26 August 2007 (UTC)

Are we sure we want a special rule for scientific articles?

OK, so there's been a lot of discussion a few sections up. So far, anything published in an academic journal was a secondary source, just like newspapers and non-vanity press books were secondary sources. While that did create a situation where an article in a peer-reviewed journal that reported the original experiment would serve as both a primary and secondary source, I'm not sure we want to change the rules so that the article is no longer a secondary source. I'm concerned that it's going to lead to a lot of flimsy "OR" allegations and removal of good cites to good journal articles. Squidfryerchef 19:32, 23 August 2007 (UTC)

I agree -- see my message above. --Coppertwig 19:36, 23 August 2007 (UTC)
I responded above, but briefly - we're not talking about a special rule, but about codifying the way these articles are generally already handled. No one is talking about removing good journal articles, but calling anything peer-reviewed a "secondary source" creates a huge problem: it puts Wikipedians in the position of selecting which, of the thousands of peer-reviewed articles published every week, are the important or notable ones. That determination should be made by review articles, scientific bodies, experts in the field, etc (per WP:WEIGHT). Otherwise you get an editor who cherry-picks a 1970's study suggesting Depo-Provera causes endometrial cancer in animals, and chooses not to present the subsequent studies showing that it has a protective effect in humans (off the top of my head, but a real example). The solution is to use secondary sources (e.g. textbooks or reviews on Depo Provera), and cite primary sources (journal articles) within that context. Otherwise the door is open to all sorts of cherrypicking which violates WP:WEIGHT. MastCell Talk 20:55, 23 August 2007 (UTC)

The primary/secondary source distinction is inherent to history and similar fields but not useful or even applicable to all other areas. That is the fundamental problem here. Journal articles in the sciences are often both primary (reporting original research) and secondary (interpreting previous research) sources. In any case, we have always treated peer-reviewed sources as reliable; the issue of undue weight is orthogonal to the issue of reliability. — Carl (CBM · talk) 21:00, 23 August 2007 (UTC)

Using textbooks etc. is only one solution. Some research may not make it into that kind of source for a long time. Another solution is the wiki editing process: Wikipedians use their judgement as encyclopedia writers to decide what is interesting, relevant and imporant to include, and if one Wikipedian puts in something about cancer in animals, another Wikipedian comes along and insists that it be balanced by a statement about cancer in humans -- or that it be removed because the article is too long, because of arguments about undue weight, etc. Scientific articles are an excellent, valuable resource for Wikipedia. Limiting ourselves to textbooks etc. would exclude a large amount of useful information. As someone pointed out (but I can't find it now), some results aren't reported in review articles at least not for a long time: scientists consider the ordinary scientific articles sufficient in many cases. I don't see any particular reason why the opinion of a scientist writing a review article should carry much more weight than the opinion of a scientist writing an ordinary scientific article. --Coppertwig 21:14, 23 August 2007 (UTC)
(ec) I don't think reliability is at issue. I'm merely suggesting that journal articles reporting novel research should be used in a manner which hews very closely to the interpretations drawn by the authors. Reliable, peer-reviewed journal articles can be (and often are) cited out of context to support a conclusion not specifically reached by the authors or contrary to the meaning assigned those papers by experts in the field. That's a WP:WEIGHT issue, but it also turns on WP:NOR and primary vs. secondary sources. To Coppertwig: I'm not saying (nor does WP:NOR say) that journal articles can't be cited till they appear in a review. I am saying that they should not be cited out of context or in a manner unsupported by any secondary sources. MastCell Talk 21:20, 23 August 2007 (UTC)
Aha! "the interpretations drawn by the authors" -- that sounds like secondary sources to me!! All secondary sources, not only scientific articles, should be used in a manner which hews very closely to the interpretations drawn by the authors. Things are cited out of context; they should not be. We can repeat that things should not be cited out of context. We'll never stamp it all out.
I don't understand. First you say you're not saying that journal articles can't be cited till they appear in a review. But then you say that should not be cited in a manner unsupported by any secondary sources. If you don't count the ordinary scientific articles as secondary sources, and if they haven't appeared in a review article yet, then what secondary source can you use to support them? I think it's too much to require that they must be supported by a secondary source. That's equivalent to excluding a lot of scientific results -- anything that isn't reported in a review article or similar type of source. A lot of good information isn't. --Coppertwig 21:34, 23 August 2007 (UTC)

I've never claimed that journal articles can't be cited unless they've appeared in a review. I am proposing that journal articles should not be cited in a manner that directly contradicts the findings of more secondary sources. If I write that the annual incidence of pancreatic cancer is 5 per 100,000 and cite a primary source/journal article, who's going to disagree with that? But if I cite an article from the Journal of American Physicians and Surgeons claiming that vaccines are responsible for all human ailments as a rebuttal to the World Health Organization, or if I cite a 12-year-old paper by a tobacco-industry consultant to rebut the Surgeon General's finding that secondhand smoke causes lung cancer, then I'm engaging in original research or synthesis. That's all I'm getting at. And these are real examples. MastCell Talk 21:50, 23 August 2007 (UTC)

It seems that a lot of the objections to the classification of journal articles as (partly) primary sources seems to be that this would exclude their usage. However, there has also been discussion as to the (relatively recent) exclusion of primary sources in any case, as opposed to clarifying their appropriate use. Is it possible that it might be better to connect the two discussion more? SamBC(talk) 22:24, 23 August 2007 (UTC)
I think that makes sense. Usage is really the main issue anyway, and classification is secondary to that. I know what this policy says, but in practice primary sources are frequently cited and rarely objected to unless they're being used as part of an improper synthesis. The thing about scientific articles would not in any way mandate that they be removed from articles where they're properly cited - it's just an effort to clarify their appropriate use. MastCell Talk 22:31, 23 August 2007 (UTC)
Whether a journal article is or is not a primary source, we shouldn't assume that primary sources cannot be cited. They clearly can, because they are, and very frequently, and by our best articles. We need to make sure, however, that we are accurately representing the difference between primary and secondary sources. Based on my review of the literature, I think the predominant usage is to call journal articles primary sources, particularly when they contain original research (which they nearly always do, by definition). When you cite the first published instance of a new idea, you are citing a primary source. That holds whether you are in the field of science, history, religion, or any other academic field. COGDEN 22:42, 23 August 2007 (UTC)
What if you have some original research and also some interpretation of others' research? The entire distinction is problematic for many disciplines. It's most useful for history and degrades quickly when you move away from that field. — Carl (CBM · talk) 22:44, 23 August 2007 (UTC)
If the interpretation of the others' research is novel and original, I think it's a primary source as to that new research. It's also a secondary source, however, as to any review of the prior research other than the novel research. For example, if a scientist reviews prior scientific results, and then applies those prior results in a new way, the first part is secondary, while the first part is primary. If the Wikipedia article is discussing the new conclusions, the source would be treated as a primary source. If the Wikipedia article just uses the source as an overview of the prior research, it would be treated as a secondary source. COGDEN 19:24, 24 August 2007 (UTC)
Right. It's a matter of wording. Not citing journal articles in a way that contradicts a secondary source, OK (within reason: if it's an important, widely-cited and accepted scientific article that contradicts an earlier secondary source that can now reasonably be considered outdated, it would be OK to cite it -- though that should probably not be explicitly stated in policy or it would be widely abused; just carefully write the policy so it doesn't prevent it.) Not contradicting a secondary source is one thing. Not using it if it's unsupported by any secondary source is something else and would be too restrictive.
I agree with CBM. Outside Wikipedia, I think the terms "primary source" and "secondary source" are used in history or similar fields but not widely used elsewhere. --Coppertwig 22:47, 23 August 2007 (UTC)
No, I think the definitions are just a bit field-specific. There's a pretty clear distinction in the scientific literature between articles which report novel findings, and those which largely or only synthesize the work of others (e.g. review articles, textbook chapters, etc). Perhaps calling them "primary" and "secondary" sources sets up a misleading parallel with historical sources, but the distinction is very real and appropriate nonetheless. MastCell Talk 22:49, 23 August 2007 (UTC)
I agree there are "research" papers and "review" papers. In the field I edit here, math, we already cite textbooks and review papers for the bulk of things, and research papers almost only when we need to cite the original work. But the review papers often have original research in them (rephrasing theorems, proving unifying results, etc.). That's why the primary/secondary distinction isn't particularly relevant. In every case, the point is to cite the best reference(s) for the ideas being described. — Carl (CBM · talk) 22:55, 23 August 2007 (UTC)
Outside of history, the terms are not used widely, but are used sometimes. Basically, they are terms of historiography, and they are used to refer to the history of science, math, as well as other fields. Science historians, for example, refer to primary sources (e.g., original writings by Copernicus, Charles Darwin, or Einstein), and secondary sources (e.g., a 1875 newspaper article explaining the theory of Evolution in lay terms, or a Discover magazine article describing the theory of Relativity). But if a secondary article both reviews past scientific or mathematical research, and adds its own original research, it's both a primary and secondary source. COGDEN 22:58, 23 August 2007 (UTC)
I agree with Carl that the most important thing is to cite the best (or I would say "most appropriate") sources for what is being described. I think we need better guidance on this, less blunt and confusing than what NOR currently seems to say. SamBC(talk) 23:09, 23 August 2007 (UTC)
(ec) There's a couple of references in the discussion at WT:SCLASS the show that the terms are used, or at least understood, in natural and social sciences. SamBC(talk) 22:59, 23 August 2007 (UTC)

I'm not sure that I agree with this sentence, but it seems to have been in there for some time: "Although most articles should rely predominantly on secondary sources, there are rare occasions when they may rely on primary sources (for example, legal cases)." If this sentence is present, then classifying scientific articles as primary sources is pretty much equivalent to excluding their use. Besides, scientific articles don't fit my intuitive idea of what a primary source is. --Coppertwig 23:03, 23 August 2007 (UTC)

Intuition isn't often the best measure. However, this whole sort of thing is why some of us thought that it might be worth having interdisciplinary working definitions for wikipedia, which is how WP:SCLASS started. SamBC(talk) 23:05, 23 August 2007 (UTC)
Also, ISTR that that sentence was in part a reaction to other sentences in the page that suggested that primary sources shouldn't be used, full stop. SamBC(talk) 23:07, 23 August 2007 (UTC)
I agree. That's why this sentence must go. It does not reflect the current consensus of the Wikipedia community, nor does it reflect current or past Wikipedia practice. We can't even come to a consensus as to what primary source means, let alone that they should be "rare". COGDEN 23:08, 23 August 2007 (UTC)
I agree - I have no intention of altering policy in a way which excludes the use of scientific papers - quite the contrary, I'd rather they were used more often where appropriate. I've been the primary contributor to 2 featured articles (acute myeloid leukemia and cholangiocarcinoma), and you'll see that the bulk of the sources I used were "primary source" journal articles (though I would hope that the review articles etc. that I cited supported their use). Perhaps User:Sambc's suggestion to tackle the usage guidelines makes the most sense. MastCell Talk 23:09, 23 August 2007 (UTC)

I agree with COgden that Wikipedia articles are based on both facts and interpretations. Facts come from primary sources, and interpretations from secondary sources. I therefore support COgden's recent revert along these lines: use of primary sources should not be restricted to "rare".

One editor's intuition as to what something means is a reasonable predictor that there will likely be (at least some) others interpreting the words of the policy along similar lines. --Coppertwig 23:11, 23 August 2007 (UTC)

I concur with Coppertwig that a well written article contains both "facts and interpretations" and it is only OR if the facts or interpretations come from the WP editor, not WP:RS. We need primary sources for facts, and some subjects are necessarily fact heavy, because they either have not yet had extensive published analysis, or they are hard subjects that are not very subjective and not subject to interpretation in RS. So the blanket statement that all WP articles should rely on secondary sources is just wrong, and does not even approach a consensus opinion. It's probably OK to say that articles should rely on secondary sources where practicable, or some such weasel words, but if it has to be weasel worded, it should probably just be dropped. Dhaluza 15:46, 26 August 2007 (UTC)

Page protected

I've reprotected this page for 7 days, since it took a shade less than 24 hours after Husond's unprotection for people to go back to edit warring. It's unseemly to do this on a major policy page.

Sort out compromise wording here and I (or someone else not directly involved in the dispute) can unprotect earlier. TenOfAllTrades(talk) 23:33, 23 August 2007 (UTC)

Here's the issue: we need to decide what language, if any, should replace these two controversial statements:
  1. Wikipedia articles should rely on reliable, published secondary sources. (in bold)
  2. Although most articles should rely predominantly on secondary sources, there are rare occasions when they may rely on primary sources.
A policy is "the codification of current convention and common practice which already have wide consensus." It is supposed to "document the way that Wikipedia works." (See Wikipedia:Policies and guidelines). If we can't come to a consensus that these two statements document the way that Wikipedia currently works, we should just omit them. Personally, I don't think there is even consensus about what a primary or secondary source means, let alone that they should be "rare" or "predominently relied upon" COGDEN 23:59, 23 August 2007 (UTC)
Unfortunately some of us believe those statements do document the way Wikipedia works. It would be much easier to settle if the disagreement was actually over how articles are written rather than the particular wording that describes how articles are written.--BirgitteSB 13:30, 24 August 2007 (UTC)
I concur. I don't believe the statements need to be rewritten or replaced. Dreadstar 17:41, 24 August 2007 (UTC)
But on the other hand, "I" don't beleive that Wikipedia should be so restrictive as to prohibit truth either. A made up example would be along the lines of "(no name, but somebody with a bunch of money) was driving while drunk, sideswiping several cars before crashing into a mailbox, and rendering a child who was a passenger in one of the sideswiped cars comatose. Another person inside the car climbs behind the wheel while the "rich person" staggers out of the car and leaves. Several, say 40, eyewitnesses observe the entire chain of events and report it. The "rich persons" publicist immediately makes several press releases which are then printed and even included in an upcoming biography of the person, saying that the "rich person" wasn't present, and the car was only occupied by the one who subsequently traded places with the drunk "rich person". The only accounts of the eyewitnesses is in a police report which is sealed by the court by the "rich persons" lawyer, and on several blogs that the various eyewitnesses maintain. The media only makes a casual mention of these, saying "various fringe elements contend that the car was actually driven by "rich person", but no proof of these claims has ever been documented"." A crimal charge of reckless driving is made against the sober driver, and in a civil suit, the "rich person" offers to pay a small settlement to the parents of the comatose child in exchange for keeping the proceedings sealed. The parents are now restricted from publicy acknowledging the truth, the eyewitneses don't get any money either way, and just their simple little observation isn't worthy of an entire book, even though one them tried to get it published.
So, by Wikipedia "standards", the only thing that ever appears on here would be the perpetuated lie, as the eyewitness accounts would be a "primary source", correct? Just trying to get some clarification on how things are "supposed" to work on here, vs. how things "should" work. wbfergus 14:15, 24 August 2007 (UTC)
Sorry, you can only put the media accounts and such on Wikipedia. Unfortunate in some cases, but consider this: what if a group of 40 people who hate "rich person" decided to lie and say that "rich person" was in a particular place when "rich person" really wasn't? (Or what if a bunch of people saw someone who looked quite similar to "rich person" and started claiming (and believing) they'd seen "rich person" after they heard that other people were claiming the same thing?) What if the newspapers rightly ignored them? Maybe false claims like that happen often to famous people. If it's sufficiently important, couldn't you find at least one newspaper or thingy leaning in the political direction that would publish it? Isn't it better to err on the side of saying nothing than to wrongly defame someone -- both from an ethical and a pragmatic point of view, since contending against a single "rich person"'s lawyers could bring the whole Wikimedia foundation to bankruptcy? --Coppertwig 16:43, 24 August 2007 (UTC)
It is worth nothing that our living persons policy encourages the immediate removal of any unsourced material about living people, whether negative or positive. Vassyana 17:31, 24 August 2007 (UTC)
Okay, I'm now starting to see why there's been so many articles, blog postings, etc. about how Wikipedia has so much false information. I always just thought it because anybody could edit (and therefore vandalize), and more interested people were needed to keep tabs on articles to revert vandalism faster. Now it appears that if any wacko gets a book published, no matter what is in it, then parts of that book can used as a reliable source. If I could get a book published saying Kennedy was actually shot from the grssy knoll by three hidden snipers and the CIA has the recovered shell casing in different calibers locked away in a safe, then by Wikipedia "standards", this could suddenly be placed in the Wikipedia article, as long as it's appropriately referenced and perhaps written as "according to some sources..."? I know of a few places here in Colorado that will publish books for anybody on any subject for a minimum order of $5,000 (US). That's all it takes? wbfergus 18:07, 24 August 2007 (UTC)
Self-published (including vanity-published) sources are excluded, and have been for quite some time. SamBC(talk) 18:16, 24 August 2007 (UTC)
Right, it would very quickly run into WP:V self-published as well as Wikipedia:Notability issues. Dreadstar 18:23, 24 August 2007 (UTC)
Okay, if a writer uses a pen-name (so nobody knows who they really are) to get their work published, and then references that book on here (under a different name), how would anybody be able to put two plus two together to determine that the author and the editor were the same person? The wiki editor could claim that since their names were different, he wasn't the author and he was just citing it, as I've seen on many different military history articles on here (the claim that is, not the actual made-up example, though there is one questionable site that I've seen cited on several articles, www.centurychina.org or something like that, under the KoreanWarFAQ). Basically, I'm just playing devils advocate here, trying to when and/or if there are certain occasions where the rules do get broken, either intentially, unintenionally, or with concensus of that articles editors. wbfergus 18:36, 24 August 2007 (UTC)
Self-published doesn't mean published by the same person who is editing the Wikipedia article (as in WP:COI); it means the author and the publisher are the same person, i.e. someone has paid to have their work published. If an author sends their work to a publisher and the publisher decides to publish it, then there is an additional step of approval, so it's more reliable, but still not necessarily a good source. See WP:RS. Just because something is in a book doesn't necessarily mean it has to be in a Wikipedia article. Also, often the Wikipedia article will say something like "A book by X states that ..." rather than just asserting controversial statements. --Coppertwig 18:46, 24 August 2007 (UTC)
Responding to Birgitte's statement that: "Unfortunately some of us believe those statements do document the way Wikipedia works" :
I think the key word here is "some". I doubt you could even say "most", and that is clearly not a consensus. Besides nobody has responded to my repeated argument that nearly all Wikipedia:featured articles make liberal use of primary sources. If it's deemed acceptable in any substantial number of featured articles, it's clearly not a policy. COGDEN 19:18, 24 August 2007 (UTC)
Besides nobody has responded to my repeated argument that nearly all Wikipedia:featured articles make liberal use of primary sources. I don't see the argument in that, those articles still rely on secondary sources. If you instead had said that nearly all Wikipedia:featured articles do not rely secondary sources, then there would be an argument. From your proposed re-write above, I am not even certain you clearly understand what original research is at this point. I understand that you believe the status quo of this policy doesn't have consensus, but by that same argument you can hardly claim that your own edits have consensus. Consider that under this status quo policy, many quality articles that you are happy with have been written. Therefore do you not think it is possible that this status quo policy actually supports those articles and is more in line with your own ideas than you have believed up till now.--BirgitteSB 19:52, 24 August 2007 (UTC)
BirgitteSB, please clarify. I think maybe you mean that all Wikipedia articles should rely on secondary sources, and that they may also rely on primary sources too: i.e. Wikipedia articles should rely on either secondary sources, or a combination of primary and secondary sources. Is that what you mean? --Coppertwig 20:00, 24 August 2007 (UTC)
Not to speak for Birgitte, but what she previously said seems to agree with my perspective that Wikipedia articles should rely on reliable, published secondary sources, although there are rare occasions when they may rely on primary sources. Exactly what the policy stated prior to Cogden's non-consensus changes. Please correct me if I'm wrong, BrigitteSB! Dreadstar 20:37, 24 August 2007 (UTC)
(ec) I think the distinction is whether they use them, or rely on them. Would the article be plausible (albeit sparser) without them? SamBC(talk) 20:38, 24 August 2007 (UTC)
Right, good distinction. How often does an article completely rely on primary sources alone? I think we would run into problems if there were limited or no restrictions on use of primary sources. I tend to think that 'rare' is is indeed the watchword on the subject of articles relying on primary sources. There can be a mixture, naturally, but to totally rely should be somewhat of a rarity, I would think.... Dreadstar 20:50, 24 August 2007 (UTC)
Endorsing the use vs. rely distinction. It should not be rare for an article to use a primary source, however the more encyclopedic portions of the article will inevitably rely on secondary sources. Direct quotes give nice illustration and insight into article but what is encyclopedic is information put context. What is parts of an event were most important, what incident was a turning point, whose decision made a difference, what lasting influences of that event are still felt today. These things answers can only be found in secondary sources which have the "distance" to look at the event within a larger context.--BirgitteSB 21:07, 24 August 2007 (UTC)
Exactly. Well put. Dreadstar 03:29, 25 August 2007 (UTC)
Wrong, as I have already commented above, in aviation, specifically for aircraft and airports, we rely on primary sources. Also many, many geographic place articles rely on primary sources. So while the majority of articles may be primarily based on secondary sources, there are significant sub-groups of articles that do not. The myopic focus on primary vs. secondary sourcing, and the wholesale depreciation of primary sources is unnecessarily distracting. Why don't we focus on how and why primary and secondary sources are problematic, rather than trying to make one-size-doesn't-necessarily-fit-all prescriptive guidelines?
The current version of this sentence is fine: "Wikipedia articles should rely on reliable, published primary and secondary sources.". An earlier version which said "Wikipedia articles should rely on reliable, published secondary sources." is at best confusing.
I think it only adds confusion to make a use/rely distinction. Two different Wikipedians can write the same article in the same style using the same proportion of primary and seconary sources in the same way, while one of them sees the facts from the primary sources as being dry and less important and the interpretations from the secondary sources as the essential glue that holds the article together, and the other Wikipedian sees the facts from the primary sources as being the important information in the article and the interpretations from the secondary sources as being the mere glue that holds the article together. Since both Wikipedians are carrying out the same edits to create the same article, they must both be following policy (or both violating it), yet to one, the article clearly "relies" on primary sources to establish facts. Actually, even if one considers the facts unimportant, can't one still say that the article "relies" on a primary source to establish certain (unimportant) facts? Maybe the people writing "rely" in the policy really mean that it's the article's existence or the article's central threads that must rely on secondary sources, not that the article must rely on secondary sources for everything, even the establishment of (little) facts.
The definition of "encyclopedia" in my dictionary says nothing about secondary or tertiary sources. I think many people consulting Wikipedia are looking for specific facts such as "What is the population of Australia?" To them, the important thing about an article is whether it contains this fact and how easy it is to find. We can specify how articles are to be organized, but we can't legislate mind control over either the readers or Wikipedians. We can allow or disallow the inclusion of facts from primary source (I strongly vote allow), but we cannot force readers or editors to think of these as unimportant. --Coppertwig 13:43, 26 August 2007 (UTC)
May I suggest that we further clarify the sentence: "Wikipedia articles should rely on reliable, published primary and secondary sources." to say "Wikipedia articles should rely on reliable, published primary sources for facts, and secondary sources for any interpretation of those facts." I think this slight expansion should cover the major points raised here, and make the distinctions much clearer. Dhaluza 15:53, 26 August 2007 (UTC)

(unindent) The only problem I see with that, personally, is that there are circumstances where primary sources are nonexistant or not available, at which point getting facts from secondary sources is fine. Primary better for facts, secondary acceptable. SamBC(talk) 16:08, 26 August 2007 (UTC)

Yes, but in that case, you are still using the source as a primary source to support the fact. A fact remains a fact throughout the chain of custody, so even if it has been repeated 10 times, the source is still being used as a primary source. Naturally, the fact becomes less reliable the more times it is repeated, so we prefer the original source to a non-original source. But if we are citing a fact as a fact, we are using a primary source. Perhaps we need to make a clear distinction between original and primary sources to clear some of the confusion? Dhaluza 16:16, 26 August 2007 (UTC)
Responding to Birgitte's comment above, "What is parts of an event were most important, what incident was a turning point, whose decision made a difference, what lasting influences of that event are still felt today. These things answers can only be found in secondary sources which have the "distance" to look at the event within a larger context". While I mostly agree with this, there is still that old adage that goes something like "the victorious get to write history". Especially with ancient history, most of the sources of information are from those who were victorious, so practically every interpretation through time has been made solely on the accounts of the victor. In many cases, everything from the vanquished was destroyed, leaving only a one-sided perspective, however right or wrong. It also makes it almost impossible to present the side of the vanquished without making conjectures from what limited information may be available for that side. As well, in many cases, such conjecture may not be worthy of an entire book or article, but maybe it could be included in a section called something like "Editors conjectures" or something. Some way of not preventing the thoughts of some of the well educated editors on here, and allowing some ideas to be presented, and then left to the reader to decide whether there is any merit in them or not. A possible case may be with a recent arguement we had over on the Korean War page. Most western sources place the Chinese casualties at approximately 400,000 killed (either directly or indirectly) with approximately 486,000 wounded. The Chinese however claim to only have lost 114,000 dead and 380,000 wounded (exactly). Just the dry facts themselves lead to many edit wars, while a policy that would allow something like "A claims X, however B claims Y. The actual Z is probably somewhere in between". I've never seen this print anywhere, yet it is logical to assume that, yet Wiki policy currently prevents it. wbfergus 15:16, 26 August 2007 (UTC)
No, just state "A claims X, however B claims Y" and leave it to the reader to form their own opinion. Your "The actual Z is probably somewhere in between" is just speculation and should not be included, because that is OR. In your example, the western sources may also have underestimated the actual casualties for a number of reasons, and putting speculation in the article, informed or not, is nonsense. Now if a respected military historian came to that conclusion after studying the evidence from both sides, then you could attribute that opinion to the expert, and it would be relevant in that context. If dry facts lead to edit wars, it's not the facts' fault, and your solution just creates a bigger problem, because it sends us down the slippery slope. Dhaluza 15:34, 26 August 2007 (UTC)

I unwittingly etided while the page was protected. Fortunately, someone brought my error to my intention and I reverted. Just reviewing this section of the debate, however, I would like to add two comments. first, above, one editor wrote, "But on the other hand, "I" don't beleive that Wikipedia should be so restrictive as to prohibit truth either." This comment is just inappropriate for Wikipedia and it doesn't even have to do with this policy - our NPOV and V policies make it clear that Wikipedia's standard is verifiability and not truth. this has been a guiding principle of Wikipedia practically since its inception. Even if someone could come up with a compelling argument that such a thing as "objective truth" exists, practically people always disagree on it. Our policies are meant to provide a framework for ensureing that people with divergent, even opposing views can collaborate on writing a great encyclopedia. This would be impossible if editors kept arguing over who is "right" (i.e. knows the truth). But regardless of our views on the truth, we can all agree that "x (a given verifiable source) claims y." All of this is explained in the NPOV and V policies. Any discussion here will be more productive and any argument more easily resolved if we just forget about truth. fell free to argue about what you think the truth is on your blog, usenet, a chatroom, or your favorite bar or pub. Second, another editor remarked that many (perhaps most) articles rely on primary sources. Thee are two reasons for this. First, in many cases it is absolutely fine (in the limited sense that "it has never violated any Wikipedia policy, certainly not this one"). It is fine to use primary sources as long as they are verifiable, and as long as they are not used to forward an editor's own view via an editor's own analytic, synthetic, explanatory, or interpretive claim. Second: in some article primary sources are used, and indeed used inappropriately. There is a good reason for this: Wikipedia is a wikipedia - anyone can edit, at any time ... which means anyone can break the rules at any time by inappropriately using primary sources, and thanks to the wiki software, some other editor can fix the mess. It also means that our hope is through constant involvement by more and more people, Wikipedia will continuously improve - and, logically, if it gets better over time, then it was worse in the past. No one denies this. We all know that four years ago Wikipedia didn't cover nearly as many topics as it did now. But Wikipedia was worse in the past in other ways - some articles really were not NPOV, and some articles were not verifiable. over time we have developed a clearer sense of the kind of franework we need to make Wikipedia work, and this policy is one of them. Of course it is possible that an article that was written in 2001 (before this policy took its present form) and hasn't been edited since actually complies with this policy, all other policies, and is a great article. But it is more likely that it has problems, and is not fully compliant with policy. The fact that many articles may not comply with policy does not mean that the policies are wrong ... it means this is a wikipedia which means it is a perpetual work in progress meaning EVERYTHING can ALWAYS be improved in SOME way. I hope that we can resolve this dispute quickly (and unprotect the page so i can put in my edits again ... and so some other editor can delete or rewrite or add to what I wrote to make it beter ... and so some other editor can make it even more better ...) so people can do what we really should be doing: making sure existing articles comply with NPOV, NOR, and V by adding proper citations or more content from verifiable sources representing other points of view, and writing new articles, guided by these policies. But whatever you write, remember - it can always be made better, and hopefully a few years from now someone will read it, and see a way to improve it. Slrubenstein | Talk 18:01, 28 August 2007 (UTC)

"This comment is just inappropriate for Wikipedia and it doesn't even have to do with this policy - our NPOV and V policies make it clear that Wikipedia's standard is verifiability and not truth." Fine, let's do it: get rid of "truth." But when we chuck "truth" out the window let's be honest enough to chuck "encyclopedia" with it. What I perceive is a mighty struggle to rationalize an ill-conceived "policy." Truth is out, logic is out, primary sources are being forced out (by a vocal minority.) Do it, but change the name to something that doesn't claim nor imply "encyclopedia." It's the Wiki-secondary-sources-only-truth-doesn't-matter-logic-doesn't-matter-pedia. And I'd like to issue a request that all those who denigrate primary sources show secondary source references to back the claims. Criminy, show primary sources, show any sources. For people who claim to be all hot and bothered about verifiability you sure do duck that responsibility with the backbone of your argument. --Minasbeede 19:23, 28 August 2007 (UTC)
Well, ideally, Wikipedia should provide the truth, but pragmatically, I think it should provide (1) the general consensus, and notable dissenting views, on the topic (2) a quick intro for people unfamiliar with, and marginally interested in the subject, and hopefully eventually (3) help identifying and finding the important works, primary and secondary, on the subject. That third goal is a long way off, but disfavoring primary sources puts it ever-further off.
For each goal, the article should review the literature - it should not present new hypotheses. Jacob Haller 19:56, 28 August 2007 (UTC)
OK, the scary thing about what you say is that it seems to indicate that flat-earthers and the like should get equal coverage , equal weight, equal treatment. As long as what's in Wikipedia can be "verified" (attributed to some source) it belongs. That opens wide the door to what NOR is supposedly meant to prevent. (Yes, I know qualifying adjectives are attached to "source" in many of the statements made but all the adjectives serve to do is to mask the issue and make possible a parroting of the policy that avoids the important issues involved.) --Minasbeede 23:03, 28 August 2007 (UTC)

Minasbeede, I will ingore your sarcasm and hypebole. These policies have massive support in the community and really do get at the heart of the project. So you don't like them? Well, guss what: you do NOT have to comply with our policies at all! You do nothave to edit Wikipedia articles! you don't to be here ate all! You reject our values and policies, well, you are a free person, just go away. Based on your idea of a propoer encyclopedia, I have good advice for you: put together your resum and apply for a job with Encyclopedia Brittancia. Go work for them, I am sure you would be happier! Slrubenstein | Talk —Preceding unsigned comment added by Slrubenstein (talkcontribs) 20:10, August 28, 2007 (UTC)

It would be inappropriate to respond to your ad hominem rant in kind. (You should remove it, Slrubebenstein: it's unseemly.) Your claim of "massive support" is exactly the kind of useless material that the Wikipedia policies are intended to prevent appearing in the work itself. You people who are so adamant in your twisted and rigid interpretations of policies flagrantly violate the policies and their spirit while you claim to be upholding and supporting them. Find me a reference that shows that any sizable body of people anywhere (outside the old Politburo of the USSR) support the notion that an encyclopedia should not be concerned with truth but only with "verifiability" (and of course with the Politburo "verifiability" meant "complies with the current party line.") --Minasbeede 22:52, 28 August 2007 (UTC)