Wikipedia talk:Use of primary sources in Wikipedia

Initial discussion

Proposal started as an idea developing from what I wrote in VPP --Francis Schonken 12:38, 24 October 2006 (UTC)

Looks good. I'd like to see a bit more about (or links to) stuff about how photographs can be used as primary sources, and the difference between using them to illustrate something mentioned in the article and using them to support a statement made in an article. Also, the difference between linking to a website with the image and uploading a fair-use copy. Sometimes a photograph is just a pretty picture. Sometimes it actually demonstrates a disputed point that isn't documented elsewhere. That sort of thing. Also, is there a similar document about secondary sources? Carcharoth 13:26, 24 October 2006 (UTC)

Re. using photographs to support a statement made in an article: I find it hard to come up with an example for that. Well, finally, maybe this one: Image:Jolanda Ceplak fotofinis.jpg used to support the statement "The photo-finish did not go Jolanda's way and Benhassi was given the silver medal and Jolanda bronze." in the Jolanda Čeplak article. Is that what you meant? Do you have a better example? Anyhow, for the example I gave, the interpretation of the image depends largely on information given outside the JPG image itself. For starters, one would have to know who the four running women in the photograph are. And where and when exactly the photograph was taken, etc. I'd still rather tend to see, in Wikipedia context (we're not the referees interpreting that photograph), the photo as an "illustration" to the contention about Ceplak's bronze medal. The secondary sources for that information are news coverage of the Olympics, etc. Without such accessory sources I don't see how the "bronze" vs. "silver" information could be "derived" from such photograph directly. But as said, if you think about a more clearcut example, feel free to tell about it. For "images used as illustration" there's less problem to find examples I suppose! --Francis Schonken 14:23, 24 October 2006 (UTC)

Re. "is there a similar document about secondary sources?" – yes and no. Not "separately", no. But in essence, core content policies like WP:V and WP:NOR center around secondary sources most of the time, don't you think? --Francis Schonken 14:23, 24 October 2006 (UTC)

Wikipedia:Instruction creep

Is there anything here that can't be said with a line or two at WP:RS? And is the current text even proper? Primary sources can certainly be used to illustrate a point, yet they can also be used to verify dry facts. I see no reason why the biography of Thomas Nelson, Jr. shouldn't be able to reference the Declaration of Independence as proof that he was one of the people who signed it. Durova 15:21, 24 October 2006 (UTC)

• Yes, but do you need to? Cripipper 19:00, 25 October 2006 (UTC)

Of course, without clear instructions, everyone ends up debating things in circles, and getting confused. That is the problem with not having clear guidelines. If you reduce it to a few sentences that you understand, I can guarantee that someone else will come along and misunderstand it. In some cases more really is more, and less really is less. Carcharoth 15:43, 24 October 2006 (UTC)

Maybe someone is confused here, but it isn't me. The appropriate uses of primary sources have been well established outside of Wikipedia. Some editors join the project who aren't aware of those uses, and a few editors deliberately ignore or distort them. Deliberate and persistent misuse is already covered under WP:DE. WP:V and WP:RS address normal use. So either this proposal is redundant, in which it should be abandoned, or else it attempts to redefine the acceptable use of primary sources at Wikipedia according to some idiosyncratic scheme. In the latter case this proposal should be torched and buried. Any potential benefit is dubious: Wikilawyers and trolls would get a new nexus of guideline to misinterpret while productive editors would get overburdened by the requirement to confirm dry facts through secondary sources and many of Wikipedia's best articles would suddenly contravene a site guideline. Durova 16:39, 24 October 2006 (UTC)

Examples, examples, examples

I asked for examples above, and while I can't come up with them myself, I really think they would help. It makes things so much clearer. Carcharoth 15:41, 24 October 2006 (UTC)

Status, for example, of government reports?

I am very much opposed to this proposal; let me give an example why.

Recently I have been engaged in editing the article on the Space Shuttle Challenger disaster. A significant proportion of this article is devoted, naturally, to the question of why the accident happened, and where blame should be assigned. The major and authoritative source on this question is the report of the Rogers Commission, which was convened to investigate the accident. The testimony that was given to the commission is a primary source; however, in historical terms, the report itself is also a primary source.

The proposed rule would mean that in this situation, the article on the Challenger disaster would only be able to refer to the Rogers Commission report for "illustration," yet would be able to consider websites which derive their information almost entirely from Rogers as reliable secondary sources. This would be a ridiculous situation, and for this reason, among others, I have to oppose the proposal. MLilburne 17:10, 24 October 2006 (UTC)

A couple more examples from my own editting that show this proposal, although well-meant, causes more harm than it prevents:

• The Futuh al-Habasa is a primary source for the history of Ethiopia in the second quarter of the 16th century (as well as for the life of Imam Ahmad Gragn), & was translated into English a few years ago. However, there is practically no secondary literature available to explicate this document, although much of the narrative is unambiguous (e.g., Imam Ahmad wa victorious in this battle, he burned down that church, etc.) General histories for this period omit much of the detail provided in this work.
• The Kebra Nagast is a primary work that is easily available as an ebook on the Intenet; yet the only secondary work that I have so far identified that discusses it is an unpublished graduate thesis -- which in theory is available thru Inter Library Loan, but I am still waiting on my request. This is clearly an ironic situation: quotations from the primary source can be verified, but those from the secondary are not easily verified. And by the terms of this proposed guideline, no one could then offer even a summary of this work -- which is an important document of Ethiopian culture. -- llywrch 19:56, 24 October 2006 (UTC)

Good points. I suggest case-by-case treatment of which sources to use. One of the main problems is that lots of these cases require editorial judgement on the part of Wikipedia editors. That could be the weak point in the chain. Do I trust this random Wikipedia editor to have used a reliable source? In effect, these guidelines could be seen as a way to guide editors who don't fully understand why primary sources sometimes are best. ie. When all the secondary sources are derived from a single primary source bottleneck. Carcharoth 21:46, 24 October 2006 (UTC)

While annoying for Llywrch, if this document is so important I would imagine that there is secondary literature available on it. My apologies if I am wrong, but it appears to me that part of the problematic issues raised stem, in part, from an over-dependence on internet-based sources. Cripipper 18:55, 25 October 2006 (UTC)

Unfortunately, you are misunderstanding my argument. My point of using the Kebra Nagast as an example was to show the extreme inverse of a case where secondary sources are difficult to access but the primary source is very easy to access. I prefer to quote from secondary sources to furnish an interpretations & explications rather than to offer my own -- but in some cases a lack of material forces me to explain or discuss primary sources. (Right now, I'm wrestling with discrepencies over the organization of Ethiopian woredas, the local administrative unit; I'm going to need to rewrite some of the articles I've contributed to show that the source is unreliable in more ways than I have so far.) Some Wikipedians I have no problem with them doing this: they are honest, level-headed & put the interests of Wikipedia over any partisan beleifs. At the same time, we must all admit (as Carcharoth notes above) that no matter how many guidelines or rules we adopt, there will be some who will find a way to twist them in order to push into Wikipedia their own peculiar beliefs or original research. -- llywrch 19:57, 25 October 2006 (UTC)

The proposal also raises havoc with all the aircraft accident articles we have on Wikipedia. Quite often the ONLY reliable evidence we have giving the cause of a notable accident is the investigating authority's accident report. This is even true for accidents where over 200 people died, and especially for accidents that aren't - well, I don't want to say "popular", but perhaps "well-known to the average United States citizen in the year 2006". The accident report is a primary source, but without it not only are accident articles unwritable, they're also unverifiable. --Charlene 04:18, 31 October 2006 (UTC)

Ugh

A primary source is acceptable if its use is

1. Verifiable (WP:V)
2. not a novel or interpretative conclusion or theory (WP:NOR)

Period. Any proposed guideline has to start with the policy basis from which it derives. Nearly everything written after those two sentences needs to flesh that out, but that is the core. The core use of a primary source is not illustration. SchmuckyTheCat 22:11, 24 October 2006 (UTC)

I agree, but I'd word it differently: the primary source should

1. be reliable (WP:RS) (The source is the verification.)
2. not be used in a way inconsistent with WP:NOR#Primary.2C_secondary.2C_and_tertiary_sources - don't use the primary source as a starting point for your own analysis/interpretation/whatever.

I guess you could also say relevant, although that goes without saying, but undue weight might be an issue.

— Armedblowfish (talk|mail) 16:44, 29 October 2006 (UTC)

Too complex nutshell

The current statement of the policy "in a nutshell" is too complex. It should simply state what the preferred course of action is, but instead, it gets bogged down in explanations. The second sentence, in particular, is entirely inscrutable, especially as it uses a highly emphasized word (illustration) in a different sense than it was used in the first sentence!

Here's an example of what the nutshell could say:

Primary sources should be given less credence than research that has been based on those sources.

Not that I'm certain of what this policy is about, just yet :-) --Yath 02:14, 25 October 2006 (UTC)

Yet this doesn't deal with the 'Look at these new documents!' trump card, on which no historian has had the opportunity to write (which was the real point I was getting at when I raised this originally). Cripipper 18:58, 25 October 2006 (UTC)

If it can't be verified elsewhere...

What I wrote at the pump

:*As mentioned below, these are primarily illustrations of the article, not sources. The letters, for example, can be used to illustrate something referred to in a secondary source. Where the problem exists, is in the numerous articles on 20th century history which have within then the phrase "As recent documents released/published/put on the web by X,Y,Z show..." A case can be made that these do not break guidelines, as they do indeed show what they purport to show. The problem exists with context and the abilty of the person using them. Sadly there are many editors out there who wouldn't know the difference between a literal and interpretive source. A tightening of the guideline to permit primary sources that are used to illustrate points verifiable from other sources, but ban them without other supporting evidence would accomodate the concerns outlined above.

It is very easy for Durova to say it is all so simple, but he seems to primarily work with earlier periods of history where the available evidence is fairly clearly established. However, for those of us working on 20th century history, this is a real problem (and pain). Would it not be fair to say that if you are drawing on primary evidence for a point that cannot be found elsewhere in secondary literature, then that is original research. The simplest way to get around this is to require the use of primary sources to be backed up by a relevant secondary source. It is my own particular bug-bear that part of the root of this problem is an over-dependence on internet-based research and sources, which then becomes self-feeding... Lack of books on the subject results in use of internet sources, but a recently-available source on the net is of higher quality than the other secondary literature so it seems obvious to use it... Cripipper 18:55, 25 October 2006 (UTC)

"The problem exists with context and the ability of the person using them."

Surely this problem exists whether the sources being used are primary or secondary? Most of my work on Wikipedia is on 20th century history, and it would seriously curtail many interesting articles if Wikipedia was not allowed to use primary sources to establish facts or to present the interpretations made by the people who participated in the events that are being discussed. A little bit of thought and discernment will make it clear when primary sources are being used in inappropriate ways. MLilburne 19:25, 25 October 2006 (UTC)

Which is what I try to communicate when I say that we should report our sources "accurately & honestly", for they are two different & independent qualities. One can quote a source very accurately -- yet twist its meaning in a novel or unusual way to arrive at a conclusion that is dishonest to the source. -- llywrch 20:11, 25 October 2006 (UTC)

Have you ever tried telling someone that they are using primary sources in an inappropriate way? Cripipper 11:08, 26 October 2006 (UTC)

I'm a historian; so yes, I have. MLilburne 12:30, 26 October 2006 (UTC)

I mean on here. It invariably turns into an edit war as one is repeatedly redirected in the direction of guidelines and policy that show primary sources as welcome. It is my own personal belief that the raft of articles and sections that have within them the phrase "As recently declassified documents show..." or some such variant are not only against the spirit of the guidelines, but also make for badly written articles that lose proportion and veer off in a POV direction. How can we try and cut down on this? I have been trying, but the guidelines as they currently stand make it very difficult. Cripipper 13:51, 26 October 2006 (UTC)

Both primary and secondary sources can be misinterpreted, depending on the competence and/or honesty of the editor. Such issues must be resolved through discussions, dispute resolution etc.. Not through prohibition of any independent use of primary sources - or of secondary sources for that matter. I do agree that primary sources should only be used when their interpretation is relatively clear and indisputable (otherwise, that would contradict the spirit of NOR and V). But apart from that, I agree with the majority of the editors above: this guideline would be harmful, and especially so in an encyclopedia where material supported with references to either primary or secondary sources is still far from universal. --Anonymous44 15:39, 26 October 2006 (UTC)

Ancient history

The other problem is ancient history, where the primary sources are often what there is, and encouraging editors to go beyuond them results in crackpottery. Consider Ephialtes of Trachis, which is, and must be, largely Herodotus. Some editor has gone to some, unnamed, secondary source, and stuck in the bit about Ephialtes having a grudge against the Spartans; this is a perfectly reasonable speculation, but it is speculation.

The other problem is exemplified by Pericles, which treats Plutarch as a secondary source, on the grounds that he quotes texts which have not otherwise come down to us. Plutarch is not evaluating them as a modern source would, and should not receive the same credit. Septentrionalis 16:48, 31 October 2006 (UTC)

Delete all talk about primary, secondary, tertiary,... sources

There is absolutely no need for such classification: we can easily do without, and all that talk about primary, secondary and tertiary sources only confuses the editors.

Instead, it suffices to point out what is essential: that claims do not go beyond the sources, as explained in WP:NOR. From that all essential stuff follows. At least, that's what I claim; you are free to challenge it.

For starters, the example that now reads:

'The transcript (while it is a primary source) can not be used as a basis to present the slant (opinion, perspective, position, observations...) of the witness as "fact". That is, unless a reliable secondary or tertiary source (for instance a judge's verdict, or a generally accepted history book written after conclusion of the court case) does so.'

Will next read:

'The transcript (while it only presents the claims of the witness) can not be used as a basis to present the slant (opinion, perspective, position, observations...) of the witness as "fact". That is, unless a reliable and generally accepted source (for instance an undisputed judge's verdict, or a generally accepted history book written after conclusion of the court case) does so.'

In this case I had to make a few corrections; and to my own surprise, the need for correction only became transparent by the deletion of classification.

Thus I hereby change my opinion: I conclude that such classification in groups of sources is detrimental for Wikipedia.

Harald88 19:11, 3 November 2006 (UTC)

Recent changes to proposal

With the recent changes, this proposal has just gotten even more complicated, and now seems to spend most of its time defining what is and what isn't a primary source for its own purposes. If the rules are this involved, it's just not going to work. MLilburne 11:27, 7 November 2006 (UTC)

I removed some text that was describing the reliability of a primary source, as it was in contradiction with other text in the proposal. ≈ jossi ≈ t • @ 23:53, 9 November 2006 (UTC)

Combine

It might be useful to combine this page with WP:ATT. While the intent of this proposal is good, we start to have too many pages on the subject of sourcing. (Radiant) 16:34, 16 November 2006 (UTC)

Well, it originated as a WP:ATT split-off. Couldn't say in which archive of the WP:ATT talk page people were discussing whether Wikipedia should use the primary/secondary/tertiary source distinction. Just before the first bodies would have needed to be carried out (metaphorically speaking), I thought maybe better to split this discussion off, work on it, and decide afterwards whether it can be useful for WP:ATT and/or a reworked WP:RS - I left notice on both talk pages.

It's not finished yet, and already too long, so anyway I'd rather first finish it, then trim it (needs trimming!) and then decide what to do.

While working on it I rather felt this was moving towards a MoS page (something of a generalised version of WP:WAF, including non-fiction primary sources). So, possibly it would be useful for Wikipedia without necessarily driving it back to a (so-called) "core content policy".

Anyway, one of the problems for using this in an unbloated core content policy page, is that I can't find a way to give a short, snappy definition of "primary source", that always works (I mean, as a useful distinction - there are hundreds of snappy definitions of primary sources that are not the least bit useful for building a guideline upon). As it is, it could only work as a more detailed guideline, not as an easy to formulate over-all principle. Also, not even anything I'd expect everybody to understand (not needed for achieving a basic quality for Wikipedia), rather something to refine quality once basic quality has been achieved. I mean, rather "FA" (replacing the current 1-sentence formulation about images in FA guidance), than "GA". --Francis Schonken 17:18, 16 November 2006 (UTC)

Come to think of it, if we'd ever need a (cursed) nutshell to summarize what this is about, I can't reduce it further (yet) than these three principles:

1. Primary sources (including direct quotes) rather serve the purpose of illustration than the purpose of verification.
2. Appropriate lay-out tools should be used to set illustrations apart from the flow of the article narrative.
3. Ideally, illustrations take between 25% and 50% of the surface of an article.

Commenting on these principles:

• ad 1: sets this quite apart from WP:V and WP:ATT type policies;
• ad 2: that's the part I'd still like to elaborate a bit; this principle indicates the proximity to WP:WAF type of guidelines.
• ad 3: the proposed limits might seem awkward, nonetheless:
  • Less than (approx.) 25% would usually not get an article through FA. The recommendation has to be read that (for instance) if image material on the author Tacitus is limited, illustrations can also be given in the format of the Pliny quote opening Tacitus#Studies and reception history.
  • More than 50% looks like a transwiki-to-wikisource candidate to me (see Wikipedia:Don't include copies of primary sources).

(Side-remark:) due to the "rather", "appropriate" and "ideally" qualifiers in the principles above, I'd think it pretty hard to get this to the rigidity of a policy principle. --Francis Schonken 17:47, 16 November 2006 (UTC)

Actually, on WP:ATT we did find a snappy exception-free description that could, in my opinion, replace all talk about primary and secondary sources and their exceptions. But that proposal only lived for a short time (it was supported by maybe 5 people) and it lost out against the legislation gang before it could mature. The maturation could have been a bit similar to the above, perhaps as roughly sketched as follows (of course much too revolutionary to have a chance in Wikipedia):

All information must be notable, and all information sources must have been published (or made public) and be reliable for the purpose for which it is cited. That implies:

1. Article topics must be based on information in trusted sources; topics that require interpretation must be based on sources that provide such interpretation.
2. In particular, this policy has been sharpened as follows: topics about scientific theories must be based on peer-reviewed publications and historical topics must be based on history articles.
3. For every claim the most reliable sources must be cited. Usually that will be the original source (if available). Interpretations must be fairly cited in accordance with WP:NPOV.

something like that would shrink things dramatically but it would imply throwing away a lot of text that has been carefully negotiated over the last years, including most (or all) talk about "primary sources". As that will likely never happen, I now "unwatch" this page and all policy pages except for WP:IAR. :-))

Cheers, Harald88 23:18, 16 November 2006 (UTC)

Where do I go now?

I just turned up at this article, and it looks pretty interesting. However, it's hard to tell if it is still current, or even relevant at all. It would be nice if it pointed to something which accurately reflects current WP policy and guidelines. cojoco (talk) 23:16, 14 February 2009 (UTC)