Wikipedia talk:WikiProject Fact and Reference Check/Archive 1

Archive 1Archive 2Archive 3→Archive 5

There were many ideas in the Wikipedia channel about this topic, I saveed the log and will extract the basis of the ideas here later (and will attribute them as well). --ShaunMacPherson 14:01, 1 Sep 2004 (UTC)

  • It is a good idea to comment about the accuracy of facts - each and every fact. But can we exclude facts on mere suspicion of being inaccurate based on our limited checking of accuracy?
That is an interesting point, a new proceedure may have to be thought out if there is a disagreement about facts or Wikipedia evolves to the point where possibily some material would be taken out because it comes from a less reputable source.
I think though that Wikipedians currently do, and likely will in the future, a good job about weighing the factual statements in the articles. Facts or opinions that are not widely held or mainstream are usually directly attributed to an individual (e.g. Scientist X has stated that Aliens have visited Earth). It would still be factual to indicate that a certain person holds a certain viewpoint or belief - All that will happen is now we can actually track down their viewpoint and factually know that they had indeed said it. --ShaunMacPherson 15:00, 1 Sep 2004 (UTC)
  • This is a very ambitious task needless to say. But it is desirable. What is desirable is some software support for this. If it is possible, we could broaden the scope to question even the neutrality of sentences. --Hemanshu 14:08, 1 Sep 2004 (UTC)

Implications for Wikipedia

There have been a few comments that people are worried that this project may interrupt how Wikipedia currently operates. I don't think that fact checking need necessarily interrupt Wikipedia, what it will seem to do is is open up a new avenue of opportunity to people who are not verbal or literary, but have excellent research and fact finding skills. Those that just want to write can do so, and those who want to verify that the autoreferencing code is working properly, and that new facts are numbered and checked have their part to play as well. (Nothing precludes people from doing both too :) ). --ShaunMacPherson 17:03, 1 Sep 2004 (UTC)

Examples

I'll be working on forming a more concrete example by actually formatting a current Wikipedia article, and making its reference page. Of course I'll have to do it manually since the autoreferencing code has not yet been developed, but hopefully it will be a better preview then the Jack example, on the main page, as to how such fact and referencing might be implemented. --ShaunMacPherson 17:03, 1 Sep 2004 (UTC)

Tools

Some ideas:

  • Some kind of auto-footnote-numbering feature would be nice (along the lines of the example (-#)?), for which a MediaWiki modification would have to be made, I guess....
  • Templates for whichever authoritative sources we choose to cite on a regular basis. Then we could put a proper bibliographic reference in the footnote without having to retype or copy and paste it manually every time....

-Sewing - talk 18:26, 1 Sep 2004 (UTC)

I like the idea, but since people are constantly adding to articles, we would have to tag checked articles to show when they were last checked. Danny 18:29, 1 Sep 2004 (UTC)

Good point. Anyhow, the idea is that the MediaWiki mod would take each (-#) or whatever and produce something like this for a finished product... -Sewing - talk 18:35, 1 Sep 2004 (UTC)
Interesting method, but only one reference could be associated with each fact with your superscript way. This way, example 2 from the main page I made yesterday Example 2 maybe a better method as you can verify each fact unlimited times with mulitiple sources. Edit the page to look at my ad hoc tabs, less ugly ones will be made hopefully :).
It would also be useful if you could make visible on the actual article the sections being quoted, but you'd need a script to do that. --ShaunMacPherson 18:15, 3 Sep 2004 (UTC)
I guess it would be a 2-step process: first insert the (-#)'s, submit the page, then go back and add the actual references??? So on the first pass, all the software would do is create a "References" section heading, automatically number each (-#) sequentially, and create a corresponding section heading. Better yet would be to use something more compact than a separate section heading for each footnote, however.... -Sewing - talk 18:41, 1 Sep 2004 (UTC)
Regarding Danny's concern, perhaps the WikiMedia utility would also add a timestamp to the article for when the references were last edited (by doing a diff on the previous version). I added an example to my sample Pacific Ocean article. -Sewing - talk 19:11, 1 Sep 2004 (UTC)

Here's the feature request for footnotes on MediaZilla: http://bugzilla.wikipedia.org/show_bug.cgi?id=192 - There's also some discussion on m:Footnotes. --Kurt Jansson 22:12, 1 Sep 2004 (UTC)

I've just went to that bug section. This is getting very complicated, can anyone explain the process which proposed additions to the Wikimedia code (what Wikipedia runs on) are put forward? I've added a comment there directing people to the project page, hopefully they see the benefit of crossrefercing articles :). --ShaunMacPherson 14:00, 21 Sep 2004 (UTC)

More tools..

  • Some nice reports that may be useful in this project can be found here, thanks to Topbanana

--Quinobi 07:04, 15 Oct 2004 (UTC)

Example?

Instead of the trivial hypothetical "blue pants" example, could someone take on a real, non-trivial article as an example of how this might work in practice? I suggest starting with something that is reasonably stable, not terribly controversial, and not previously chock full of references. Maybe mayonnaise? -- Jmabel 19:36, Sep 1, 2004 (UTC)

Hi! I did an example the day before yesterday but it may not be that visible on the main page so this is a link to it: example 2. I find it looks a little gawdy, perhaps when the coding requotes the statement, it takes out wikiformatting? Feel free to copy the text underneath and lable it example 3 etc. if you have an improvement. --ShaunMacPherson 18:10, 3 Sep 2004 (UTC)
Hmm, I think I prefer with the footnotes more. Mind you, I still think it would look better on a subpage. I know we're trying to get rid of those, but for some articles- especially the larger ones- having the references in a section on the page would be quite annoying, since there would be so many. I prefer the footnotes because they're easier to follow. With footnotes, if I want to make sure that spiders really do have eight legs, I just click on the footnote. With this, I have to scroll down and look for the statement. Well, this is talk:, so that's just my opinion- feel free to argue! -[[User:Frazzydee|Frazzydee|✍]] 18:14, 3 Sep 2004 (UTC)

Potential problems and complexities

I can see a few possible difficulties with fact-checking in this fashion. Many of the problems mentioned on m:Footnotes apply. Assuming every fact is to be referenced in some way, I think it will be important to have the capability to tag a large block of text with a single reference; for instance, the very short stub (found by random-page browsing) Stubbekøbing contains this text:

  • Stubbekøbing is a municipality in south Denmark, in the county of Storstrøm. The municipality covers an area of 156 km?, and has a total population of 6,836.

Even this short stub contains many facts:

  • Stubbekøbing is a municipality
  • It is in south Denmark
  • It is in the county of Storstrøm (which by implication is also in south Denmark)
  • It covers 156 square km
  • Its population is 6,836

Referencing each fact separately, especially for long articles, would be a monumental task. Delimiting which fact(s) are being referenced seems like the only reliable way to do it. However, whenever a new fact is inserted inside a delimited chunk, new delimiters must be specified and/or a new reference added (increasing the complexity of our already complex wikicode markup).

This leads into another problem - tracking changes in the article, and ensuring that references are not wrongly associated with unrelated facts (or vice versa). Aside from re-checking articles after each addition or alteration, I can't think of an obvious way to prevent this from causing problems.

One possible way to eliminate the need for most references is to use references only for facts that may be unintuitive, not widely known, under dispute, or otherwise contentious. Of course, this opens a whole new can of POV worms: how to decide when a fact is incontrovertible enough to not require a reference.

Any facts with a URL as a reference may quickly become outdated, especially if the reference is a news article that may not be archived for more than a few weeks. Obviously, additional information such as author, publication media and date would be strongly encouraged.

I like the concepts discussed on m:Footnotes, and think they would go a long way to solving the lingering question of accountability, aside from the uglification of wikicode caused by them. A simple list of references alone would give us more credibility in many cases. I can think of no publication that references every significant fact, so we would likely be one of the first to attempt it.

Some brainstorms:

  • Use a form of meta-wikicode for all references. The meta-wikicode would not normally be visible when editing the article (much in the same way that formatting marks are not visible in a word processor unless their visibility is enabled by the user). This would allow the wikicode to maintain its current level of complexity for most editing tasks; users doing fact-referencing could view the meta-wikicode for the purpose of inserting references, delimiting facts, adding additional comments, etc. Whether these references/footnotes are displayed within the article text (which could be disruptive to readers not interested in seeing tons of links to all the references) could be a user option "show/hide references"; toggling display could pretty easily be implemented with CSS. Undoubtedly this would require significant implementation changes, but this seems to me the best of all possible worlds.
  • Use some variant of the diff system that we use to view article changes; we can already deduce what user or IP address contributed exactly what bits of text to any article. A new input box could be added to the editing page for contributors to document their sources as they add or remove information; those references could be included in the article as footnotes, or as meta-markup such as that mentioned above.
  • Use a separate references page for every article, which would contain offsets or numerical references into the article text, i.e. (Paragraph 3 sentence 5: Joebob Jones' article "Foo" in the Weekly World News, December 17, 2001), or just a reference that correlates with an identifier inserted into the article wikicode. This could be tricky to implement in a manner flexible enough to adapt to ongoing changes in the article, but might also avoid some additional complexity in the article's wikicode.

Overall this seems like a very worthy endeavour. I think it could work if we give it enough thought. -- Wapcaplet 01:13, 2 Sep 2004 (UTC)

You present an extremely important point there- if we reference every fact in an article, it could get pretty messy- even with footnotes- it'll litter the page. Can we limit one footnote per sentence, and put two references in one footnote? Obviously, we're going to have to use some common sense. If something is widely known, it need not be referenced (i.e. the humans have two legs example). -[[User:Frazzydee|Frazzydee|✍]] 21:35, 1 Nov 2004 (UTC)

I am uneasy about the practicability of this idea. To work we would need to define the form the references are to be given, and ensure everyone sticks to them, eg Joe Bloggs, Clog Dancing for Dummies, Complexity Press, (2004), or Bloggs, Joe ---- . There is no consistency about how references are already written. If references are footnoted, means must be made to automatically update the numbering if new edits and references are made. In the New Oxford Dictionary of National Biography, to which I am a contributor, footnotes are not given, but a block of references is given at the end, including printed sources, portraits, sound recordings, motion pictures etc, and I suggest we adopt the same principle. To fact check and reference every statement in a paper is a big addition to the work of editors, and could well impact on the length of articles submitted due to the amount of extra work needed to do it comprehensively. Apwoolrich 06:31, 2 Sep 2004 (UTC)

Some of the facts to be referenced in the spider article are actually matters of definition. Are we going to give citations for statements like, "Humans have two legs"? Sometimes we will need citations for shady areas, e.g., the idea of "races" of various non-human organisms in biology was subjected to a long debate and part of the reason for that debate was that not all participants realized that different specialties in biology habitually use the term in different ways, avoid its use, etc. The places where references are likely to be most useful are matters of objective fact that are not well known, e.g., that spider "silk" is a protein structure. People who find that matter interesting would benefit from references that give more information than would be appropriate in a general article on spiders. But all of the information that any well informed student of the subject will agree with, the number of legs, the lack of wings, etc., etc. is found in compact form in standard reference books such as B. J. Caston's How to Know the Spiders. P0M 20:26, 21 Sep 2004 (UTC)

A proposed strategy

Very little effective fact checking will be achieved on our massive body of information merely by arbitrarily starting somewhere, breaking things down to atomic facts, and checking them. As in software quality assurance, we need to spend our resources in such a manner as to get the most "bang for the buck". I would suggest that once we've decided what we will consider as acceptable verification (or falsification) of a factual statement (and that will not be a simple matter: sources can contradict each other), there are several reasonable strategies to choose appropriate facts to check.

The key principles here are (1) nothing should be immune from fact-checking but (2) we should try to direct a disproportionately large amount of resources to areas we think are most likely to contain falsehoods.

Here's what I'd propose:

  1. At least 5% of fact-checking time devoted to thorough checking of articles chosen truly at random, e.g. using the "random page" feature.
  2. At least 10% of fact-checking time devoted to thorough checking of known controversial articles. This is not limited to the articles that are overtly labelled as controversial, but would include topics related to, for example:
  3. At least 5% of fact-checking time devoted to random assays of articles chosen truly at random: after using the "random page" feature (or equivalent) use some comparably random method of picking three facts from the article to try to verify. If those verify, move on; if not, see the remark below about neighborhoods.
  4. At least 10% of fact-checking time devoted to checker-driven assays of articles chosen truly at random: after using the "random page" feature (or equivalent) the checker should look for the claims in the article that (subjectively) seem most likely to be false. If the first three or four verify, move on; if not, see the remark below about neighborhoods.
  5. At least 20% of fact-checking time should be dedicated to articles which one or another wikipedian (not necessarily one usually actively involved in this fact-checking project) has identified as "suspect". This might work somewhat like cleanup: anyone can put an article "in the hopper", but the people doing the work decide which articles actually to work on. Reasons to identify an article as suspect might include:
    • sources were cited but can't be found (e.g. a dead web link as a citation).
    • one or more factual errors already identified (and possibly corrected).
    • edits by people who have been known in the past to insert poorly researched (or outright false) information.
    • "It just smells wrong."
      Sorry to stick my ideas in the middle of your lest, but this is a good idea. Perhaps a new tag that is the equilivant of "Please check this, the facts seem unusual" :). --ShaunMacPherson 18:21, 3 Sep 2004 (UTC)
  6. At least 20% of fact-checking time should be dedicated to various neighborhoods of where false statements have been found: e.g. same article, same author, closely related topics, etc. Over time we will discover which of these "neighborhoods" is the most productive of further identified factual errors.

That leaves 30% of resources to be allocated more arbitrarily, e.g. by people just picking something they are interested in checking and going for it.

Obviously none of this is set in stone, and the percentages might change with time or other categories might be added: e.g. if we are trying to turn part of Wikipedia into a published book or CD-ROM, that would merit getting priority. -- Jmabel 06:39, Sep 2, 2004 (UTC)

Some have suggested that the featured articles should be checked before becoming featured articles, and past featured articles could be varified as well since they are the epitome of the best of Wikipedia.
It would also be useful to sign up some people with electronic encylopedias. I only have encarta, anyone know of other good ones we can do fact checking with? Perhaps we should create a list of sources we will use to keep track of the references? It would be useful if the coding could do this too. --ShaunMacPherson 18:21, 3 Sep 2004 (UTC)
As a strategy I suggest you adopt the 80-20 rule. Focus effort on the articles that are most controversial and do the fact checking there. That would give the project the most positive effect on Wikipedia and gain the most publicity and so potential helpers. Also, if this project is going to work it has to be shown to work on the most controversial articles. :ChrisG 20:18, 13 Sep 2004 (UTC)

As I point out above, there are too many separate factors here for a guideline as simple as 80-20. -- Jmabel 21:43, Sep 13, 2004 (UTC)

I think your overestimating the resources (i.e. interested parties) that are avaliable at the moment to support this project. You should start small and prove its worth. The strategy you suggest implies that you have a fairly sizeable group working on the project. : ChrisG 08:52, 14 Sep 2004 (UTC)
Its hard to start crossreferencing without the approperate <<tabs>> in place to autogenerate the quoted facts to be referenced. We could go around putting in comment tabs around facts and copying the quote to the reference section created below the article but that would tend to ruin the article ;).
What we need is to get the techs interested, and I've already spoken to a few. --ShaunMacPherson 14:00, 21 Sep 2004 (UTC)
I agree with jmabel's general plan, and add the m:Instruction creep article to the strategy discussion and project guidelines here. Quinobi 10:43, 15 Oct 2004 (UTC)

There's a related discussion going on here. You can see the entire thread by clicking on the subject line.--Eloquence*

I noticed that many people are getting very interested, and eager, to make Wikipedia a more authoritative source of information. I also have been on some of the mailing lists and there has been discussion about refereeing articles like the old Nupedia. If you see any discussion or comments about referencing, fact checking, foot notes, please leave a brief comment about this project. Hopefully it will be a nexus for developers and wikipedians to discuss how to make wikipedia a better and more authorative source of information. --ShaunMacPherson 13:57, 21 Sep 2004 (UTC)

Generated Facts?

What about for fact that are not in the format needed. For example, say you have the date of birth of Napolean, and the year that he was exiled, and from those you get his age. Could the fact checking mechanism explain the math used to get the new fact? Jrincayc 01:32, 22 Sep 2004 (UTC)

You'd quote that, and then quote his age from a source I guess. --ShaunMacPherson 08:14, 2 Oct 2004 (UTC)


How about this as a way forward?

I've been thinking about this problem with my back brain, and I've come up with this suggestion using existing features of MediaWiki.


Basically the idea is create a namespace for articles about primary source material. This namespace would record all sorts of information about the primary material, for example:

  • Title
  • Author(s)
  • Type (e.g. article, book, magazine, letter, white paper, newspaper, webpage)
  • ISBN(s)(Could be multiple if different additions)
  • Web address (External pages and especially Wikisource link if available)
  • Abstract
  • Wikipedia article if encyclopediac
  • Facts


The facts are added as sections within that article about the primary material. These facts can then be reused throughout multiple articles. If someone reads a primary article they can create and amend the related facts in one go, which is the natural way to discover and check facts. The talk page could be used to discuss any questionable facts. You reference the fact by linking to the article and the appropriate named section (for paper media the page number).

This namespace could be used for all sorts of primary material, e.g books, webpages, articles, magazines. I would imagine there would be different templates for each type of primary material. I think it quite likely that academics would be very interested in such a primary source namespace. Adding a Reference or Source namespace to Wikipedia would only require requesting it from the developers.

Possessing a seperate namespace for articles and referencing of primary source material has a number of advantages. It is far more sustainable, because the references will properly checked as a whole against the primary material. And providing the primary material is rigourously checked, it would be fairly trivial to check a wikipedia article for correct referencing to the articles in the Reference namespace. Clearly such a namespace with appropriate referencing would strengthen the academic credentials of Wikipedia.

Obviously the linking mechanism is a bit ugly and obtrusive still; but would be an easy change to the code. It would also be very useful if the wiki-ref links allowed internal Wikimedia links so that What links here would tell us which articles are referencing the facts, which would obviously be vital to check and update references.

In the future there would be a number of ways developers could create a list of footnotes within a Wikipedia article. The data being organised into sections could be manipulated in a number of ways or even moved to a proper database. This would mean that people could work on this project without fear of wasting their time. 

:ChrisG 21:28, 4 Oct 2004 (UTC)
Excellent work. That is very close to the idea I had. In your version you click super script note and goto the reference. The problem with doing it this way though, with super scripts -> reference (instead of to the fact statement), is there can only be 1 reference per fact unless you have a string of superscripts for each sentence. With example 3, you can have multiple references for each fact. All that would need to be done is a redirect to the statement instead of the reference :o).
I like in example 4 how it automatically brings you to the statement in question. I wasn't able to do an example with it in #3 but now I think I can using your example. I also really like your idea of actually having a seperate page with the information seems very much superior, as it does not clutter the article. Plus having a seperate page like you suggested allows us to have our own discussion page on factual issues.
Looking at the code though there is a lot of junk in the edit view in example 4. Even with example 3 it looks very junky with tabs in the actual code. We need away to do clean <<brackets>> around fact statements, and have the ability to hide these brackets for people just editing the article and not doing fact checking, but have these brackets follow the text around. We really can't go forward much, I don't think, until a <<bracketing>> fact statements are coded into MediaWiki code. I have not heard from the tech people I spoke to so if you know anyone (Timwi, TimStarling, other coders) please encourage them to devote some time into coding a bracketing solution. The #1 criticism is always the authenticity / reliablity of the facts in Wikipedia so I think they should be focusing on this issue. --ShaunMacPherson 09:43, 5 Oct 2004 (UTC)

I can see the advantage of having multiple references for one fact collected together. But I think that it is working back to front. Trying to keep multiple references to a single fact correct would be a nightmare to do as it would be dissassociated form the primary sources.

A fact relates to a specific primary source. It makes more sense to collect all the facts relating to a primary source attached to the article about the primary source, because it would combine the process of reviewing a primary source and noting down useful facts in one process. I'm sure somewhere on Wikipedia some other people were interested in starting a WikiReview project and so they would be interested. It would also mean that if a person reviewed a primary source properly, it would be suitable for use in multiple article, maximising the value of doing it. People looking for facts in articles would also have a seperate namespace they could search to look for suitable facts.

Obviously for controversial articles you would see people jumping between both namespaces, creating a fact checking primary articles and then adding references on the wikipedia article. But that is to the benefit of everyone; because the arguments about the facts can take place in the reference namespace, not the wikipedia article talk page. I think it would remove a lot of controversy. Clearly there is a lot of junk using this method; but the important thing is it works now, so you don't have to wait for developers to add features. In addition it would be quite easy to ask developers to clean the process up, by amending the [reference] syntax so we could use it also to refer to internal wikipedia links and automatically making them superscript. Developers like easy changes. Allowing referral to internal Wikipedia links would mean the 'What links here' would identify where a primary source was used to support an article allowing for easy fact checking.

I can understand your desire for a bracketing syntax and in the best of all worlds it would be the ideal solution. But it is a huge change to make in the code, and so you would need a committed to developer to work on it; and it would be a nightmare to do. It looks very simple when you bracket a specific phrase as a reference; but once people start editing the document you have a nightmare:

  • The update process would have to check if information has been amended within a bracket.
  • Flag this in some way
  • What if you want to create footnotes for different parts of a phrase, e.g. "<<<<Jesus>> Christ>> was <<born in <<Nazareth>>>>". How on earth to you make sense of fact that I want to create footnotes for Jesus the name, Jesus Christ the title, born in Nazareth and Nazareth.
    • Then what happens when someone edits it like this, "<<<<Jesus>> Christ>>, also called <<Jesus the <<Nazarene>>>> was <<born in <<Nazareth>>>>"?
  • If you try to hide the bracketing syntax from editors then thats another level of complexity, and you cause confusion about what editing to make.

It would be hard to find a developer interested in solving that, because it is so difficult. Whatever solution they come up would almost certainly not be liked by everyone. Its a complex requirement and the best way to build towards it is incrementally.

It is far simpler to add a reference after the fact you want to reference. People understand this from books; and many people will be careful with the references and indeed check them as they go. Fact checking proper should take place in the reference namespace.

From my point of view this project as its first priority needs its own namespace in order to start fact checking and reviewing of primary sources. Once that is set up, people writing articles will soon start referencing the reference namespace and creating articles about suitable primary sources. The method of linking to these references is secondary and a technological problem. 

:ChrisG 14:40, 6 Oct 2004 (UTC)

Submit a proposal

I would really like to get started on this project, but I can't tell if there's a consensus for how to do so. Could everyone with an opinion summarize it in 50 words or less, or just add your name after someone else's summary if you agree with it totally? I'm probably going to do something with the music of the United States series using a user subpage to list cite facts. I'll start with music of the United States before 1900 with a reference page at the imaginatively titled User:TUF-KAT/music of the United States before 1900, if someone wants to see it. Tuf-Kat 02:41, Oct 13, 2004 (UTC)

I think that project as currently proposed is ill-conceived and misunderstands what fact-checking is about. It is not about redundantly confirming the obvious. It is about verifying or refuting the reasonably questionable. -- Jmabel|Talk 18:50, Oct 13, 2004 (UTC)
Change the page if you think it can be worded better :o). I really was hoping more people would take an active role in fixing the project page up instead of leaving me to do it heh. Go crazy everyone and make it better, there is always the history if the members want to revert some changes.
What I think the project is about is fact checking all facts on Wikipedia, please put what you think it should be about.--ShaunMacPherson 04:06, 15 Oct 2004 (UTC)

I've already explained my proposed approach above (#A proposed strategy). It is based on the principles of quality assurance. It looks like others here have a very different idea in mind. I think it's epistemologically flawed and I've said as much. -- Jmabel | Talk 07:00, Oct 15, 2004 (UTC)

The key principles here are (1) nothing should be immune from fact-checking but (2) we should try to direct a disproportionately large amount of resources to areas we think are most likely to contain falsehoods. (Quote from #A Proposed)

I agree with this completely. I guess there would be people who format the articles with footnotes, people who check facts, and people who respond directly to 'hot' topics with a lot of conflict, usually the same topics that contain a lot of falsehoods. I'd guess most of the time would be usefully spent in this 3rd catagory, on directing energy at areas that contain falsehoods, like you say.

None of this can be effectively done though until a useful way to do footnotes are incoroprated into mediawiki. I suggest a <<bracket approach>>, but I'd take anything :). --ShaunMacPherson 07:53, 15 Oct 2004 (UTC)

But, also, merely finding an independent citation does not "prove" the truth of a statement. It's one thing to get articles well-referenced (and I'm all for that, and I probably do about as much in that direction as anyone on the English-language Wikipedia), but it's another to actually be able to say they are anything more than referenced: that requires expertise on the subject matter. See, for an interesting example of how tricky this can be, the notes in Armenian (people) about the references for the population statistics. -- Jmabel | Talk 18:44, Oct 15, 2004 (UTC)

That is why I think the fact checking, or rather gathering of references to cite, should take place in a reference namespace. That way the argument about whether a reference is accurate is argued about in the article about the source. Arguing about reference material in Wikipedia is currently too far away from the original source material; and it may happen across multiple articles to no advantage to anyone. If we have a reference namespace the argument about the source material is carried out in one place. Any facts or references derived from that namespace could be used in multiple articles and it would have a clear audit trail. :ChrisG 14:13, 16 Oct 2004 (UTC)
Yes, that could be very useful. In fact, I could imagine a very useful set of discussions on the reliability or otherwise of various sources that do not necessarily merit articles of their own. -- Jmabel | Talk 19:58, Oct 16, 2004 (UTC)

As I see it the reference namespace would be for academic studies, books and any other primary material. Of course, it wouldn't have the source material, it would be articles set up in a structure format about the source material, e.g:

  • Title
  • Authors
  • Publication date
  • Publication details
  • Abstract
  • Useful references.

I would imagine different formats for different types of primary material. The talk page could be used to discuss any aspect of the article; but I would presume it would be most useful to discuss the subtleties of referencing and fact checking.

A very small percentage of this primary material would also be suitable encylopediac material, which would not be a repetition of the reference namespace, because it would be presented in a different less formal way.

I can see a crossover here with various other project ideas, Wikiversity, WikiResearch, WikiReview etc. So perhaps it wouldn't be a reference namespace; but instead its own Wiki with its rules; and probably higher academic standards. I would imagine such a Wiki could well be more appealing to academics than Wikipedia itself. This disadvantage of a seperate Wiki would be that 'What Links Here' would not tell you who was referencing the source material. :ChrisG 23:54, 17 Oct 2004 (UTC)

read

how do you get a free acount