Wikipedia:Verifiability

This is an attempt to summarise an existing guideline that we've followed for a while but not written down. Mind you, in writing it down, we may decide it's rubbish!

One of Wikipedia's rules to consider

The goal of Wikipedia is to become a complete, accurate encyclopedia. Verifiability is an important tool to achieve accuracy, so we strongly encourage you to check your facts. However, don't be too keen to remove unverified information at the cost of completeness.

When adding information

Fact checking is tedious, time consuming, and not particularly rewarding. It is unfair to make later editors dig for sources, particularly when the initial content is questionable. Those who write articles likely to be deemed in need of fact checking, for whatever reason, should expect to assist by providing references, ideally when the article is first written. Because of this, it's important to make it easy to verify the accuracy and neutrality of your content. Citing your sources is an important part of this, but not the only factor. Another good rule of thumb is to be specific (and avoid weasel words). For example:

A human rights spokesman said that the incident was part of a wider pattern of violence in the region

This is difficult to verify, because it's hard to know where to start. Many spokespeople may have commented about the incident on many dates and on many occasions, and it's unreasonable to expect someone to check all these statements looking for the one that matches. A better phrasing would be:

Eliza Twisk, of Amnesty International, described the situation in an interview with Channel 4 news on July 8 2000, saying that "This is all part of a growing trend in Europe of violent protest and equally violent response". [1]

This is easy to verify: one could contact Eliza Twisk, or Channel 4, or Amnesty International. As the exact quote is given, rather than a paraphrase, this can be fed into various search engines. Finally, a URL of a transcript is given.

Degrees of verifiability

There are degrees of verifiability. At the one end, there are facts that can be verified fairly quickly by most editors, requiring only resources available over the internet, or at the local library. At the other end of the scale are facts that can only be verified by subject matter experts.

In general, consider the kinds of people who are likely to edit the article in question: the article should be verifiable by these people. Therefore, an article on a sociology topic might include content that can only be verified by a sociologist - perhaps referencing some standard sociology text. However, it should probably not include content that can only be verified by a physicist, because physicists are likely to be spending their time reading and editing our physics articles.

If you are writing on a well-studied field, then it's possible that most of the editors will be reasonably acquainted with the topic, and you can be a bit more relaxed about verifiability. However, if you are writing about a more obscure topic, then you may find that many of the editors have never previously heard of the thing you are writing about, and you should take this into account.

Checking verifiability

There are several reasons you might want to verify something in an article:

The author has a record of contributing inaccurate or misleading information.
The author has a conflict of interest.
There are other errors in the article, and the entire thing needs to be checked.
The article is the subject of an accuracy dispute.
The subject area is one where errors are frequent.
The statement is implausible on its surface.
The statement is key to the entry as a whole.
The statement is overly vague.

Here's a suggested procedure for verifying content.

If you feel the urge to remove a statement from an article, first check the bottom of the article for references.
If there are any, check the sources. If you can confirm the statement using them, leave it in; otherwise, continue.
If there is a talk page, check that. The statement may already have been verified, so there's no need to repeat the procedure. However, if the reference was only in the talk page, move it to the article to help people who might want to check it in the future.
Use your common sense to work out what other resources would help, and check them. If you can confirm the statement using them, leave it in; otherwise, continue.
Move or copy the statement to the talk page, explaining that you have not been able to verify the statement, and stating what sources you have checked.
Optionally, check the article history for who added the statement in the first place, and leave a note on their talk page telling them that their statement is disputed, and directing them to the appropriate talk page.
Anyone may now feel free to try to verify the statement and produce a reference on the talk page.
If you only copied the statement, wait a week (or other random amount of time), and if no-one has found a reference in that time, remove it from the article altogether. Don't worry, it'll still be on the talk page.
If someone does find a reference, the statement should be put back into the article, with the newly found reference. To make it clear which statement used which reference, it might be worth numbering the references and then referring to them in the article like this[1] or like this¹. If no-one finds a reference, the statement can remain on the talk page indefinitely.

Once you've successfully verified something, consider whether you can edit either the article, or the talk page, to make it easier for the next person.

Dubious sources

For an encyclopedia, sources should be unimpeachable. An encyclopedia is not primary source material. Its authors do not conduct interviews nor perform original research. Hence, anything we include should have been covered in the records, reportage, research, or studies of others. In many, if not most, cases there should be several corroborating sources available should someone wish to consult them. Sources should be unimpeachable relative to the claims made; outlandish claims beg strong sources.

Sometimes a particular statement can only be verified at a place of dubious reliability, such as a weblog or a tabloid newspaper. If the statement is relatively unimportant, then just remove it - don't waste words on statements of limited interest and dubious truth. However, if you must keep it, then attribute it to the source in question. For example:

According to the weblog Simply Relative, the average American has 3.8 cousins and 7.4 nephews and nieces.

This is similar to how we try to achieve a neutral point of view.

Obscure topics

Verifiability is one problem with articles on obscure subjects. By concentrating on verifiable subjects, we also concentrate on important subjects. The disadvantage of this approach is that we may tend to concentrate on recent trivia--celebrity gossip can be verified from current newspapers, whereas events in the past can be harder to locate documentation for.

However, just because some information is verifiabile, doesn't mean that Wikipedia is the right place to publish it. See what Wikipedia is not. In particular, many Wikipedians feel that verifiability is insufficient to determine whether an obscure topic should be covered in Wikipedia. See Wikipedia:Criteria for inclusion of biographies and Wikipedia:auto-biography, for example.