User:Pgallert/On reliable sources

For the official Wikipedia policies on reliable sources, see Wikipedia:Identifying reliable sources and Wikipedia:Reliable source examples.

The problem

Articles should be based on reliable, third-party, published sources with a reputation for fact-checking and accuracy. This is the rule in a nutshell but it opens a can of worms: Except for a few clear cases it is hard to find sources that fulfill these criteria. Strictly speaking, the RS rule restricts sourcing in Wikipedia to the following:

1. Other well-known encyclopedias
2. Books widely prescribed in university-level courses
3. PhD theses
4. Peer reviewed articles in scientific journals
5. A few dozen reputable newspapers
6. Web content by notable authors or organisations, writing about their area of expertise

That is not a lot. It excludes most of what is used in Wikipedia today, particularly:

1. Most books (the publisher seldom checks facts)
2. Most newspapers (it cannot be determined whether they are really uninvolved)
3. Most web content (their reputation for accuracy cannot be established)
4. Government publications about their own country (not third-party)
5. Government publications about other countries (they might have an agenda)
6. CIA world fact book (What they know might be reliable, what they publish, not)

The solution

Can't we use any source that presents facts without interpreting them the wrong way? Sure, but we must make sure it is the facts, and we must make sure they are presented the right way.

We, the writers of an encyclopedia. Don't blame the sources if you used them to support a statement which turns out to be wrong. The list of sources that do not need to be checked is short, see above. All other sources must be scrutinised by the writer who uses them to produce encyclopedic articles, and that takes time, effort, and common sense.

But how do we scrutinise our sources? Where and when do we look for spin? What do we have to look out for?

(Un)Reliability of sources is relative

How often do we find questions like "But how reliable is source X?" in various non-article namespaces of Wikipedia? Thousands of times, of course, and seldom a straight-forward answer will follow. This is because no source is reliable per se, for every statement. So what can be done? Let me introduce (Peter's?) Two laws of publishing:

1. Nothing ever is published without a reason.
2. For every proposition there is exactly one source that is the best reference for it. This source can be found, and it is the one and only source Wikipedia or any other encyclopedia should use to support the proposition.

Two major questions to ask

Instead of trying to define a long list of reliable and unreliable sources and coming to an answer nothing short of "it depends", one can ask two simple questions:

1. Has the author knowledge in this subject area?
2. Why has this been published?

If the answer to (1) is "No" or "Don't know" then the source is not reliable, assuming your "Don't know" answer is the result of some elaborate (re)search. Examples:

• A court reporter writing about computer viruses for the first time very likely has not produced something reliable on this occasion. No matter how fantastic the newspaper is with court reports, or otherwise.
• Travel books can be very reliable sources—on travel information. Anything else they contain, like the country's history or an evaluation of its politics, is likely plagiarised from somewhere, or a complete invention.

If the answer to (1) is yes, proceed to question 2.

Answering (2) is difficult, of course, but think about Peter's first law: There must be a reason. If the answer to (2) raises suspicions about neutrality, conflict of interest, bias, or anything else that is fishy, find a better source. If there is no-one else who makes this claim, dismiss the information as a possible deception or exaggeration. Examples:

• Government reports on the legislation in their country, on history, actually on anything, is not an iota more reliable than the private web page of your neighbor's little sister. Probably less reliable. They want to present themselves in a good light. The better governed countries will just omit unpleasant facts, all others might even have deliberate lies in their official information.
• If money is involved, scientific articles can be fishy, too. These days, universities do not pay for most of the research that is carried out by their faculty. It is sponsored by industry, and the flow of money dictates the results. This list is growing by the day. Would you believe an Altria-sponsored cancer study? And after you found out that Altria was previously named Philip Morris? That Marlboro is one of their brands? Aha.
• Even if no money is involved, scientific articles can be fishy. For two decades, no research published by employees of the Allensbach Institute could defeat the theory of the Spiral of silence (1974). Well, this theory was proposed by Elisabeth Noelle-Neumann who happened to be the director of this institute. Surprising, isn't it?

If the answer to (2) is such that you do not doubt its accuracy in any way, then you might just have found a reliable source. This can lead to some weird examples of reliable sources, backed by no other policy than Ignore all rules, but sometimes this is the only feasible choice. Consider this:

1. You want to write how many members RNP, the Really-New-Party, has. What is the worst place to look? Right, its own web site. They might inflate the numbers. What is the second worst place? Their opponent's web site, they might deflate the numbers.

   But what if all political parties publish the same number of RNP members? One (RNP) has an agenda, the others have no clue. But the situation that none of RNP's opponents seems to doubt their membership figures just makes this information a bit more reliable. Much more reliable, actually, than any news article about RNP, where you likely have no way to determine the journalist's, or the editor-in-chief's, party affiliation.

2. What could be less reliable than an advertisement? "Palmwag Country Lodge, situated in beautiful Kaokoland 100km south of Sesfontein, requires the services of a conservation specialist [...]".

   You cannot use this as a reference for the beauty of Kaokoland, of course, but you could use this to specify the location of Palmwag Lodge; Why would they lie? And actually, it does establish some credit towards the diversity, possibly the fragility, of the landscape: Why else would they hire a nature conservationist?

3. The Farmers' Association of Seeis refurbishes part of a cemetery and puts up an information plaque, stating the circumstances of this renovation. Is this a reliable source? Well, for one it is not published in the context of this Wikipedia content guideline, even though it is published according to the Wikipedia article about publishing. That's confusing. Furthermore, it is of course not independent, and we can safely assume that the Farmers' Association of Seeis did not establish a record of fact-checking and accuracy—this plaque might well be their only publication, ever.

   But the farmers are undoubtedly the experts on their own clean-up work on the cemetery, and why would they lie? Whom would you believe if a major newspaper claimed a different year of refurbishment? See, a reliable source. (See the Seeis article on how to actually do a referencing like this)