Searching For A Word Versus Searching For The Thing

A search box matches text, but what you want is almost never a piece of text.

The most useful thing to understand about any search system is also the least intuitive: it is matching strings of characters against documents. It does not know what you mean. Modern systems are extremely good at guessing what you mean, so good that they conceal this fact for most everyday queries, and then fail without warning on the queries where it matters. You want a fact about a category of thing. The system finds pages containing your words. When your words and the thing coincide, everything works. When they diverge, you get nothing useful and no explanation.

They diverge constantly. The thing you want may be called something else entirely by the people who write about it: an industry has one word, regulators have another, the public has a third, and the historical record has a fourth that nobody has used in forty years. The thing may have no single name at all and be describable only as a relationship between two other things. Or the word you chose may be shared with something far more popular, in which case the popular meaning will swamp every result and you will conclude that your subject is undocumented when it is merely outnumbered.

Once you see this clearly, the countermeasures are obvious. Name the thing several ways in several attempts rather than repeating one name louder. Add a word that only your subject's world would use, so that the popular meaning falls away. When you find one good page, harvest its language: it will be full of the exact terms the field uses, and those terms are the keys to the rest of the material. You are not searching for information. You are searching for the words that information is written in, and then reading the information you find.