Sadly, real documentation has been lacking from at least TRS2004, which came with an extensive user-guide and content-creator-guide, and other nice quick guide. I'm glad I still have books and PDFs tucked away. Even with the updates and changes made, much of that information is just as relevant today as it was back in 2003-2004 era.
While I totally agree, I don't think that many users fully realise the problems involved in providing and maintaining adequate user documentation, which is probably why many software developers skip it or provide it at a very minimal and/or inadequate level. There will, of course, be many different interpretations of what "adequate user documentation" is.
Going from printed (on paper) to digital (e.g. PDF, Web) documentation is not as great a saving in time and effort as one might think. I speak from experience in the commercial production (writing and editing) of both paper and digital documents. The development time (writing, editing, proof reading, fact checking, getting permissions, etc) involved in both forms is much the same. While the digital form does not have the extra delays involved in the printing and distribution, it has other "delay factors" such as ironing out the differences between different platforms and software packages. PDF, Web and Wiki are not the universal "one size fits all" display formats that some may believe them to be.
Keeping existing documentation up to date is not a simple process either. In the case of software documentation each new version release or Service Pack can force entire sections (even chapters) to be rewritten and new sections added. Each change requires the usual checks (proof reading, corrections, etc) but also "back checking" with the rest of the documentation for consistency. One of the issues I always had when editing the work of multiple writers into the one document was enforcing a consistent style. There were times when I had to rewrite the entire contribution of an author.
In short, adequate documentation is expensive to produce - it takes resources and it takes time. Unless you are Micro$oft, Apple, Adobe, etc, both could often be better used in improving and updating the product. I think that most users would argue that getting a feature to work is more important than documenting why it doesn't.
And what happened to the often quoted claim that "real geeks don't read manuals"?
PS: I quite like the Trainz Wiki but it is very under utilized, both in creating new entries and reading those that are there. As I have found, keeping it updated takes time and effort.
My thoughts.