1script - 6:40 pm on Aug 30, 2010 (gmt 0)
You give a great example with this .txt file. The pure distilled content of a .txt file should make it very desirable to bots because most modern HTML marked-up pages serve at best 40% of content and the rest has to be thrown away.
Just bare text served up as naked as the day it was created
A .txt file is also easier to parse because there is nothing to parse, it's ALL content which is I think where it becomes very relevant to the theme of this particular discussion: title tags are and have always been in the most parse-able part of the page's code. I've always put them right after the <head> tag and I'm sure most of those SEO types hanging around here did the same. The lower you go on the page the more weird stuff like embedded CSS, scripts and so on gets on the way (and that's assuming the markup is error-free to begin with which it often isn't) and hinders the bot's ability to distill the content they are after.
This is a pure speculation on my part but what if importance of titles in the past was only brought about because they were always most available to G*bot, unlike the rest of the content of the page? Now, with the much celebrated Caffeine muscle G* may now be able to parse pages better (apply CSS rules, execute JS and so on) and therefore importance of title tags is diminished because there is just more other content from the page to index?