Finding five-year-old writings on the internet is more like 5th century historical research than you would think.
Actually, I'm really rather serious there. I'll be quick with this one, so bear with me.
The last couple of posts I've made have seen me going back to old message-board postings, tracking down dead blog entries and the like. Now, when it comes to google, I have m4d l33t skillz like you wouldn't believe. As Rutger Hauer says in possibly his finest work (the video to Kylie Minogue's On a Night Like This), "That which you find impossible, I find easy." I can get old cached stuff, and I know about the Wayback engine and similar projects. But there is a wealth of real, valuable information that is gone forever.
Coincidentally, I just finished reading Peter Heather's Fall of the Roman Empire,and the similarities between the problems of sources for historians of the late Roman Empire and somebody looking for stuff only two or three years old is surprising. The originals are gone, but digests exist, as well as commentaries which quote more or less directly from the original. Where you find apparently unrelated sites giving near identical quotes, your confidence that you are looking at the original rises.
Re-quoting can occasionally be traced, especially where idiosyncratic changes are made, or a particularly creative interpretation is either cited or copied. Caution is essential, and the context of the commentary that surrounds the quote tells one a lot about how reliably representative of the original this exceprt is likely to be. Close reading is invaluable as a tool.
So when looking for some of the posts for Khaldun/Burke referred to in "Shell Games - Part 2", below, the originals had gone, as part of a deliberate move on the part of an involved actor: the host and target, SOE). But other posts survived which quoted and commented upon Burke. From those, I was able to recover enough to make useful citations. And, on the upside, it didn't require the idle Porphyrogenitus to set teams of civil servants to the task.
And here's something that a careful archaeological historian can avoid with discovered texts. Accessing cached google entries often actually destroys them if the site no longer exists. It's as if reading a book burns each page once you turn to the next. Google goes back, checks this old page that is apparently interesting, finds the original is no more, and flushes the contents.
Of course, there are no real internet palimpsests available. While actually getting hold of the server's disk array in a lab might theoretically make overwritten texts available, in reality this is not an option. Once that forum - World of Warcraft -style - is written over with newer posts, then the original is gone.