>
"instead of filling my memory with facts I store it with pointers to facts"[To push the analogy further] Unlike the managed memory model of the Encyclopedia, one of the problems with the internet as memory is that memory addresses are so frequently overwritten and the entire system is essentially non-redundant. This is to say that, unlike books web pages tend to change of disappear after publication breaking links and each web page is - for the vast majority of practical purposes - a unique individual manuscript rather than the product of a printing press.
This makes the internet as memory suffer from amnesia as pages are rewritten, overwritten, and discarded in a matter of months even at relatively stable websites and this tendency toward frequent revision encourages a lower level of investment in the depth of data provided on the web. As a case in point, I recently found a genealogical link from 2004 which pointed to the contents of a specific box in an archive in New Jersey that contained a photograph of my great great grandfather while he was serving in Company B of the 95th Ohio Volunteers during the American Civil War. Now the page returns 404, and the search function of the archive website does not hit on my ancestor's name at all - the detailed information which was accessible from the web did not survive the intervening updates of the website.
Of course my personal example hardly touches the scale at which the internet suffers from amnesia - even the deletion of Geocities is probably a tiny fraction of the information which the internet forgets at what feels like an ever increasing pace.
And given the ephemeral nature of the internet's indices - it doesn't even make sense to talk of remembering a Google or Bing search page given that it is specifically constructed for a unique time, place, and consumer. If the internet is memory, then the search engines are oft-demonic homuncului who provide access.