- From: Benjamin Franz <snowhare@netimages.com>
- Date: Wed, 16 Aug 1995 15:29:28 -0700 (PDT)
- To: http wg discussion <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
On Wed, 16 Aug 1995, Balint Nagy Endre wrote: > Hi all, > > I'm here again, and want to summarise discussion on cache disabling > techniques. > > Why people are against caching documents? > 1. I can only guess intentions of people, never seen personally, but I can > imagine only one cause: they want precise access statistics. Shopping basket applications and dynamic documents. *FAR* more important than precise stats. It never even occured to me that it might provide more precise access stats when I designed a site to explicitly defeat caching (<URL:http://www.psiloveyou.com/>, for those who care). The problem with caching is it makes dynamic documents (as in some shopping basket applications) hell. Between Mosiac's "Expire:? What's that?" behavior and cache corruption in Netscape, I couldn't run a reliable site without explicitly munging URLs to prevent caching in addition to expiring large sections of the site instantly. Believe me - I tried. It wasn't until later discussion by people trying to claim AOL has millions of accesses hiding behind single hits to their proxies that it occured to me that it was also a way to defeat the stats unfriendly behavior of caching browsers/proxies. Incidentally, based on the stats from the cache defeating site - AOL undercounting is no more than a factor of two or three. I would speculate that the massively broken nature of their browser, combined with a huge speed gap between 'Native AOL' graphics and inlined Web graphics turns AOLers off the WWW. Paying high per hour charges to download graphics from the web seems a no-brainer (to me anyway). > 2. they aren't paying for non-cached and otherwise cacheable requests. I am not at all sure how to parse this. At first I thought you meant that the provider charged by the hit, then I thought the opposite. Could you clarify what you mean? -- Benjamin Franz
Received on Wednesday, 16 August 1995 15:18:49 UTC