![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
At 12:08 PM 4/24/98 +0900, Martin J. Duerst wrote: >If you define a directory system as a system that gives multiple hits, >you are definitely right. But the technical view that "it doesn't always >work, so it's useless" is completely wrong. This is an user interface My definition was in terms of taking "search" specifications, rather than "lookup" specifications. The simplest distinction is whether you are permitted to do partial specifications for a key. For directories, you are. For mapping services like the DNS, you are not. Hence this is much different from a user interface issue; it pertains to the definition of the underlying functional engine. >Also, it's important to understand that currently, doing a search >in a yellowpage-like search engine takes significantly more time >than guessing, including the time it takes to get the search page, The fact that searching takes time is the reason we do not want to postulate a real directory service as a replacement for the DNS, in the middle of every email and web reference. >So such a search always looses to an estimated 2-3 guesses (most >probably even 4-5 guesses), and it definitely always looses to So the basic rule is that a search engine is faster than guessing, if the guessing isn't trivial. I agree. (Quick. What is the url for Southwest Airlines?) >The ccTLDs are not a problem. I have "guessed" many domain names in >Switzerland and Japan, the countries I am familliar with. If you >wouldn't live in the US, I don't think you would see that as a big >problem. I'll let you know in a couple of months, but I've already been told by others in other countries that they find things like the web browser default guessing of .com to be quite maddening. >I agree that it's absolutely unclear how guessing might scale with >more TLDs, but one thing is rather clear: If it doesn't, and there I wasn't claiming that it was unclear how well guessing scaled. I am claiming that it doesn't scale. The only thing that is unclear is at what scaling point it fails. (By many measures, it already has.) d/ __________________________________________________________________________ Dave Crocker Brandenburg Consulting +1 408 246 8253 dcrocker at brandenburg.com 675 Spruce Drive (f) +1 408 273 6464 www.brandenburg.com Sunnyvale, CA 94086 USA
Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.