Google's netnews speed (was Re: pyBoards.com)

[Aahz]

In article <3F05FFFE.DB98BF44 at engcorp.com>,
Peter Hansen  <peter at engcorp.com> wrote:
>Aahz wrote:
>> In article <3F05C39D.3F65B89C at engcorp.com>,
>> Peter Hansen  <peter at engcorp.com> wrote:
>>>
>>>Google Groups likely takes its feed directly of a Usenet site, and Usenet
>>>in general suffers from large propagation delay.  It is that delay, caused
>>>by messages filtering slowly across the Usenet network, which leads to the
>>>"5 answers" problem, not Google Groups itself.
>> 
>> Actually, Google uses at least two or three feeds (including
>> stanford.edu, which has a super-competent news admin), and Usenet
>> propagation is actually little short of e-mail speeds these days,
>> depending on where you're located.
>
>That might be, but empirically I've seen posts that I've made take 
>hours to get there, and I've seen posts that were made to the mailing
>list (and copied to me directly) take hours to get there.  In both 
>cases, there is an "upwards" propagation delay towards whatever hosts
>Google is milking before stanford.edu or the others will see it...

Peter, you're thinking less clearly than usual.  ;-)  My point was that
the delay does *not* come from the time it takes to propagate to Google,
but from the delay required to index and publish the index.  Google's
indexing technology is nothing short of amazing, but it is poor for
realtime updating of the index.  In the past, they have generally
updated the index only twice per day -- and quite frankly, given *my*
experience with full-text indexing (I used to work for Verity), I
consider that to be *frequent* updating given how bloody fast Google's
index is.
-- 
Aahz (aahz at pythoncraft.com)           <*>         http://www.pythoncraft.com/

Usenet is not a democracy.  It is a weird cross between an anarchy and a
dictatorship.