__/ [www.1-script.com] on Tuesday 11 October 2005 15:20 \__
> Roy Schestowitz wrote:
>> Does everyone realise that PageRank (i.e. backlinks principle /
>> was conceived almost a decade ago? There must be a better way, a new
>> paradigm altogether (and I am not referring to my last suggestion in
>> particular). I remember people coming to this newsgroup proposing
>> improvements to PageRank (Amir?), aggregating search engines or using
>> bandwidth to crawl even more pages (e.g. Yahoo bragging about having
>> crawled more pages... so what!?!?!).
>> Microsoft are already suffering due to lack of innovation. Their
>> O/S still sticks because of questionable practices, but that's a topic
>> another day. Either way, Google have messed about and revolved around
>> same key principle endlessly, almost to death. A few lines of python
>> have mushroomed to become millions of opaque LOC's. If the key idea is
>> inherently flawed, no improvement will save it in the long term. It
>> end up like Windows Vista, which has just been re-build from scratch as
>> the code (accumulated over the decades) was fundamentally rotten.
>> You can put a pig in a dress, take it out to dinner, but it's still a
>> pig in
>> a dress, not a girlfriend. Is PageRank merely a pig in a kingdom where
>> 'pigotry' is seen as beautiful (hence bigotry)? Is Google still a naked
>> king < http://tinyurl.com/94cpc >?
> Hi Roy,
> That pig in a dress vs. a girlfriend tirade somehow did not vibe right ;-)
> Gotta use less obvious analogies in a shady business of search engines ...
I was in O/S mode when I wrote this. Windows XP is Windows 95 in a dress in
my humble opinion.
> Anyways, getting back on topic: who said the original 9-years old PageRank
> still dominates? It has been improved upon many times already.
Yes, but it is fundamentally the same. The best you can do is put more
layers on top and wind up like Longhorn, which was a trainwreck.
> indication is the fact that Google still delivers results almost as
> relevant as they could be given the tiny amount of original data you send
> them in your two or three word search query. I already cannot remember a
> time my queries failed to deliver at least workable starting point of a
> research if not the exact answer right there, sometimes even without
> opening the links!
Don't forget hand-tweaking! *wink*
> If you think the algo needs lots of improvement, try to
> recall AltaVista circa 1997-98. What a joke those search results were!
I was actually going to mention that in my blog tomorrow (in drafts now). I
should also be able to compile and upload that proposal draft for knowledge
engines. Owing to Davemon's advice, I even added a Wiki to http://iuron.com
Roy S. Schestowitz | Ballmer O/S - so furious it may crash
http://Schestowitz.com | SuSE Linux | PGP-Key: 74572E8E
3:25pm up 47 days 3:39, 3 users, load average: 0.21, 0.26, 0.26