Google Index to Go Real Time | Hacker News Reader

Google Index to Go Real Time

readwriteweb.com

33 pointsAnon8415 comments

Is anybody else having problems w/ GOOG's search results often being stale, several year old links? Particularly when searching for issues with software projects? I've noticed it more recently, and I'm not sure if it's because of the topics that I'm now searching for, or if it's a more general GOOG problem.

They really need to modify their algo such that freshness is more of a factor.

Slightly off-topic: Why are you using the ticker-symbol?

No reason really. Shorter to type = )

For some reason I got in the habit over IM, and wasn't thinking about it.

OK. I thought this was some commentary like talking about M$.

I've noticed that new things they've indexed perform extremely well for a day or two, then fall like a rock if they're not from an authoritative source.

After the couple days that fresh results are towards the top, Google seems to put more relevance on good incoming links. In order to have those links, the content has probably been there a while. That's probably why results are a bit stale.

The main problem I have about Google searches is that they usually lead to self-help forums instead of original authors. For example, a search for "Asus P6T BIOS" doesn't show Asus's own site until page two or three of the results. The result is similar when searching for some Linux kernel internal or gcc feature (like mudflaps). The problem with using links to calculate relevance is that people rarely link to the original author of a program, but forum systems and mailing list archives have tons of links in them to the same posts.

This should integrate really nicely into any of the MVC frameworks, just tell the hub whenever a model is created or updated that has its own page which you want to be indexed.

I understand you can generate and submit sitemaps, but pinging a hub after save makes a lot more sense to me than periodically generating a potentially huge sitemap and submitting the whole thing.

So if I get this right, it means that sites running this protocol will submit to Google right away? It's RSS in reverse?

So this means that instead of random search engines coming to you, the search engines have to rely on the sites knowing they exist. Brilliant way to keep out small-time competition.

EDIT: Nevermind, I got corrected.

I think it's both-and, not either-or.

the search engines have to rely on the sites knowing they exist

No, because sites wouldn't decide who to notify; they would push changes to any interested listener.

So Google would find them their usual way, then send them an "I'm interested" request, and the site would add them?

Yeah I guess that makes sense, nevermind.

"If Google can implement an Indexing by PuSH program, it would ask every website to implement the technology and declare which Hub they push to at the top of each document"

and they all choose google, making google, like, permanent supergod of the internet, instead of just a presently dominant overlord.

commenting because its too late to delete and this was a particularly stupid observation--all interested search engines will stumble on the feeds and subscribe.

Seems like a slightly different way of doing things. You can already submit an XML sitemap via Google web master tools, in a way it's accomplishing the same thing (telling them what to crawl). It's not real time but the premise is the same.

Edit: Also, this would definitely lead to more use of PubSubHubbub which I think Google is pushing us towards

This is big.

Reply on news.ycombinator.com