[wp-hackers] Page searching examples [was: 2.4 planning discussion]

[Andy Skelton]

On 10/3/07, Knut-Olav Hoven <hovenko at linpro.no> wrote:
> > WordPress, open source software that typically runs on shared hosting,
> > should be able to produce search results on par with the secret
> > algorithms of the world leader in search, a company with market
> > capitalization of 182 billion dollars?
>
> Yeah, but WordPress don't have to search 182 billion websites either... just
> one. And we do get access to the raw data from the database, Google does not
> (hopefully). ;)

Google's index of billions of sites is part of what makes Google's
results so strong, even when the results are limited by "site:
example.com". Also, Google indexes the entire page, whereas WordPress
only knows about the contents of the post. The output of plugins and
widgets, such as UTW or Related Posts, can help Google determine
relevance. WordPress has no facility for searching pages by their
final output.

> > It has been agreed that if WordPress had a relevence-based search
> > engine built in, we wouldn't have any trouble integrating pages into
> > the search results. That would be a nice thing to have but it's really
> > not a realistic goal at this point.
>
> That's probably because the basic WordPress philosophy has been to reject
> patches larger than 2 lines, telling the contributers to make plugins
> instead. (I'm pushing it to the edge, but it's still true)

No, you're blaming something that is not at fault. None of the
solutions to date have fit the bill.

> > There are fine ways to integrate Google searches into a blog's
> > template. I think that's more our speed. Competing with Google Search
> > just isn't.
>
> Not a good idea. Core functionality should consist of framework-stuff, not
> tight integration with other systems that may (not likely, but may) change
> their API's.

It may not be for core but it's something that themes can do. I've
done it. It works very nicely with Adsense, too.

I still stand behind my assertion that we lack a good
relevance-ordered search system.

Andy