The problem potentially with Jimbo’s desired scheme is that if the results are ‘controlled’ by anyone, ala Wikipedia, we’ll end up with situations where people will game Wikia search the same way that trolls and idiots game the most crucial Wikipedia articles, on real people, politics, and pivotal social matters like global warming, abortion, and the Iraq War. If it turns into another scenario like Wikipedia, with cliques of Jimbo’s “trusted people”, it’ll be a farce.

DMOZ.org is a better example of how to do what he’s after, for a starting point.

The discussion on this post seems to be a debate on whether or not Google is, in fact, ‘broken.’ Broken’s probably a bad word to use, since Google is obviously not in a useless state, but there’s many things people don’t like about it, particularly the number of ways of raising Pagerank without raising actual popularity or usefulness of your site. I hardly think this makes Google ‘broken’, but it’s certainly flawed. Of course, everything’s flawed, and there’s always room for improvement. The question for any potential Google competition is very obvious, but much harder to answer – how will you do it better?

Google was successful on a very simple, but important, idea – have an algorithm that has a way of determining a site’s value without just scanning for key words in meta tags or in the page itself. Initially this was simply link count, though of course that’s been refined over time. Still, the Pagerank system alone doesn’t work well in some cases – especially when you’re researching something controversial… the more supported position is of course, always going to get linked to more. In cases where it’s the WAY more supported position, the less popular side won’t even appear on the first page.

DMOZ was mentioned, and while that’s a great directory, very few things are going to appear there unless they’re already easily searched for, or well-known in their community – and that’s if you have GOOD editor on the topic. Assuming the topic in question has a good editor, this often makes it a great place to quickly find some relevant sites, and it DOES do a great job of filtering out the crap that gets good search results but has no real value, but it’s hard for a relatively unknown site to ever get known through something like DMOZ. It’s a great supplement to a search engine, but it really can’t replace one. It’s also based heavily on each category having (usually one) editor. Even assuming the editor is very professional and mostly unbiased, you run into another issue – what happens when he simply gets sick of updating his section? An entire category can go years without updates, because not only do people have to realize it’s not getting updated, but someone with an interest in fixing that problem – and enough knowledge to do it – needs to appear.

The wiki approach, in theory anyway, makes it much easier to keep stuff up to date, and the idea of a discussion page for all the search results allows you to quickly scan for dissenting opinions on the results – very good when you’re trying to fully understand all sides of an issue – especially when the discussion page may have a link you want that the search results don’t.

The problem is that for that approach to work, you need a good system of checks and balances, and you need VERY fast responses to vandalism (Wikipedia gets hit often enough, but a search engine as popular as Google would get hit CONSTANTLY.) Making page edits take a few days to actually commit would probably help (especially if you have section editors DMOZ to check periodically for crap.)

Ideally, I think you need a mixture of the wiki approach and the DMOZ approach – let anyone edit it, but make sure someone’s there to actually review the edits – ideally more than one someone. Also make sure that there’s a way to make sure all of these chief editors are actually still active (simply checking login frequency should work) and have some sort of way of removing a heavily biased section leader.
Of course, none of this works if new but relevant pages aren’t discovered… easy enough in a topic where there’s a strong community and the editors are probably active in said community (though this creates another issue… if the community itself is controversial, you need to ensure links to sites AGAINST it can appear too…), but much harder for a more academic topic, or simply a relatively obscure one… so you’re still going to need crawler bots finding stuff, and then people manually sifting through it.
This raises another problem though… manually sifting through bot results is not exactly interesting, and unlike wikipedia, this would be a for profit company – people might be more than happy to submit known links on a favorite topic, but VERY few people are going to do raw data processing for free. You’d need a pretty big staff to fully support the effort.

Search engines are one of those things where almost anyone can think of an improvement. Actually getting it implemented is another story, however. I do think it’s only a matter of time before a viable Google competitor appears. Wales has a chance, but he’s going to need more than just wiki mechanics, and it’s almost certainly going to take a lot of trial and error to get right.

As far as I can tell, the main argument by Jimmy Wales goes “Google is broken, so we’re making a search engine that works more like Wikipedia does”. My head hurts just thinking about that.

No, I’m not saying that Wikipedia is bad. I’m just saying that, in its own way, it has at least as many problems as Google does.