CSE HTML Validator

I've noticed that search is becoming all but useless in many cases. I just get huge loads of things that are useless to me.

Others have noticed this. Like Jeff Attwood of StackOverflow, who may have started the recent change at Google. That's the change that has seen "syndication sites" (that grab somebody else's content (legally) add adverts and do nothing else) demoted by Google. That will hit Google's advertising revenue, so it's a sign that the pressure was high.

Now I see an announcement. You will soon be able to permanently ban a web site from your own search results (from Google). I've wanted this for a very long time so I think it's a good move. See http://bit.ly/dNyPm0.

It's not a pure blessing. You need to be signed in to use it. If you don't want somebody else to be tracking your web usage (and cashing in on that data), it's not a solution.

I take it as a sign that Search Engine companies have sensed that users are angry with the rubbish they get and are ready for a change.

I've come to the conclusion that a search engine that I pay for, that I control to some degree, and where I have meaningful control over the algorithm is the answer. This move goes some way toward that, but I suspect won't make me or many others much happier.

At the moment they're not using the switch off signal in rankings. Long term I imagine they will. (There are much better ways to handle that than demoting everybody's ranking!! but I suspect they won't be that smart.)

Implications:
1) Maybe we will get increasing control. It's a devil's bargain if somebody else get's your details as part of the deal though.
2) If your site annoys a lot of users, you might be in trouble. It might even happen soon. With a feed back signal at last, black SEO will no longer be able to promote complete garbage as it scams the search engines. There will be payback. Hopefully we might even see a resurrection of worthwhile content.
3) If we're lucky we'll see an "arms race" from the engines, giving you more control. A slow demise of "one size fits all" and "no feedback" would be good.

I also hope it is a shift toward better search engine results.

I am often amazed at the amount of garbage that can come up on the first page of search results. I'd normally expect the garbage to be several pages away from the first page.

I do like the idea of having users rank sites. You could then use that data to fine tune results by eliminating (or lowering the position of) low ranked sites, but then you might have to rank the users themselves as to how trustworthy they are, and there's the opportunity for malware to take over one's computer and secretly give high rankings to junk sites using many people's accounts.

Not sure about the paid search engine idea. I am not sure that would "fly", but who knows, Google could try it and see what happens. Website rankings from paid users might be trustworthy enough to use in computing whether a site is "junk" or not.

The ranking of users is, in my view, absolutely critical.

There's a self sustaining factor here. Just look at web content that describes the web, for example. So many people have been swallowing untruths for 10 years or so that their view on things are so divorced from reality as to be dangerous. Ideally you need to approve people individually and escape from the often wrong belief that "crowds have wisdom".

The issue I see with paying is that somebody pays anyway. At present it's advertisers who feed the coffers and plaster web sites with largely irrelevant content. People seem happy with a few percent clicking on an advert. Just think about that. You get 3% click through. That means 97% of visitors are indifferent or actively hate the adverts. Not particularly smart!

Adverts paying for it hasn't worked very well. It's helped to flood the web with rubbish. It's part of a lack of personal control.

To get control you can build it yourself or pay somebody else to. (I'm sure there's other models too.) The models don't need everybody to adopt them. Just enough to make them self sustaining.

We'll see.

I've come to the conclusion that a search engine that I pay for, that I control to some degree, and where I have meaningful control over the algorithm is the answer.

The only way you're likely to get that kind of control is via engines that support SPARQL. I like http://sindice.com/ + http://sig.ma/

more: http://www.w3.org/wiki/TaskForces/Commu ... rchEngines

Start running your own virtuoso endpoint.

In the past I've concluded that RDF and the Semantic web are a good idea (though a bit underpowered) but are lightly used.

When I last looked for RDF and microformats in previously identified "worthwhile content" I, unfortunately, didn't find any.

It's very domain specific I know, but in what areas do you find that the "best of breed" content is accessible using Sparql?

I'd be very interested in other people's experience.

(One of the reason's I really like XHTML is that it opens the potential to make the web better searchable.)