Free Software Activists Take On Google Search

alphadogg writes "Free software activists have released a peer-to-peer search engine to take on Google, Yahoo, Bing and others. The free, distributed search engine, YaCy, takes a new approach to search. Rather than using a central server, its search results come from a network of independent 'peers,' users who have downloaded the YaCy software. The aim is that no single entity gets to decide what gets listed, or in which order results appear. 'Most of what we do on the Internet involves search. It's the vital link between us and the information we're looking for. For such an essential function, we cannot rely on a few large companies and compromise our privacy in the process,' said Michael Christen, YaCy's project leader."

Re:Question

[CanHasDIY]

Will one client be able to view the queries of its peers?

If yes, how is that an improvement? If no, how does it work?

From TFA: [yacy.net]

It is fully decentralized, all users of the search engine network are equal, the network does not store user search requests and it is not possible for anyone to censor the content of the shared index.

However, that seems to be all the information there is on the process... doesn't quite assuage the ol' paranoia circuits, does it?

Cool, but what's in it for the peers?

[91degrees]

While these things can succeed on the backs of some philanthropic individuals, it's just human nature that to get a decent community, you need to benefit the supporters in some way.

Doesn't need to be any formal system. Free software, for example, seems to be based more on the honour system than anything else, but people do develop free software because there's something in it for them - software tailored to their needs. What is the incentive for being a search peer?

Comment removed

[account_deleted]

Comment removed based on user account deletion

No control over disk usage

[markdavis]

This whole concept seems quite fascinating/interesting. Ironically, two questions came to my mind immediately:

1) How much bandwidth does this take?
2) How much disk space does this take?

Neither question is answered on their FAQ ( http://www.yacy-websuche.de/wiki/index.php/En:FAQ [yacy-websuche.de] ), although they addressed the disk space issue thus: "Can I limit the size of the indexes on my hard-drive? For the moment no. Automatically limiting that size would mean having to delete stored indexes, which is not suitable. "

Yikes! I am not sure how many people will want to run a local YaCy client when there is no control over how much disk space it uses (or, apparently, bandwidth). It still has a lot of promise, though.

Got to get off my lazy butt...

[xTantrum]

...and start coding my ideas. First itunes, then fb and now p2p search. Just goes to show ideas are a dime a dozen its just who implements it first. Can't wait to see how this turns out though. P2P is really how the internet should be structured as much as possible.

Re:Well

[Anonymous Coward]

At least it actually is in the interest of search providers like Google, Yahoo and Microsoft to produce useful results in order to achieve / maintain a large userbase.
Not so much in the interest of somebody who simply sees a distributed search engine as his chance to drive fews to his blog / ad collection / malware site.

I'm not seeing why this should be tried.

[cshark]

Haven't we learned from gnutella, and the others, that this kind of thing just doesn't work? That it'll get overwhelmed by spam, hackers, you name it? I'll try it because I always try new p2p type stuff. But I'm really hoping they have a good security team.

Re:Well

[Rei]

The whole "portal only as an afterthought demo" seems to me a huge flaw as well. You think your average person is going to install this on their computer just so they can do web searches? Not-going-to-happen. People who want to run it, will. People who don't or don't know how, won't. They're the 99.99%. They need a portal. Clients should automatically be putting themselves in the portal-switching queue.

As for the capabilities, I just tried it out [yacy.net]. The results are *extremely* few and very poor. "Dog" gets five hits, for example. You'd almost think it was a joke. Hopefully this was a load problem or a problem due to a lack of scaling in the system thusfar, and not a design flaw.

At least their frontend doesn't seem designed with injection in mind. Start off a search with ' (such as 'Test) and watch what happens to the peer listed at the bottom of the page. I doubt that particular issue is exploitable, but if this a habit of one of their coders...

Re:Java...

[vadim_t]

Ugh, yeah. Another cool project is going to be held back by Java.

Way back, this happened with Freenet. I thought it was a cool idea, but the darn thing wasn't happy with all the 256MB I could give it. Even now, Java is still a considerable load on laptops with 4GB RAM.

I think that for best adoption they should have concentrated on making it small and light. If it can be run in say, 64MB RAM then you can install it anywhere. And it's quite likely that a good part of why Freenet was so horrible when I tried it, is because it made a lot of the machines it ran on swap like crazy.

Re:Cool, but what's in it for the peers?

[TheRaven64]

I sketched out a few designs for a decentralised search engine (but didn't implement them, so kudos to these guys for actually bothering), and one of the ideas I had was to allow nodes to return sponsored links (e.g. Amazon referrals). The client would display these for the top few nodes and track the reputations of individual peers. The more users who liked the search results that you returned, the more of them would see your sponsored links. If you came up with a ranking algorithm that did a better job than existing ones, then you'd get a bigger slice of the advertising space. It's essentially the same business model as Google, just on a smaller scale.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:Java...

[Lazy Jones]

cool project is going to be held back by Java.

You know, I'll take "cool projects held back by Java" any time over equally cool projects written in C that need to be patched 5 times a year for the next 10 years because of sloppy programming leading to arbitrary remote code execution vulnerabilities. Please, just let software written in C die with dignity, the language had its decades of glory before everything was accessible over the 'net ...