[liberationtech] DuckDuckGo vs Startpage [was: Help test Tor Browser]

[Axel Simon]

On 27/06/13 01:02, Mike Perry wrote:
> The Doctor:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On 06/24/2013 09:16 PM, Daniel Sieradski wrote:
>>> Has there ever been any effort to create an open source search
>>> engine that is entirely transparent in both its software and
>>> practices? (dmoz.org
>> doesn't count!)
>>
>> ...YaCY?
>>
>> http://yacy.de/
> 
> YaCY and other FOSS engines (in a sibling thread someone mentioned
> another that I already forgot) are also something that I will accept
> search plugins for the Omnibox, but their result quality, index depth,
> and crawl frequency are no match for either StartPage or DDG.
> 

There's also Seeks.
http://www.seeks-project.info

It's “An Open Decentralized Platform for Collaborative Search, Filtering and
content Curation”.

From what I understand, Seeks tries to do several things at once:
- Provide search results by aggregating them from different sources such as
Google, Bing and other seeks nodes.
To jumpstart the available results and achieve good quality, they decided the
best thing to do was just to grab good results where they were, so by default
nodes will ask Google for results.
But more backends can and are being developed.
- Keep things decentralised. The nodes share results with each other, this is
the basis for the general Seeks network's crawler, if I understand correctly.
- Enable users on a node to express their like or dislike for the result of a
search.
This means over time the node learns and will curate results for a given user.
Dislikes are kept to a node while positive search results are shared between
nodes to build up the general search engine's results.

In terms of pure privacy, this does sound like only half a solution : if you run
the node on your laptop, seeks is just querying Google for you really.

But one can share a node with more people or even use a public node. There are
several listed at:
http://seeks-project.info/wiki/index.php/List_of_Web_Seeks_nodes

In this case, a public seeks node acts like a proxy for new search requests. And
for requests that have already been asked, it will give answers on its own
without querying external engines.
There are also instruction on how to anonymize a Seeks node on the wiki.

The project is really interesting, even if a little less active today than it
was 18 months ago.
But it works and you run it on your server.
You could probably set it up as a hidden tor service too.

I've cc'd Beniz, who runs the project, he probably has far smarter things to say
on the question. :)

Cheers

axel


-- 
Axel Simon

----------
mail/jabber/gtalk: axelsimon at axelsimon.net
twitter / identi.ca: @AxelSimon