[Freedombox-discuss] Leaving the (proprietary) cloud - my roadmap for FB

[Amr Gharbeia]

On Sat, 2010-10-09 at 01:25 +0200, Jonas Smedegaard wrote:

> True, the code is free and open, but it uses other search engines as 
> basis - the P2P is a regrouping of search results from central engines.

Seeks is still in phase one of its development, and currently does
little more than what you described. By the time, it reaches its fourth
and final phase of development, it promises a fully-distributed model.
The fourth phase seems to make very good use of a project like
FreedomBox.

>From Seeks wiki: http://seeks-project.info/wiki/index.php/Roadmap

"There are four main envisioned milestones for Seeks:

"Codename SOLO 
"Seeks provides a websearch proxy and a local metasearch engine. The
meta engine aggregates websearch results from several mainstream search
engines. The aggregation is based on consensus among the search engines.
At this point Seeks provides a plugin system that allows to easily build
sets of URL interceptors and webpage modifiers. Applications rank from
ad-blocking to dynamic webpage enhancements or aggregation. The
metasearch engine comes in the form of such a plugin.


"Codename HIPPY 
"Seeks provides the basic collaborative functionality on top of existing
search engines by connecting people that search the web with similar
queries. The similarity-based pattern-matching technique used is known
as locality sensitive hashing (LSH) and we distribute it among the
peers. Existing search engines results are re-sorted and enhanced with
the information fetched from the peer-to-peer network, such as ratings,
other relevant results, related queries of interest, and direct chat
opportunity is provided to users performing the same, or approached
queries. Enhanced with a clean and modern interface within the browser,
we believe these core capabilities should be of interest to number of
people, and we wish they draw in users, testers and developers in
sufficient numbers.


"Codename DANDELION 
"The next step is key to the Seeks project, and introduces what we
believe is one of its most beneficial features. Seeks will propose a
self-publishing mechanism accessible to anybody with a browser and an
Internet connection. Instead of relying on a search engine for linking
keywords to web contents, Seeks will let users register any URL of their
choice using their own set of keywords. Users querying the peer-to-peer
network (DHT) will thus be recommended web content without using any
existing search engine. This operation would be the combination of a DHT
lookup plus a push on the peers of interest. The operation should not
cost more than a file lookup on your favorite peer-to-peer application,
(i.e. ok-fast). Users that would register their personal web content or
that of others under unsatisfactory keywords or queries would see their
keyword associations naturally rated down by other users, in a move that
we believe should lead to a better match of keywords and queries to the
true content of a web page.


"Codename PLUM 
"The final step proposes a decentralized web information index to
gradually re-capture public information currently stored in private
corporate facilities. We propose to implement small software extensions
to common Web servers, such as Apache. These extensions would allow web
servers to locally index their webpages and share the indexes with other
web servers and users, in a decentralized manner, on the Seeks network.
The consequence is that over time, Seeks would evolve a parallel search
engine, processing queries against a decentralized database of
information rated by the community for the community."

-- 
Amr Gharbeia
http://gharbeia.net