> The thing is - if there's anything even slightly centralised about it - how is
> one going to get the money to fund it?
>
> I mean the bandwidth charges'd be astronomical.
We could set up a Free (in the GNU sense of "Libre") search engine not
centralized.. in the way of software like napster.. we have to think and
find a solution.. we could make a worldwide network of bookmarks for
example.. of course powerful bookmarks.. well indexed bookmarks..
A posible way to accomplish this could be:
1.- We can have installed lets called it "GNU Smart Search" GSS for sort.
This software could reserve 10 - 100 Mbytes of our hd for its use
2.- each time we find an interesting site we bookmark it with
this software.
3.- This software could have implemented a spider which uses idle
network time to index our bookmarks and lurk the internet a little
bit. (the spider could only lurk our most local internet)
4.- When this software goes on-line it tells a central server it is
ready to recieve queries. (this can be dynamical and adaptative).
5.- lets say this means we have to ashare 5% of our bandwith for queries
6.- The central server recieve queries and forward them to each GSS client,
the client then respond to the original machine which launched the query
so the central server can be always as free as possible. (this strategy
can be improved. but we can think, cant we?)
7.- we can configure our GSS client to tell the server what contents we can
search better, geographical data about ourselves, how much bandwith we
can afford to share.. and so on.. so the GSS server can implement a more
clever forwading engine. (in fact as clever as we could be able to
implement)
Lets say:
Well 10 Mbytes * 10000 machines online = 100 Gbytes of searchable information
But lets be more optimistic:
100 Mbytes for GSS client * 100000 machines online = 10000 Gbytes!!! far more
than a lot of current search engines!!
and 1000 machines online means 1000 spiders working.. if we forget about
people bookmarking, their machines still can be working indexing a lot of
sites. This means a huge amount of information.. How much internet sites do
we really can find with current search-engines? 2%... 15% maybe? I doubt
we can reach much more. and what sites current search engines will present
you the most? comercial sites? informative sites? I think mostly comercial..
It is an idea, if somebody likes please tell me and we can start writting
this at sourceforge.. I am going to open this project right now.
Of course this GSS system could be improved to the limit of our imagination.
maybe it wouldn't be the most fast search engine but we could look for
strategies to make it reasonable fast. what I have described here is
only a first scheme.. and as disk space and bandwith became cheaper we can
have the most monumental resource to index the whole internet.
We can ask some LUG's to run servers.. many of them have free servers for the
GNU/Linux community.. (at least this is the case in Spain). The servers could
implement better strategies to load balancing and to eliminate duplicates.
Their only and main task should be to fordward queries and very few data,
which can be compressed so we safe as much bandwith as possible at this end.
This could be a great relief.. since the fact is that everything in the
internet is being monopolized, that all the "old" enterprises at the end
belong to very large ones.. and that the natural conclusion is that we will
finish with an internet whose main resources will be owned by only a couple
of huge companies.. the lack of freedom could be horrible then..
And be aware about microsoft .NET strategy...
As smaller companies are being swallow for large ones at the end we will
finish without room for freedom. and we only will see the internet that large
monopolies want us to see. As you very well point out to afford the
resources needed to index a reasonable amount of internet sites is impossible
for smaller companies.
Regards
Roberto
------------------------------------------------------------------------
Roberto Diaz <[EMAIL PROTECTED]>
http://vivaldi.ddts.net
Powered by ddt dynamic DNS
Powered by GNU running on a Linux kernel.
Powered by Debian (The real wonder)
Concerto Grosso Op. 3/8 A minor
Antonio Vivaldi (so... do you need beautiful words?)
------------------------------------------------------------------------
