Re: [Bug-gne]External Servers and Illegal/Extreme Content

[Tom Chance]

We ought ot be wary of what RMS when he warned us not
to re-invent the wheel. New unproven technologies,
like hosting a resource on a distributed network,
could in theory be very good for GNE but they could
also be fraught with problems because its never been
tried before. We don't want to end up with a system
that isn't working 5 years down the line when we have
5,000 articles to host and organise.

Servers should definitely be automated. The system
should be setup so that the only maintenence the
servers require would be the occasional defrag and
hardware work. But this could be done really easily by
simply using existing technologies like just using
Perl, a database, and apache. For each mirror you have
a main server (with the articles on), a bank of perl
interpreters (easy to expand should you need more
processing power, just add another machine!), and
either an index with that mirror, or a central index
on the main server held in GNU.

This is pretty much what has been discussed over and
over again, but people just talk theoretically without
thinking of how it could be done, saying "let the
rocket scientists do that bit". Its like managers
trying to make programmers work (ever read Dilbert?).
Rob Scott and I wrote up fully how you'd do the above
solution in server terms, though at the time we
thought it'd be good to store articles in a mySQL
database instead of XML with a db index. The rest of
the idea still holds though, and is properly worked
out, not just some theory in the air with no
substance. It's still up at:

http://www.state-embers.co.uk/alexandria/index.htm

Tom Chance



--- Christopher Mahan <address@hidden> wrote:
> I agree with Imran. The system could self-replicate,
> self-propagate, 
> self-index and self-link all by itself, once it is
> built the right way.
> 
> I am not sure that the current system of mirroring
> would do the trick, but I 
> don't know enough about that to judge.
> 
> What I envision would be a group of servers (in the
> same room or across the 
> world, or in orbit for that matter), that could
> communicate with each other, 
> update each other as to the most stable locations of
> any articles, and point 
> to each other.
> 
> If the system was done right, I could see that any
> article would be stored 
> on at least 4 separate servers, and that the system
> would check every 
> article every day (in the xml, no need for taking
> the server's cpu time to 
> run scripts to html-ise them), and if only 1, 2, or
> 3 copies were found, 
> then the system could issue the article to be held
> on an underutilized 
> server.
> 
> Also, servers with popular articles could offload
> some to other less-used 
> servers to alleviate traffic choke points.
> 
> It would almost be a self-balancing DNS server,
> where only GNE-authorized 
> servers communicate with each others.
> 
> This gets a lot closer to distributed computing, and
> since an article is on 
> a server one day but may not be on it the next, we
> might bypass some legal 
> issues, I think.
> 
> As far as the servers, I think that there would be a
> standard set of scripts 
> (php, java, asp, perl, tkl, python, wsh, ad nauseam)
> that a server owner 
> might be able to install in aspecific directory, and
> the scripts would then 
> self-authenticate with other GNE servers, and the
> GNE servers might start 
> using that server as one of their own (upload
> article copies, remove 
> articles, update translations, versions, etc...
> 
> If that server went offline as a result of whatever
> (war or the rumor 
> thereof, a cut cable, or a polica action), none of
> the articles would be 
> offline.
> 
> The Gnutella people must be listening...
> 
> Now, this system might be considered an expert
> system, semi-intelligent, and 
> I am sure we could get some good AI theorists to
> figure the nitty gritty and 
> heuristic rules.
> 
> This also means that once an article is submitted,
> it might be very hard to 
> eliminate from the system. Which is, in my opinion,
> exactly what we want.
> 
> Chris
> 
> 
> 
> >From: "Imran Ghory" <address@hidden>
> >Reply-To: address@hidden
> >To: address@hidden
> >Subject: Re: [Bug-gne]External Servers and
> Illegal/Extreme Content
> >Date: Sat, 17 Feb 2001 18:56:22 -0000
> >
> >On 17 Feb 2001, at 9:55, Tom Chance wrote:
> >
> > > > > It's the "knowingly" bit that worries me.
> > > > Presumably we know/have seen
> > > > > the content, that's why we're choosing the
> server
> > > > carefully,
> > > >
> > > > No, you misunderstand what I'm suggesting, the
> GNEP
> > > > won't
> > > > decide which material goes on which server,
> it'll
> > > > only decide which
> > > > material it will accept.
> > > >
> > > >  If an article is rejected by GNEP the author
> is
> > > > able to submit it to
> > > > one of the alternative servers, which are
> willing to
> > > > carry
> > > > controversial material.
> > > >
> > > > That way at the worst only the server which
> accepts
> > > > it can be held
> > > > responsible.
> > >
> > >
> > > That's a bit of an odd, and very messy way of
> doing
> > > it. It implies not only that GNE will be the
> central
> > > server and any mirrors or "contraversial"
> mirrors will
> > > be little offshoot projects. How, for starters,
> would
> > > you propose you link to all of this material?
> >
> >Have the IDs @ server_name and have DNS style
> tables available.
> >
> >Read the section entitled "An encyclopedia located
> everywhere" in the 
> >original
> >RMS announcement.
> >
> > > Expecting a new index for each server,
> >
> >No, the index will be part of a front end system.
> The encycopedia data 
> >would be
> >a the back end.
> >
> > >a new web site,
> > > or anything similar is ridiculous and will make
> > > navigating GNE as painfully laborious as the
> WWW.
> >
> >The user won't see any of this, they'll just have a
> front end which they 
> >can
> >access and the front end can search multiple
> servers and return data from
> >across a range of servers.
> >
> > > In your way GNE also still has responsibility
> for the
> > > article as much as if the server holding it was
> a GNE
> > > server, in my view anyway.
> >
> >How ?, leaglly it will be a server run by person
> xxx who isn't affiliated 
> >with the
> >FSF in any way except running software written by
> them.
> >(IANAL)
> >
> > > We have to have a central
> > > submission form, that will then put references
> into a
> > > central index,
> >
> >That would mean that one body would still control
> the material which isn't
> >consistent with the aims of the project.
> >
> > >so that the resource remains coherent
> > > and without broken links and missing materials.
> >
> >If we have mirrors, we won't loose material.
> >
> > >It
> > > will also then be much easier to mirror from
> server to
> > > server.
> >
> >Not really, once we've coded the system we can make
> it pretty much
> >automated.
> >
> >Imran
> >
> >_______________________________________________
> >Bug-gne mailing list
> >address@hidden
> >http://mail.gnu.org/mailman/listinfo/bug-gne
> 
>
_________________________________________________________________
> Get your FREE download of MSN Explorer at
> http://explorer.msn.com
> 
> 
> _______________________________________________
> Bug-gne mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/bug-gne


__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/