headers
David MENTRE | 10 Aug 2003 18:08
Favicon

Start of code for the server

Hello,

I've continued today to code the demexp code related to the server (srv/
subdirectory) and the make infrastructure.

Right now, the code does *nothing*. It's just a bunch of data structures
and functions. I would have prefered not to release it yet, but as we
are supposed to play the free software game (release early, release
often).

It would be nice if people of this list could:

 1. get source code
       cvs -d:pserver:anoncvs <at> subversions.gnu.org:/cvsroot/demexp login

         (no password at login prompt)

       cvs -z3 -d:pserver:anoncvs <at> subversions.gnu.org:/cvsroot/demexp co demexp

    For details on CVS: http://savannah.nongnu.org/cvs/?group=demexp

 2. try to compile it (please read the README)

 3. report on this list of any particular failure, issue, etc. or
    success 

The code should compile right out of the box on any debian box (with
necessary packages given in readme).

Thanks,
(Continue reading)

Permalink | Reply | 1 comment |
headers
David MENTRE | 10 Aug 2003 18:18
Favicon

Some questions on Condorcet voting

Hello,

I'm about to try to code the voting part of the server, using the
condorcet voting. 

I've started to look at the different links provided on the Expérience
Démocratique web pages (http://exp.dem.free.fr/).

Right now, the most precise description I have found until now is:
 http://www.electionmethods.org/CondorcetEx.htm (en)
 http://electionmethods.org/fr/CondorcetEx.htm (fr)

I have not read all the details, but it appears to exist two algorithms
to solve Cyclic Ambiguities : Basic Condorcet and Schwartz Sequential
Dropping (SSD).

Quoting the web page: "Condorcet is simpler, but SSD is the state of the
art.". Should I implement SSD?

Yours,
d.
--

-- 
 David Mentré <dmentre <at> linux-france.org>
   http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
 GnuPG key fingerprint: A7CD 7357 3EC4 1163 745B  7FD3 FB3E AD7C 2A18 BE9E
Permalink | Reply | 2 comments |
headers
David MENTRE | 10 Aug 2003 18:33
Favicon

Re: Start of code for the server

David MENTRE <dmentre <at> linux-france.org> writes:

> I've continued today to code the demexp code related to the server (srv/
> subdirectory) and the make infrastructure.

A forgotten point: Isa, I haven't yet integrated your code. I glanced
through it and I have small remarks. But most importantly, I would
prefer to integrate a modularized code, not the big file which exists
right now.

Isa, can you also take a look at the server code (.ml.nw) files? I think
it would be nice to put your interface in the same literate programming
form (i.e. .nw files). What do you think of it? If you agree, I'll add
minimal noweb tags. You'll have time then to add necessary comments if
needed.

Yours,
d.
--

-- 
 David Mentré <dmentre <at> linux-france.org>
   http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
 GnuPG key fingerprint: A7CD 7357 3EC4 1163 745B  7FD3 FB3E AD7C 2A18 BE9E
Permalink | Reply | 0 comments |
headers
David MENTRE | 10 Aug 2003 18:22
Picon

Re: Some questions on Condorcet voting

David MENTRE <dmentre <at> linux-france.org> writes:

> I have not read all the details, but it appears to exist two algorithms
> to solve Cyclic Ambiguities : Basic Condorcet and Schwartz Sequential
> Dropping (SSD).

And there are two variants for SSD: "plain" version and the "cloneproof"
version.

> Quoting the web page: "Condorcet is simpler, but SSD is the state of the
> art.". Should I implement SSD?

Should I implement the cloneproof version?

Yours,
d.
--

-- 
 david.mentre <at> wanadoo.fr
Permalink | Reply | 1 comment |
headers
David MENTRE | 10 Aug 2003 20:42
Favicon

Complete computation of a condorcet voting

Hello,

For testing purpose, I need a sample of condorcet voting (complete, with
intermediate matrices computation, cycles resolution, etc.).

I can do one myself but if you already have such a computation, it would
be nice.

Yours,
d.
--

-- 
 David Mentré <dmentre <at> linux-france.org>
   http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
 GnuPG key fingerprint: A7CD 7357 3EC4 1163 745B  7FD3 FB3E AD7C 2A18 BE9E
Permalink | Reply | 0 comments |
headers
David MENTRE | 11 Aug 2003 19:07
Favicon

Condorcet voting: explanation needed

Hello once again,

Last spam :) for the code, I plan to include an explanation of Cordorcet
voting. Right now, I plan to copy the explanation at
http://www.electionmethods.org/CondorcetEx.htm (with their
authorization), but if you have a better explanation (a more
mathematical one for example) it would be nice.

Yours,
d.
--

-- 
 David Mentré <dmentre <at> linux-france.org>
   http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
 GnuPG key fingerprint: A7CD 7357 3EC4 1163 745B  7FD3 FB3E AD7C 2A18 BE9E
Permalink | Reply | 0 comments |
headers
David MENTRE | 16 Aug 2003 17:49
Favicon

Re: Re: Some questions on Condorcet voting

David MENTRE <david.mentre <at> wanadoo.fr> writes:

>> I have not read all the details, but it appears to exist two algorithms
>> to solve Cyclic Ambiguities : Basic Condorcet and Schwartz Sequential
>> Dropping (SSD).
>
> And there are two variants for SSD: "plain" version and the "cloneproof"
> version.

Ok, I have finally implemented both of them (see autotest-srv.ml.nw for
code example). 

I have followed rules in the Debian Constitution
(http://www.debian.org/devel/constitution) to implement SSD. The
ElectionMethods.org web site is not very clear. 

I have implemented some hairy test cases (cyclic 3 parts tie) and it
seems to work. However more checks are needed.

The scoring of defeats could be improved, following once again
definitions in the Debian Constitution.

I have tried to document the code in a literate programming way. So I
would consider as a bug if you cannot understand what I have done.

Yours,
d.
--

-- 
 David Mentré <dmentre <at> linux-france.org>
   http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
(Continue reading)

Permalink | Reply | 0 comments |
headers
David MENTRE | 18 Aug 2003 23:16
Favicon

No much code in the coming week (and some homework)

Hi,

I won't be able to code this week and next week-end, so don't expect new
code neither response from me until the middle/end of next week.

But if you have some spare time, I can give you some homework :)

 - think about security issues. More important than precise security
   mechanisms, we need to clearly define the security objectives we
   want. Think about confidentiality of vote, counting and (independant)
   re-counting of votes, authentication of participants, authentication
   of source code and compiled binaries, authentication of server from
   clients and clients from server (for both general interaction and
   voting), allowing participants to change there vote albeit let the
   vote unkown for the voting system, etc. We need to write the security
   section of the specifications.

   Of course, having security mechanisms and protocols (and even
   libraries) to do that would be a plus. :)

 - write an english version of the web site (or at least a small
   description of the project). Would be useful to point to from the
   development web site (http://www.nongnu.org/demexp/)

 - find or write english description of condorcet voting, actual or
   expected performances, needed metrics to evaluate performances,
   etc. A reflexion and examples of pairwise matrices reflecting bad
   situations (cyclic ties and the like) would also be much appreciated 

   BTW, I insist on the fact that the actual is documented (or it least
(Continue reading)

Permalink | Reply | 0 comments |
headers
David MENTRE | 18 Aug 2003 23:25
Favicon

A question on computation of delegates' votes

Hello,

I plan to use the following algorithm to compute delegates votes:

 - compute pairwise matrix from a delegate vote (i.e. his ordered list
   of responses)

 - multiply this matrix by an integer scalar of which value equals the
   number of delegators having chosen the delegate

 - add this matrix to the sum matrix of previous and next votes

 - apply usual algorithms to determine the winning response

Does this algorithm seems correct to you?

Yours,
d.
--

-- 
 David Mentré <dmentre <at> linux-france.org>
   http://www.linux-france.org/~dmentre/david-mentre-public-key.asc
 GnuPG key fingerprint: A7CD 7357 3EC4 1163 745B  7FD3 FB3E AD7C 2A18 BE9E
Permalink | Reply | 0 comments |

Gmane