Re: [Demexp-dev] TOP invitation to programmers.

[Frederic Lehobey]

Hi,

On Mon, Oct 02, 2006 at 05:34:40PM +0200, ketty . wrote:
> I have been made aware of the TOP Politics -group:
> http://top.xwiki.com/xwiki/bin/view/Main/
> And a recent post to unite programmers:
> http://groups.google.com/group/top-politics/browse_frm/thread/505c11b28d85cbf7/c30b4ae282e4fbde#c30b4ae282e4fbde

Thanks for these links. First time I hear about leparlement.org.

Now quick answers below (maybe this discussion should be moved on
demexp-en as it is not strictly related to software developpement).

Quick answers below.

> With this response about DEMEXP:
> I know that project rather well, I'm subscribed to their mailing list
> and participated quickly to it before starting "parlement".
> 
> My troubles with it:
> 
> - voting method is "Condorcet", which is often considered slightly too
>  complex for most people

It is correct the current voting method is Condorcet (and this choice
is not random) but other voting methods might be implemented
too. Patches are welcome.  :-)

> - special powers to some administrators whose role is to organise polls

Not really. The only special power is on classification (needed for
the future delegation). But everyone can raise questions and provide
answers.

> - technology is a very good one, but sadly it is not used much in the
>  world and will have troubles finding programmers

It somewhat true. But works like Augustin's one might open us to the
much larger world of PHP / Drupal programmers. Actually, the server
has a public API so every kind of client might connect.

> - no possibility to use the whole system as a forum where every
>  question/answer is just one more post

Yes. The system is intended on purpose (let's do one thing but do it
well) as a voting tool, not as the place where the debate takes
place. The debate might occur on many other already existing places
(Drupal, forum, wikis, and so on...).

> - few considerations for security

... for the moment! Because precisely we do not want to deal with this
in an approximative way (which is always bad for security). The idea
is as follows: let's work to build a tool that is functionally
complete (it is not yet the case with respect to delegation and
localisation) and THEN specify the appropriate security algorithms for
our needs. We have already reviewed the crypto litterature, but it
seems our problems (constantly open voting with delegation) has not
yet been addressed much. If you look back enough in the mail archive
(before our wiki existed) you will find we have even devised some
(naive) cryptographic algorithms. But we have failed so far to have
them studied by professionnal cryptographers (it is a tough
problem... and nobody is yet paid for it).

> But, they have a group of intelligent people involved in a cool social
> idea/ideal.

Thanks.  :-)

> 1. Voting system.
> I think not one system would be perfect for all questions. My vision is to
> be able to set the voting system used on a per question basis. At least that
> is my vision of the technology, in practice you might wish to restrict the
> posibilities of users, or you might not.
> This would be feasonable with demexp, no?

Currently, no. But everything is open. It is free software. This need
has already been expressed. Please provide the patches.  :-)

> 2. Special powers.
> I am all for special grouping of users. For example you might want to
> differate between users above and below 18 years of age (or another
> arbitrary age). And you might wish to group users into geographical areas as
> is common in politics today. The techology should make possible such uses.
> How it is used in practice should be decided by users.

Same as above. Beyond the technical points, philosophically, I
somewhat disagree with this point of view. But it does not prevent it
from being implemented in the software.

> 3. Technology.
> I very much like ocaml and similar languages, and i would not want demexp in
> java. :)

The clients might be coded in whatever language.

> 4. Security.
> This is important indeed. But not only the demexp-server is responsible for
> the security. We have the physical server etc to worry about too. The
> security issue is not something that can be finished from the start, but
> something that will be an ongoing issue.

I completely agree. Consider it as a Graal (after several years and
many thousands of users): a distributed server. But it will be for
demexp 4.0 or above...  :-)

Experimentally,
Frédéric Lehobey