[geeklog-devel] About the Spam-X GSoC project

[Dirk Haun]

(BCC'd to all those who applied for the Spam-X project)

Some students have been asking me about the Spam-X project[1] and why it
wasn't selected. There was also some interest in working on it outside
of GSoC.

First of all, let me be blunt: The Spam-X project was our easiest
project idea for the Summer of Code. So I was really disappointed with
the overall quality of the proposals we got for this project.

Specifically:

1) Only one person noticed that Akismet and TypePad use the same
protocol and that therefore these two modules wouldn't require a lot of
work (since there already is an older Akismet module)

2) Hardly anyone thought enough about testing. Even after asking every
applicant to expand on that topic, only one proposal contained some
decent information.

Most people seem to have missed this paragraph in the project description:

>We would like to see a short evaluation of these services as one result
>of this project. For a proper comparison, the modules should probably be
>installed in parallel but not be used to actually delete spam. So a sort
>of evaluation mode could be introduced (and also added to the existing
>SLV module).

I've tried to hint that we could install the modules for evaluation on
geeklog.net and other sites that get a lot of spam. That would allow for
a good real-life comparison of the services.

3) Nobody realized that the Mollom module could be a problem (i.e. may
require more work) since it has an "unsure" categorization in which case
it wants to display a CAPTCHA.

4) The plans for SWOT weren't much better. Only one proposal made me
think that the student "got" what SWOT is about, the others were mostly
just repeating what was in the project description.

This being our easiest project, I was really hoping for someone to come
up with a nice twist or an additional idea to set themselves apart from
the other proposals. No such luck, I'm afraid. And since the few good
bits came from different proposals, we decided to skip this project entirely.

</rant>

Having said that: We're still interested in getting these modules
implemented. So I was thinking about throwing some of our bounty money
at it. Two or three people have already expressed interest in working on
modules (before I mentioned money), which brings me to the question how
we should handle this. Any suggestions?

I was initially thinking of a fixed sum for all the modules from the
first half of the project (i.e. without SWOT). But now that several
people have expressed interest it would probably make sense to split it
into smaller bounties per module.

Bounties are a nice idea. It only gets complicated when multiple parties
are interested in the same project ...

bye, Dirk

[1] http://wiki.geeklog.net/index.php/SoC_more_Spam-X_modules


-- 
http://www.geeklog.net/
http://geeklog.info/