We have discussed extensively the topic of CAPTCHA techniques for combating spam. The last round of discussions came after the fact that we featured a project related to captchas in our GSoC and OPW project ideas pages.
See Bug 62960 - Prototype CAPTCHA optimized for multilingual and mobile https://bugzilla.wikimedia.org/show_bug.cgi?id=62960
I think we can make this a good GSoC project idea, basically by
* acknowledging that we have a problem with our deployed captchas, and the investigation of alternatives is welcomed
* limiting the scope to mobile + multilingual, two key areas where our current implementation is clearly lacking.
* limiting the expectation on efficiency to anything better than the current implementation, not to a perfect system
* limiting the expectation on deliveries to a prototype, not to a full solution to be deployed in five months from now
Pau Giner and Emufarmers are willing to co-mentor, but they lack the experience required in programming. If the Language and Mobile teams are interested in this prototype, could they pull/pool the mentoring time required for this project? Is there anybody else willing to help?
In my non-expert opinion, the fact that we are aiming to have a prototype reduces a bit the demand for code perfection. Still, programming experience is needed to evaluate properly our current candidates, and also to assess whether the plans drafted during the project are technically appropriate.
The current proposals, sorted by date:
https://www.mediawiki.org/wiki/User:AalekhN/GSoC_proposal_2014 by Aalekh Nigam
https://www.mediawiki.org/wiki/CAPTCHA/Image_completion_captchas by Thanuditha Ruchiranga Wickramasinghe
https://en.wikipedia.org/wiki/User:Shashank2016/Multilingual,_usable_and_eff... by Shashank Jaiswal
Even if it is tempting to evaluate their CAPTCHA methods proposed first, in fact that doesn't matter as much as evaluating the skills, experience and collaboration style of the candidates themselves.
This is why we need now your help, even as no-mentors, proposing technical tasks to the candidates, and helping evaluating them.