BloJJ

Good initiative. Maybe I should lobby for us doing the same for code at IDSIA... for paper-writing, however, I've been using Google Documents for a long time now - just dumping the LaTeX source in an online word processing document, and sharing it with collaborators. It works just as well as you would hope for, and has made various collaborations much easier. Once, four of us wrote a paper in one long night, all working simultaneously on the same document! I don't know if there's an option for making the document completely public though, but I think there is.

That's a pretty good idea, actually... although I prefer my good old emacs with all the syntax highlight goodies to the Google Docs WYSIWYG interface.

Yeah, I agree that it would be even better if GDocs highlighted the syntax as well, and possibly even did the compilation!

I don't see Google Docs compiling LaTeX sources in the near future but that would be COOL.

This is not exactly on-topic but I was curious what you thought about Amazon's EC2 service as a platform for evolutionary computing:

http://www.amazon.com/gp/browse.html?node=201590011

The reason I am interested in it is because of the ability to scale the computing resources in several dimensions, and because I am outside of the university system and have no good access to high-powered computing hardware for my research. Do you know of anyone in the EC community taking advantage of Amazon's services? Is it even suitable for this?

Hi, Dave. I also considered it. The concept is kinda interesting, but, unfortunately, it's for pay, so it's difficult to check it out on a whim. It would be ironic too, to use EC2 for EC (1).
However, if the issue is scaling computing power for those who don't have it available, using AJAX or BOINC is an alternative; all you need is a host with access to a database. Did you check our paper on the subject?

Yes; I saw your paper, and I also attended your talk at GECCO - you presented several interesting options for distributed computing. Another talk I attended mentioned using the multiple processors on high-end video cards for EC. So there are many options to pick from, it seems!

I have been checking out ECS, and from the purely theoretical point of view, it's no big deal: virtual machines available on demand. Plus, I'm afraid doing any experiment with would end up being expensive, even with the low machine-hour cost. And it does not fit well with the university paying processes.