Several news sources have started reporting on a 2007 research project by a group of MIT students who found they could accurately predict Facebook users' sexual preferences based on the people they were friends with. The project, referred to as "Gaydar," sampled data from 1,600 men (only 33 of whom were out as gay on Facebook) to create an algorithm that supposedly predicts whether a user is gay or not. However, the research methodology behind this unpublished study seems a bit dubious to me.

The initial test of Gaydar correctly predicted that 10 of the researchers' friends - who weren't "out" on Facebook - were gay. That's a pretty decent success rate, but a tiny, tiny sample size. Only 33 gay men out 1600 total can't possibly be reflective of the entire population. According to Wikipedia, 4% of voters in the last US election self-identified as gay, lesbian, or bisexual. Assuming that those numbers are pretty conservative, only 2% of the people in Gaydar's initial sample were gay. I'm not convinced it's really time to start panicking that you could be inadvertently outed based on your Facebook friend list.

Personas is a fascinating piece of web-based art that visually represents just how much the internet knows about you. Type in your name, and Personas will analyze the text of the Google results, breaking your web presence down into various categories. These groupings include everything from "online" to "sports" to "illegal," and each one has a color, so each person's name results in a unique band of colors.

If you have a unique name, Personas can be disturbingly accurate. If not, it's still visually interesting, but other people who share your name can muddy your results. The flaws in this kind of data analysis are part of the project, though: Personas is about showing how smart the web has become, and how far it has to go before software can intelligently trace our digital histories.

Here's a question for all our elderly readers: Do any of you remember the primitive era affectionately called 1995, and hearing your college professors speak hopefully (or possibly lament) that soon all the information and media ever created would be up on this web thing and easily accessible and available free of charge? Do you remember how many people went out and bought those state of the art 486s and bleeding edge Pentium I computers, and signed on with AOL or Compuserve or Mindspring to fire up Netscape, stumble on to Yahoo! only to discover the truth.

Even back then, there was a lot of stuff online that was technically information or visual/audible media. It was free, much of it, anyway, as well. I spent way too many hours watching an oddity called a webcam update at shockingly fast one minute intervals, as it delivered grainy black and white stlll images of some forgotten webmaster's painted turtle in California to my desktop in Northern Virginia.

As far as exotic, fine art work or rare, priceless tomes of great knowledge went -- it wasn't all accessible online, or necessarly free if it happened to be available. But for a good portion of the '90s, people who hadn't been online much, or were in denial, insisted it was out there.

There dawns the new century, and the myth of "it's all there, free" started to fade away with the old beige Pentium I and II computers. Things went the other way, though. Every day there was more information on the internet from all sorts of sources, and some of it (shock, awe) was free, or at least accessible to some degree. Is it irony or karma? Who knows? Many people are floored, now, to discover how much useful, cool, credible information is available online free of charge.

So just in time to go back to school (or to impress your friends with your innate intelligence), I've found a few sites and tricks for getting really great information online without additional tuition fees.

YouTube is fine for videos of kittens dancing and celebrities make drunken fools of themselves. But sometimes you want something a bit... geekier.



MIT has launched MIT TechTV, a video sharing site for the university's community. You need an MIT email address to upload videos, but anybody can view or share movies from the site.

The videos are powered by blip.tv and cover a range of subjects from news and research to lectures and humor.

[via Digital Inspiration]

Last week the folks over at Creative Commons released version 3.0 of their licensing suite for user-generated content. The bulk of the changes center around clarifying the existing licenses, and addressing the growing internationalization of Creative Commons content. With 3.0 also comes a compatibility structure that will allow them to identify and certify other licenses as CC BY-SA (Attribution Share Alike) compatible. They have also attempted to address concerns voiced by Debian and MIT.

Since the beginning of the Creative Commons in 2002, the language of the licenses had always been based around US copyright law and "generic" in nature because it was not specific to any particular country's laws. So when it came to applying CC licenses to works from other countries, the licenses had to be "ported" to conform to the law of those countries. In fact, the core license has been ported to 30 other countries, or "jurisdictions", to date. To this end, CC have spun the old "generic" core license into two parts: "Unported" (which is based around the language of international intellectual property treaties.) and a separate United States specific license.

If you are already licensing your content under a CC license, you may want to take a look at the updated licenses. But if you haven't taken a detailed look into the Creative Commons yet, there's no time like the present.