Software Carpentry

Helping scientists make better software since 1997

YouTube for Data

TechCrunch is running an interesting report on Swivel, which bills itself as “YouTube for data”.  Anyone and everyone can upload data sets to share with the rest of the world, link together, and what have you.  It’s going to be a great resource for teaching…

…but I’m less enthusiastic about it for science, unless its creators have some verification and validation magic up their sleeves that they’re not yet talking about.  Still, it’s another sign of what Richard Dawkins called “the evolution of evolvability”: every new kind of thing (segmented bodies, eyes, web services) gives evolution a new affordance on which to act.  Taggable public upload is turning out to be a very mutable affordance…

Written by Greg Wilson

2006/12/05 at 07:21

Posted in Content

2 Responses

Subscribe to comments with RSS.

  1. A lot of folks started saying YouTube for data, but Dmitry and I both chuckle because it wasn’t us. We are going to focus on making data something you can have fun clicking around in, instead of something you need a manual to operate. Dmitry and I both studied physics and on the one hand we agree with you, Swivel will probably not be for science: the competition is pretty fierce with Mathematica (a very inspirational piece of software) on the one hand and GNUPlot on the other. However, we hope that the same honest, open peer review process that drives good science in academia will also drive balanced and fair data verification and validation at Swivel. The cool thing about a Web site versus a piece of software like GNUPlot is that a Web site is half communication vehicle. So, we hope that the validation and verification actually accelerates at Swivel. Please swing on by once we open up.

    Brian Mulloy

    2006/12/05 at 09:30

  2. Wow, that was quick turnaround 😉 I’m very interested in Swivel
    because I teach an open source course on basic software engineering
    skills for scientists and engineers (; having
    high-quality data sets available to mess with will make life a lot
    easier, and provide “good practice” examples for grad students. My
    concern is quality control: what are you planning to put in place
    to give people an idea of how trustworthy data is? Or is your model
    to become “Wikipedia for data”? 😉

    Greg Wilson

    2006/12/05 at 09:45

Comments are closed.

%d bloggers like this: