Totally agree with the author. For a Matlab-flavored large scale processing DSL, check out the matrix library in scalding: https://github.com/twitter/scalding/wiki/Introduction-to-Mat...

IMHO, two things that will make it perfect: 1)Support of onMouseWheel 2)Support of mobile gestures BTW, does anybody know of anything supporting those?

I like to think of Mr Khan efforts as something similar to those of the mentor in Sophie's World. http://en.wikipedia.org/wiki/Sophies_World He can teach the audience on how to think about subjects, but I doubt he can be deep on all of them. This is very helpful but is not scalable. I wonder how this effort will become mature if it continues to get funded and promoted.

There is a whole literature about it. Check http://en.wikipedia.org/wiki/Discounted_Cumulative_Gain

Hi,we are launching a service exactly like the one you were referring to. You can try the API already. Check out the docs https://sites.google.com/site/thinkersrus/products-1/science... and the demo http://www.thinkersr.us/demo.html

Oh nice, looks interesting and worked pretty well from my few tests. Sadly, with the amount of requests I'd be doing (10000+ just for training), I guess I should build myself something on my own

Funny thing, we have a start-up in submission to YC that should help with that. We hope to release a demo to HN in a couple of days.

Awesome... can I have a beta invite? :)

Hi, here is the link to detailing what you can try out right now. The official beta launch is going to be later on this week, but you can get started using it anytime you like: https://sites.google.com/site/thinkersrus/products-1/science...

Enjoy! :)