Dr. White, who heads this up, spoke yesterday at Strata . . . it was the most bizarre talk I heard in the 3 days by a mile. At one point, he actually put a slide from a Wired peace that bashed him, then spoke for about 45 seconds about how the piece was wrong when no one in the audience really cared. He's just an odd duck.
On this, he put forth a really solid vision of what he sees coming out of this - a classic melding of Machine Learning dark arts with the ease of use of a traditional, modern web framework. It's a pretty bold vision, but he's got the cash to spread around to certain contributors.
If I had piles of govt money to spent I would rather support the development of already established OSS projects with adequate teams and would try to influence further direction by contributing "scientific basis" and "rational reasoning" about functionality I think is necessary. That means I would enjoy "free" community testing and support and probably even bug-fixes.) The mantra for Open Source is "make something for others to [re]use", which, I think, govt officials could never understand.
I completely agree with your emphasis on the community as one (if not the) defining attribute of what makes open source special. I think, however, that attitudes towards open source in government are definitely starting to shift. Events like the (3rd annual) Open Source Summit [1], and efforts like XDATA contribute greatly to overall awareness of the importance of understanding and embracing open source in government.
As far as supporting established OSS projects, this is definitely a big part of what XDATA is doing. Several of the projects on this list have well-established communities at the Apache Software Foundation (e.g.: OODT, Spark, Shark, Mesos, Tika) and others have strong ties to university research programs (Stanford, University of Washington, USC, UC Berkeley, etc.)
In a sense, what XDATA is doing is helping to connect these communities and funding them to come up with ways to collaboratively leverage their software and skills to solve data-intensive problems.
Not surprising Julia is DARPA sponsored. Julia has a big future ahead of itself in the military and government. It has perhaps more of a future than any other modern language in those industries.
> Julia has a big future ahead of itself in the military and government.
Hmm, curious, why would that be? Or rather, what are the features that make you think Julia has that other languages and technologies don't have that would be killed in military and government?
I work within that area and if anything the military and government love their red tape, certifications, protection profiles, rubber stamps that it would seem like the last industry to jump on a new and largely yet unproven language. You wouldn't believe the number of Windows XP and RHEL 4 and 5 servers still running.
The difference between the high quality math guys and the high quality distributed jocks at that level is much larger than I think most people think. I come from the engineer side, and it's really bizarre to hear people w/ Ph.D.s in machine learning or physics give talks where they will mess up what is really a junior level distributed programming concept. Happened twice at Strata that I saw. Mind you, that's for BASIC things, not someone trying to right a highly concurrent computational system across hundreds of nodes. No lie, I had a conversation with a machine learn jock that asked me point blank when we couldn't make Matlab go faster.
On the flip side, the distributed engineers might even be worse at the math than the data jocks are at programming. A middling concept for a machine learning guy, like say a Bayes error, is like trying to write concurrent state machine in assembly.
tl;dr - the pool of people that are both elite math jocks and elite system jocks is vanishingly small, as in I would be shocked if that number was out of the triple digits world wide.
Julia has a lot of promise in that it gives the math guys familiar syntax, but appears to have the underlying capability to be very fast on shared memory architecture. It's already got the handles to call out to python / C / Fortran bindings.
On this, he put forth a really solid vision of what he sees coming out of this - a classic melding of Machine Learning dark arts with the ease of use of a traditional, modern web framework. It's a pretty bold vision, but he's got the cash to spread around to certain contributors.