Nearly 16 years ago, I got call more-or-less out of the blue from a fellow named Rob Utzschneider, who wanted to interview me for a parallel processing start-up named Applied Parallel Technologies. Rob and his co-founder, Ed Zyszkowski, had just received funding under the Advanced Technology Program, a program run by NIST, an agency of the U.S. Commerce Department. The intent of ATP was, "...to spur the development of path-breaking new technologies by providing cost-shared funding for potentially valuable but high-risk R&D projects." Over a two year period, ATP put about $1.1million into Applied Parallel Technologies (APT - confusingly similar to ATP...), which paid our salaries, bought us workstations, gave us an Internet connection and who knows what else.
Over time, APT morphed into Torrent Systems, acquired VC funding, and was eventually acquired by Ascential Software and, ultimately, IBM. The system we developed - Orchestrate - is still at the heart of IBM's Information Server software. Given its continued survival, we must have done something right.
Evidently NIST thought so, too. I just stumbled over an ATP report describing the performance of the first 50 completed ATP projects. Our work earned 4 stars - the highest possible rating - and was among the 8 top-ranked projects. The project summary brought back a lot of memories.
So - I guess that, at least in our case, a government program actually did just what it was supposed to do. The ATP grant got Torrent off the ground, and we were successful enough that venture capital followed. We built a working system that helped bring parallel processing into the mainstream. The MapReduce software used by Google is a variation on the flow-based, data parallel concepts embodied in Orchestrate.