DARPA (the Defense Advanced Research Projects Agency, just called ARPA during some periods of its history), is legendary as the sponsor of what became the Internet. More generally, it has an outstanding reputation as being that great oddity, a government agency that works, creatively and efficiently. (This Wikipedia article on DARPA is generally consistent with the agency’s public perception.) And so, when I was thinking about ways the US government could usefully support the technology industry, my thoughts naturally turned to DARPA, and the possibility of creating civilian DARPA analogues.
But a glitch soon appeared in this line of analysis — I found I couldn’t think of anything much tha DARPA has accomplished in the past 30-35 years. I remember some flashy AI projects in the 1980s, like the Autonomous Land Vehicle (ALV), but those didn’t amount to much, despite a quarter-century of more or less continuous effort, continuing to the present day.
Googling around, I found a 2003 study assessing DARPA’s effectiveness, but focusing more on issues with widespread adoption than on DARPA-sponsored advances themselves. It gives DARPA a great deal of credit in the areas of stealth aircraft and, probably more important, standoff precision strike. But if I understand correctly, most of the DARPA-backed research breakthroughs in those areas were made in the 1970s, after which other parts of the DOD took over. Another article adds one biggie to the list — GPS (Global Positioning System). But other examples, like speech translation, have been crawling along for so many decades now they hardly rank as proof of success. (The telling jokes in this blog post may be half a century old.)
Maybe I’m being unfair. Getting from ARPANET to Netscape Navigator took 25 years. GPS evidently took even longer to mature. If that’s how long it takes for DARPA initiatives to be obvious big successes, maybe I’m just being too impatient.
