ARPA Programs Don’t Work

Lately, there has been a craze over reviving ARPA style programs. There’s the original ARPA and DARPA, the newer ARPA-E and ARPA-H, the UK’s ARIA. There’s the forgotten HSARPA for homeland security, IARPA for intelligence, and ARPA-I for infrastructure. Recently ARPA-C was proposed for climate research.

There’s just one problem with all these alphabet agencies: they don’t work. Let’s look at the record. Starting with the original: ARPA.

A popular myth is that ARPA is responsible for the personal computer and the internet. Both of these claims are false. ARPAs role was at best marginal.

In the case of the personal computer, what happened is that 10 years after its creation in 1958, ARPA was so unsuccessful that Congress had to step in and limit funding to purely military applications. This forced many computer scientists that had previously been employed by ARPA to move to private research facilities such as Xerox PARC, where they invented the personal computer. So ironically, it was only by defunding ARPA that we got the personal computer.

Not to mention that the subsequent rapid development of the personal computer was almost entirely driven by private corporations such as Apple, Microsoft and Commodore.

Next, the internet. The misconception here is to claim the internet is the protocol. ARPA did indeed fund TCP/IP and create ARPANET.

But the internet is not just the protocol. The internet is the physical mass of servers, optical fibers, and data centers that connects us all. And here again I have bad news for ARPA: it was companies like Cisco, Lucent and Nortel that invented and built out the physical internet infrastructure, with little or no support from the government.

The key question is the counterfactual. In a world where ARPA doesn’t exist, would these companies have developed the protocols themselves?

I think the answer is yes. The reason is, an invention only exists after it has meaningfully impacted people’s lives. If the internet was invented by DARPA in the 1960s, why did it take until the dot-com boom (and subsequent bubble), for it to meaningfully impact people’s lives?

Ultimately, what Progress Studies cares about is improving quality of life. And when looking at the ARPA record, it’s hard to find a single instance of it actually working. The world would be better off without ARPA.