How are they becoming less relevant?

The DARPA grand challenges have bootstrapped:

1. Self driving cars and got them working as a prototype. [1] 2. A rapid Orbital Launch Program (Still in progress after only one competitor left)[2] 3. A subterranean challenge to map and search underground environments [1] 4. A way to use social networking to find something that is time critical [2] 5. Onion routing which is used in Tor. [3] 6. Cyber Grand Challenge to automate the exploitation and securing network systems [4]

Note that competitors also have open sourced what they have done. [1] https://en.wikipedia.org/wiki/DARPA_Grand_Challenge

[2] https://en.wikipedia.org/wiki/DARPA_Network_Challenge

[3] https://en.wikipedia.org/wiki/Onion_routing https://en.wikipedia.org/wiki/DARPA_Grand_Challenge_(2007) [4] https://www.darpa.mil/program/cyber-grand-challenge

> The DARPA grand challenges have bootstrapped:

I have a contrarian opinion about the impact of DARPA and related programs (ARPA-e especially). To some extent, there is a confusion of cause and effect. DARPA is good at identifying possible nascent technologies and funding them. They then have a flag planted, and can claim "we helped invent X", which is hard to disprove. But was it because of DARPA involvement, or was DARPA merely there at the right time? They have definitely funded plenty of projects that go nowhere.

ARPA-e makes this strategy explicit. They'll give money to an up-and-coming startup between rounds and then claim all later investment as "follow-on funding". They then report "follow-on funding" as a success metric to congress, which seems easy to game. ARPA-e has been around for over 10 years and I don't think they can point to anything coming out of the program as a legitimate game changer.

I do think the applied research programs do some good and has some impact, mostly as a way to convene the broader technology community. But for the most part I would rather have the government focus on funding basic research instead of applied research.

What's the difference between "helping invent X" and "identifying possible nascent technologies and funding them"?

Nascent technologies aren't a fait accompli. Without funding, something expensive like a launch system isn't going to get anywhere no matter how clever an idea someone has on paper.

At least in the self-driving car case, it's hard to say they didn't have a big impact. For universities to allow, fund, and support students to form engineering student teams, there usually needs to be some kind of official competition that the team can compete in. In this case, those university teams made major progress on what was at the time seen as an intractable problem. Google and its ilk then headhunted all of the students in the top performing teams to launch their self-driving car division, and now here we are.

My question is... do they really? They happen to be around when tech moves, but would it have moved without them? It's a historical counterfactual, and hard to argue about. I'm skeptical and I'm more skeptical of ARPA-e.

How are they less relevant? Well it's a bit left-field but I believe part of the reason why the political climate has become more pro-trust within the last twenty years is that international tech dominance is a national security boon. DARPA is not going to be asked to create ground-breaking tech in search, tracking, social media, tech generally, because the tech giants are already providing more than defense strategists could possibly have hoped for, on the world stage.

So instead we have these contests, the insights of which are handed out to existing companies to help them make incremental improvements. But the Onion browser isn't going to be rolled out to secure political autonomy throughout the world, or small business competitiveness because the data tracking is too valuable for surveillance programs. The national security/intel value of big tech companies has given those companies so much good will that govt officials/politicians have been unwilling to consider innovations that would undermine them.

I’m not arguing applications I’m arguing innovation. They do the initial research and then companies monetize .

Back in the early days of the Internet, I met some of the Bell Labs people working in that area. What they didn't like about an IP-based network was the jitter. That was totally unacceptable for voice. They wanted something with reliable clocking, and had come up with Datakit. That sends all packets for a given call over the same path, in order. You don't open a virtual circuit unless all the nodes have enough bandwidth for it. Telcos still use Datakit, and Asynchronous Transfer Mode is a successor to it.

I never dreamed that people would accept the degradation of telephony to 1 second of delay jitter with random dropouts and echoes.

I remember when cross-oceanic long distance calls required both parties to shout as loud as they could into the phone -- and often still not be heard well enough to make out what they were saying.

Modern IP telephony often has very high quality voice reproduction (my wi-fi to wi-fi Fi calls sound fantastic), in exchange for some timing issues. Echoes usually get solved in software (usually), and dropouts seem to be the main complaint.

In exchange, my wife can call her family in South Korea for approximately nothing, using the same data backbone as we use to watch movies and read web pages.

You can have ip over time multiplexing data link layer, like the IP over ATM. I believe they were still used in the core voice networking.

IP won because they are more flexible, and open. And the Internet cement the win because of that.

That's way planning ahead too far works less and less successful for bigger and bigger project. Too much dynamic is embed in the long wiring process. It becomes impossible to have a plan work out correctly.

1 second of delay jitter with random dropouts and echoes are typically caused by problems at the ends (poor wifi/cellular signal, too much load on cpu during a high res video call, laptop overheating, etc). UDP protocol allows to make a call even in the presence of such issues, whereas circuit switching with reliable clocking would simply not work at all.

These days, I suspect there's a lot of ISP last mile issues with everyone working at home.

Acceptance of the degradation couldn't be done in a day. It required societal acclimatization.

That's a good point. University labs are more nimble than a military contractor. The PI at a university directly does the work with their students (in some big labs, it's arguable that the PI manages the students that directly do the work). Whereas the contact at a military contractor is an accounts person, with lawyers and managers between them and the people doing the work.

Cost-wise, universities usually charge an indirect cost rate of 50-65% (which many people improperly think it means is the percent of the grant money taken by the university), whereas the military contractor charges an indirect cost rate of 120-200%.

So basically, about 70 cents of every dollar spent at a university directly goes towards the work (and trains students as a side effect; and as a triple bonus the cost paid towards student's tuition has no indirects on it at all), while only 30 cents of the dollar spent at a military contractor goes towards the work (and probably much less towards the people actually doing the work).

On the other hand, is one "unit" of student effort equivalent to one unit of staff scientist work?

I would say no: early-PhD students in particular are still learning and have other responsibilities (coursework, quals, etc) to boot. You can also interpret your numbers as suggesting that an experienced person is 2-3x more productive than a trainee, which seems....quite reasonable, actually.

I think that's the other "secret": the combination of large budgets and tight timelines encourage the use of more experienced staff and smarter division of labor: hire a programmer for this, contract out to a machinist for that. This is very different from the model used on most NIH grants, where grad students are used as jack-of-all-trades.

Author here. Good point about DARPA serving a political-economic role. It's a different lens on the coordination role.

Is the corporate-political landscape becoming more trust-based than the Military-Industrial Complex that Eisenhower warned about?

I don't understand the comparison in reference to the question of 'Why Does DARPA Work?' or my comment.

Military-Industrial Complex is where defense industry profit motives drive unwarranted political influence that shapes defense policy, foreign policy. Military-Industrial Complex drives defense spending up.

Trust is about the erosion of market competition, which results in decrease product value-price ratio. In the context of defense product, it means that the products that the consumer (here military/govt) can buy from the market become increasingly expensive and weak. Which creates an opportunity for consumers to do-it-better-yourself-and-for-less, which to me, explains DARPA.

To a certain extent the MIC as you define it would explain the erosion of the market wouldn't it?

Also I've read your comments a few times now and I'm not sure if I'm completely understanding your view of what is becoming more "trust-based". Are you saying that DARPA is more trusted by the consumer (the federal government) than the market entities (the contractors)? If so, isn't that exactly what government entities of all types are created for? These agencies exist as part of the government working for the government because it's difficult to have an effective private market created for these similar purposes.

I think it makes the question about "what makes DARPA successful" so much more interesting from if we acknowledge the limitations of the sector. Many agencies, entities, or initiatives of various types have failed when given similar mandates and similar market conditions. That's even if you believe DARPA is truly as effective as is claimed at achieving it's mandate of innovation in the Defense and National Security sector...

You raise an interesting point that the author shies away from addressing.

Can you elaborate on the phrase "becoming more trust-based" ? Do you mean that decision makers in this landscape rely more on the strength of their connections than their individual understanding?