Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why Does DARPA Work?

benjaminreinhardt.com

Post reply on HN
Print view View original

31–40 of 98 posts

Re: Why Does DARPA Work?

#31

Post by siege_engineer »

currymj wrote: post #18

A bill, the Endless Frontiers Act, has been introduced in both houses of Congress to make a section of the National Science Foundation that works more like DARPA, while massively increasing its funding. The idea is to extend the DARPA model to many more non-military areas, just as suggested in this article. https://www.sciencemag.org/news/2020/05/us-lawmakers-unveil-... Certainly a timely article. I don't know whethe…

Author here. I actually wrote a discussion of the Endless Frontier Act: https://benjaminreinhardt.com/the-endless-frontier-act/

I agree that the ARPA model is not suitable for all (or even the majority of research) and the act doesn't inspire confidence that shifting the NSF towards a DARPA-like model would do well.

Re: Why Does DARPA Work?

#32

Post by Animats »

riazrizvi wrote: post #17

IMO this analysis gets caught up looking for answers in process when the real problem that DARPA solves is a political-economic one. ATT didn't invent the internet not for lack of process, they were extremely innovative. They held back on innovations because they were making huge sums of money overcharging long distance mainframe-mainframe data link rates in the 50's-70's. DARPA succeeded as a trust buster. They have…

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.

Re: Why Does DARPA Work?

#33

Post by hitekker »

riazrizvi wrote: post #17

IMO this analysis gets caught up looking for answers in process when the real problem that DARPA solves is a political-economic one. ATT didn't invent the internet not for lack of process, they were extremely innovative. They held back on innovations because they were making huge sums of money overcharging long distance mainframe-mainframe data link rates in the 50's-70's. DARPA succeeded as a trust buster. They have…

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?

Re: Why Does DARPA Work?

#34

Post by peterwoerner »

thoraway1010 wrote: post #20

Small anecdote, but good experience with DARPA and SBIR both. One thing I liked - they seemed pretty low overhead / low hassle and outcome / objective focused. That contrasts VERY strong with most govt contracting which is under allowable cost or cost reimbursement setups where the absolute most important criteria is to bill enough costs to draw contract in the right cost buckets (which can be super annoying if local…

I used to work for a company doing SBIR work. We finished a contract underbudget returned the surplus money and got audited because of it.

I don't know all of the accounting details, but SBIR has some leeway (because its supposed to be research) but it's not perfect. However, you are allowed to use surplus money to fund development of things which are related to what you actually put in the proposal/contract. So if you have a $100k contract, and meeting the requirements only required on the $75k, you should and do spend the rest of the $25k on extra features. But you have to spend all the money.

I think the real thing about SBIR money, is that three people get phase 1 money and only one or two will get phase 2 money so you really have to deliver in order to get selected. Then if you get the phase 2, there basically isn't phase 3 money, so you better get to a product that is ready to be sold. Finally if can't show on paper a return on investment (e.g. nonSBIR funding or revenue from sales/acquisition) they cut you off from the SBIR spigot. So the incentives align with getting work done.

Re: Why Does DARPA Work?

#35

Post by lazyjeff »

riazrizvi wrote: post #17

IMO this analysis gets caught up looking for answers in process when the real problem that DARPA solves is a political-economic one. ATT didn't invent the internet not for lack of process, they were extremely innovative. They held back on innovations because they were making huge sums of money overcharging long distance mainframe-mainframe data link rates in the 50's-70's. DARPA succeeded as a trust buster. They have…

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).

Re: Why Does DARPA Work?

#36

Post by jsizzle »

bitcharmer wrote: post #29
dnautics wrote: post #11

Earlier quoted context omitted.

Perhaps I have a flair for the dramatic, but consider my statement that I haven't met a competent pm in the division I was in (n~5). The incentives are not that well-aligned. Think about it this way. If you're really smart, why do you become a DARPA PM, and not a PI? Yes there are a few good reasons, but the population it applies to have largely been selected out by the postdoc phase (where I would say you have had s…

What does PI mean in this context? I've worked in software projects for 20 years and not once have I stumbled upon the term.

Principal Investigator, basically the lead researcher

Re: Why Does DARPA Work?

#37

Post by tia4tia »

tia4tia wrote: post #30
thoraway1010 wrote: post #20

Small anecdote, but good experience with DARPA and SBIR both. One thing I liked - they seemed pretty low overhead / low hassle and outcome / objective focused. That contrasts VERY strong with most govt contracting which is under allowable cost or cost reimbursement setups where the absolute most important criteria is to bill enough costs to draw contract in the right cost buckets (which can be super annoying if local…

SBIR? Totally corrupt. Please don't ask. Can not, will not, must not tell. I can tell you one thing. If not the best but the second best or third best proposal is accepted, IT IS a problem for the US in the long run. Side node: there is money on China! Fell free to down vote me down. Again Please don't ask. Can not, will not, must not tell.

I can tell you one thing about SBIR. "We have occasionly accepted new researchers"

Really? SBIR is about blue sky research and funding blue sky ideas, potentially to risky for VC. It should be about NEW ideas and not about funding for contract research of buddies. And for SME and not for "independwnt" spin offs of large corporations.

In China, I can get you funded.

Re: Why Does DARPA Work?

#38

Post by IggleSniggle »

riazrizvi wrote: post #17

IMO this analysis gets caught up looking for answers in process when the real problem that DARPA solves is a political-economic one. ATT didn't invent the internet not for lack of process, they were extremely innovative. They held back on innovations because they were making huge sums of money overcharging long distance mainframe-mainframe data link rates in the 50's-70's. DARPA succeeded as a trust buster. They have…

With all due respect to the incredible amount of intelligence, knowledge, and insight in display on HN, your comment, with it's culminating sentence, may be the most insightful thing I've read on HN.

I don't care if this comment doesn't meet HN guidelines. I needed to celebrate your comment.

Re: Why Does DARPA Work?

#39

Post by dnautics »

darpa_commentor wrote: post #23

DARPA is effective, but this document paints a rosy picture that is far from the truth. I've been a part of DARPA projects both commerical, university and on the government side. The real process is: 1. PM gets picked due to knowing someone or being a former employee. 2. The biggest test of a program isn't if it's doable or a good idea, but if its able to be transitioned to another government agency with deeper pocke…

this aligns with my experience. NIH has a better track record than DARPA, which has a better track record than NSF and the worst is the DOE. The DOE knows this and is trying to cargo-cult DARPA through stuff like ARPA-E. Predictable results have ensued.

Re: Why Does DARPA Work?

#40

Post by IggleSniggle »

Animats wrote: post #32
riazrizvi wrote: post #17

IMO this analysis gets caught up looking for answers in process when the real problem that DARPA solves is a political-economic one. ATT didn't invent the internet not for lack of process, they were extremely innovative. They held back on innovations because they were making huge sums of money overcharging long distance mainframe-mainframe data link rates in the 50's-70's. DARPA succeeded as a trust buster. They have…

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 bandwi…

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