DARPA - Our Research | Hacker News Reader

DARPA - Our Research

darpa.mil

67 pointsrfreytag21 comments

As principal investigator for three of the listed DARPA projects (APAC, CRASH, STAC), I have to say that DARPA is by far my favorite research agency to work with.

Every DARPA project I've worked on goes through the same three phases:

1. This is impossible.

2. This is impractical.

3. Holy shit! It works!

DARPA is unique in many ways: all agency positions are temporary (2-4 years usually), so there's no technical stagnation.

There's a high tolerance for risk. It would be remarkable for more than half the teams on a project to reach the objective.

And, they actually give you all the funding you say is necessary to complete the objective.

I certainly have my gripes with DARPA too, but in comparison to virtually any other federal research agency, it's a pleasure to work with them.

One thing that many people may not realize is that DARPA doesn't award grants, it awards contracts. You're obligated to achieve your deliverables, in contrast with NSF/NIH/etc where you might discover something new and follow a different path than you anticipated in your grant proposal. This is not a bad thing, but something to be aware of.

Second, the reporting periods are much shorter. You'll have a call with your DARPA PM every two weeks. This is for DARPA to provide any assistance they can to address roadblocks and move forward, but it does keep the pace relentless.

I also find DARPA projects to be the most ambitious and interesting projects I've worked on so far. Much less risk averse and highly imaginative.

> DARPA [...] awards contracts. You're obligated to achieve your deliverables

Deliverables can be a report if you have nothing more. Easier than handling the changing demands of angels and VCs.

DARPA contracts pay you to do the promised research on the topic you proposed and then report your NEGATIVE or positive results, in writing, to three entities: PM, DARPA Director, and DTIC (Defense Technical Information Center). DARPA's compliance requirements are minimal and very reasonable.

As performer on three DARPA contracts I can say the PMs were professional, sharp, and gave me great flexibility to pursue results. I did the research behind at least 8 products, probably unfundable due to risk, with the support of DARPA. Best of all, I have ownership of the products and my entire company post-funding.

Quickly developing a trusting relationship with your PM and the staff is key to a good result; as with any customer. DARPA PMs are all current in their fields and know their time to produce results is limited (2-4 years) so they expect "high risk" submissions that might deliver revolutionary results.

The only catch is taking research results to market is non-trivial. However, even there DARPA is getting better at helping you find partners and follow-on contracts.

Completely agree - we've got a fantastic PM under the new BTO and that makes a huge difference to the success of the project.

I like parts of the darpa model, like sufficient funding to accomodate failure, and fast timescales.

But it can be distracting for some people, where lab members will be pulled from more basic science questions to more engineering 'how can I make this closed-loop stimulator hit these white matter tracts?' questions.

I've been fascinated by the ambition and innovation in most DARPA projects. How does someone who is in industry rather than an academic context get involved in a DARPA project? More info on my background for context: http://codinginparadise.org/about/

The work that DARPA does is incredible and we should be proud, as a country, that we pursue transformational research. One of the big weaknesses of the private sector is that there are few opportunities to pursue high-risk high-reward research and research that must be done on long timelines.

The stuff DARPA funds is fundamentally important and has an incredible track record.

That said, their primary customer is the DoD which has demands for technology that is sometimes rather creepy. Without going too far down that route and forcing a tangent on the thread I'll just plug here that technology is power and that power can be used for good and evil, for noble and ignoble and for personal or public gain. We need to cheer on DARPA for its development of technology and keep its customers accountable for how it applies that technology.

Do you know of non-DARPA organizations doing the kind of high-risk, high-reward work DARPA does but without the military angle?

I do not. Microsoft Research and Google Research and IBM are examples of places where some of this research is done - and there's Universities. MITRE does as well as other semi-public orgs.

But it's the case that all of these places also sponsor work and work with the military. I guess the answer is yes that it exists, but no it doesn't exist (AFAIK) completely military free.

I don't have very good visibility into pharma companies and agro companies. I know that they do some long term research and I don't know (I would suspect) whether they have partnerships with DoD folks.

Also something to keep in mind: DARPA PM's are only temporary jobs. PM's are not contractors, but the residency only lasts 2-3 years. If, as a PM, you start a project, at the end of your term, your project is handed off to a new PM.

This helps keep the org fresh, new talent coming in and out all the time and there are side benefits in reducing potential corruption. You come in with your big ideas, you get a chance to try them, and then you're out to make way for the next person.

Do you swap projects, or do you move on from DARPA entirely?

It depends. The PM on the (now completed) DARPA project I worked on had at least one other completed project under her belt by the time ours started, but moved on to other things outside DARPA near the end of our project.

I believe you're correct that PM positions are temporary, but most projects are 4 years long, so I think the 2-3 year estimate is too short.

Speaking as a Canadian, DARPA is one of the few things that make me wish I was an American.

Is this defense funding at work?

Yes - often times towards basic science and areas that are deemed "too risky" for NSF/NIH/DOE/NASA/etc and not towards weapons. Important to note that DARPA has funded the research that led to GPS, the internet, critical components of cell phones, autonomous vehicles, and many other pieces of core technological infrastructure.

The creation of the new Biological Technologies Office (BTO) this past year is particularly exciting. Example projects I find exciting:

Neural prosthetics for psychiatric disorders: http://www.darpa.mil/program/systems-based-neurotechnology-f...

neural prosthetics for memory: http://www.darpa.mil/program/restoring-active-memory

GPS was developed primarily to improve the aiming of nuclear weapons, afaik, let's not fool ourselves.

Neural prosthetics sound like they want to create automatons, but it's possible that they want to deal with the military's epic mental health crisis.

That's without a doubt a major motivation of neural prosthetics. Rehabilitating soldiers who've lost limbs due to IEDs, etc, is one of the explicit reasons for DARPA's REPAIR program as one example. I don't see that as a weaponization though, and find it a morally defensible motivation for research, though I recognize that many members of this community in particular may avoid projects with any connection to defense.

No Tor?

Projects related to Tor were funded as part of the Memex project. DARPA only funds transformational research, not iterative improvements to existing technology, so simply funding "Tor" is not a thing they would do.

Reply on news.ycombinator.com