Robots and DARPA
Ezra Klein offers up some neat video of a self-reassembling robotic chair:
On the continuum between today’s friendly iPod and tomorrow’s murderous T-1000 that you blow apart with a well placed grenade only to watch it calmly reassemble itself before your horrified eyes, this self-constructing robot chair is pushing uncomfortably close to the T-1000.
I’ve been reading Peter Singer’s Wired for War: The Robotics Revolution and Conflict in the 21st Century and I’m becoming somewhat less concerned about the looming robot slave revolt and somewhat more concerned about serious misapplication of social resources. Singer makes clear something I hadn’t previously understood, namely that military applications of robots isn’t like military applications of the internal combustion engine—a useful technology being used by the military simply because it’s so useful. Rather, in the United States the military actually represents the leading source of funding for basic robotics research (via DARPA) and the leading client for cutting-edge robots.
I think this is pretty problematic in pure economic terms. DARPA—the Defense Advanced Research Projects Agency—is by all accounts an effective government agency and lots of technologies originally developed with DARPA funds have useful civilian applications. But at the end of the day, having research funded by DARPA rather than through some other mechanism tends, at the margin, to channel work to military applications rather than to civilian ones. Now on some level, that can fine. All the productivity-enhancing or quality-of-life-improving technology in the world wouldn’t be worth much if we were groaning under the Stalinist yoke. But realistically, a shortfall in high-tech military gear is far from the most pressing issue facing the United States of America. There are a number of respects in which the USA is lagging behind some group of nations or another, but military technology really really isn’t one of them. And over time our disproportionate focus on military-related research and military-related technology is going to undermine the very economic base on which our military strength—as well as our living standards more broadly—depends.
43 Responses to “Robots and DARPA”
Register
RSSLogin
Topic Cloud
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
February 9th, 2009 at 1:31 pm
there’s a reason that “wired for war” singer publishes under “p.w.”, and it’s so that he will not be confused with the ethicist “peter singer”. help him out a bit, hey?
February 9th, 2009 at 1:31 pm
Without getting too specific, let me just say that, through family members, I’ve had an up-close look at the way DARPA money gets spent, and you’re right to be suspicious.
It funds a lot of basic research. But a lot of that research is of questionable utility.
February 9th, 2009 at 1:34 pm
Perhaps, but I think the bottom line is that the military spends a small fraction of its resources on research. That’s appropriate. Cutting the miilitary spending on research will never in a million years end up increasing NSF, DOE, NASA, NIH, whatever funding dollar-for-dollar, it will just be gone.
So I don’t see what point you’re really making.
February 9th, 2009 at 1:37 pm
I’ve been looking for a robot to chase those young whippersnappers off my lawn. Maybe a DARPA robot will do the trick.
February 9th, 2009 at 1:39 pm
This is nothing new. Among the things funded in early stages by DARPA or its predecessors:
1) computers (which were developed to do the calcuations necessary to build atomic bombs)
2) the internet (which was developed to link together the computers doing the calcuations necessary to build atomic bombs)
3) Lithium ion batteries (ie the things in your laptop, cell phone, and ipod)
February 9th, 2009 at 1:40 pm
Ted, basic research is by definition of questionable utility. If it had known utility it would be applied research.
In defense of DARPA, they did give us the internet. I think we should have groups like DARPA to do public-private R&D for civilian applications. The problem is when it comes to development, which is much of what DARPA does, the government has never really been willing to fund too much because it’s considered interference in the market. That’s why the federal government funds very little development outside of DOD. The real issue isn’t our militaristic ideology, but our free market ideology.
February 9th, 2009 at 1:52 pm
> DARPA funding gave us USENET which is a
> subset of the internet. Linking two
> servers via tcp/ip wasn’t exactly rocket
> science, either.
Um, DARPA’s RFP for a communication system using a meshed network capable of sustaining the destruction of n% of the nodes of the network without data loss is what gave us TCP/IP in the first place.
February 9th, 2009 at 1:55 pm
FWIW, without looking at the actual budgets, I’m guessing DARPA is relatively small in the scale of defense spending. It also does have a history of producing good stuff (I mean, computers and the internet weren’t ALL DARPA’s by any means, but they’re also worth what, 10, 20% of GDP nowadays? Maybe more? That’s a serious payoff.)
I think there’s a concern in the general neighborhood here, but I’d be looking at cutting applied military R&D and purchases. You can do a lot of DARPA for the price of a single F-22…
February 9th, 2009 at 1:55 pm
Their budget is $3.2 billion which seems somewhat small potatoes in the overall research funding business.
From what I’ve read they seem to just throw money at crazy ideas and see if anything neat happens – stuff that could either help them kill people or be the internet… rather than being specifically about weapons.
On the other hand they were probably the villains in Real Genius, and I’m not sure how many eccentric geniuses wwe have on guard to stop them if they get out of hand.
February 9th, 2009 at 1:57 pm
This is just Matt’s refexive anti-militaryism. He can’t help himself, but to bash anything and everything related to the military. Programs like DARPA are the reason that the US has the strongest military in the world. Matt may not like that fact, but I imagine he would like it much less being on the other side of that coin. DARPA’s funding is relatively small in the scheme of things, but it has produced innumerable advances in basic science that private industry will never fund. And, that basic scientific research leads to every development of the modern world. Again, it is easy to bash these developments if you never have to fear living the life of a 3rd world slave laborer.
February 9th, 2009 at 2:00 pm
Wow, the chair rising from the floor is the spookiest image I’ve seen in years. Turn it off! No more!
February 9th, 2009 at 2:02 pm
Many federal agencies that do R&D use the concept of Technology Readiness Levels (TRL) to demark the gradations between basic research and more applied (and commercially useable) research. The Wiki summary is good and there are references to some documents from NASA and elsewhere in government. http://en.wikipedia.org/wiki/Technology_Readiness_Level
For NASA a long standing problem is determining just how far up the TRL ladder it should take a fruitful project, and when it should be turned over to the private sector or to organizations better positioned for actually deploying something. Usually TRL 6 or 7 is as far as NASA goes. Examples in aeronautics developments supported by NASA research include high bypass turbine engines and winglets.
http://www.nasa.gov/centers/dryden/about/Organizations/Technology/Facts/TF-2004-15-DFRC.html
for winglets background
February 9th, 2009 at 2:03 pm
But at the end of the day, having research funded by DARPA rather than through some other mechanism tends, at the margin, to channel work to military applications rather than to civilian ones.
An elaboration of this intuition is one of the major elements in Disciplined Minds by Jeff Schmidt.
February 9th, 2009 at 2:03 pm
I’ve lived off a lot of research grants. Thankfully I don’t anymore. Here’s my perspective.
I work in the field of natural language processing. You wouldn’t think this is a military field, but by far the largest source of government funding for research in this field, by my experience (I didn’t apply for the grants) was from the military and intelligence agencies. We got numerous NSA grants. We got army grants. We got air force grants. There was some NSF funding and we had something from a medical institution of some sort, but the lion’s share by far was from intelligence and armed forces budgets. This means if what we were doing couldn’t be used in some way by these agencies it didn’t get funding. You can imagine why the NSA is a big funder of NLP. We had some entanglement with the Lincoln Group/Iraqex for a time (Google it).
It’s my impression that research money has to come from the military and intelligence because these are the only branches of the government allowed to have any fat in their budgets. The company I worked for got bought by a large corporation so now I work on harmless if not helpful projects to boost their bottom line. I feel a lot better about my work.
February 9th, 2009 at 2:05 pm
My impression, as a Computer Science PhD student, has been that applying for DARPA funding (as oppose to say NSF) means more money, but also much more demands/pressure. This wasn’t always the case, and I do get the sense that professor’s attitudes towards DARPA has soured some in the past decade.
Getting an NSF grant is quite competitive, and hence professors tend to devote a large chunk of their time to grant writing. The grants themselves are typically not that large (say enough to support one or two graduate students for a few years). Once you have a grant, however, you are only required to submit annual progress updates, which is typically brief. The main “accountability factor” for NSF grants is simply that if you don’t publish valuable results, you probably won’t be able to get more grants in the future.
In contrast, DARPA grants can provide much larger sums of money for projects of military interest (for example, a massive project targeted at robotic limbs was announced several years ago, and has been yielding results). Grant recipients are likely required to produce quarterly progress updates, and to travel to meetings and progress reviews. In other words, DARPA funded research is more susceptible to external pressure.
In the case of robotics, I do think a lot of fundamental tools (for example, better object recognition, automatic path planning, object manipulation) could be developed in the context of a DARPA grant. The primary difference would likely be the type of robots being used, and the hardware resources these robots had available (military robots are fancy and expensive, also much harder to destroy in the case of an uprising)
February 9th, 2009 at 2:11 pm
“From what I’ve read they seem to just throw money at crazy ideas and see if anything neat happens – stuff that could either help them kill people or be the internet… rather than being specifically about weapons.”
Worekd with a guy who was a former program manager at DARPA, who said half-jokingly that if a proposal didn’t violate at least one law of physics, it wasn’t radical enough for DARPA to fund.
More seriously, Matt doesn’t understand the path of adoption of a technology. After basic research, you need people willing to fund a high-risk technology that carries a high risk of failure. Venture Capital doesn’t cut it (they need to be 3-5 years after from a product) and the great industrial R&D centers (Bell Labs, DuPont, Xerox Parc) are a shadow of their former selves, save maybe IBM’s. DARPA is filling a void in the technology development path.
February 9th, 2009 at 2:17 pm
Guess one of the lines cut to nothing in the stimulus bill by the vigilant centrists? $100m extra funding (here, Excel line 177) for the Energy Department’s civilian ARPA-E.
February 9th, 2009 at 2:19 pm
I worked in a biomedical engineering lab where one PI had a DARPA grant, and most of what I learned from that has already been recounted above (extreme levels of scrutiny in the process, the quarterly meetings being the most onerous).
But the other side of it is the one to keep in mind when looking at the work DARPA funds. At MIT a representative for DARPA gave a speech, where he said “if all the work I fund has anything above a 10% success rate, I’m not doing my job.” Which is to say it is official DARPA policy to fund outlandish proposals, so that the military has access to the most cutting edge discoveries in the world. If 90% of that money goes to horrible, time wasting dead ends, that’s just the cost of doing business.
February 9th, 2009 at 2:29 pm
One huge obstacle to development of non-petroleum energy sources in this country is the US government’s heavy hand on control of nuclear energy.
February 9th, 2009 at 2:32 pm
We’ve already got people fighting our wars halfway around the world by piloting UAVs from a desk in Nevada. You can even win medals for that kind of combat now.
Just think of the future– entire robot invading armies, proudly funded by patriotic American taxpayers.
February 9th, 2009 at 2:33 pm
In partial defense of DARPA, blue sky research funding is just not very politically palatable in general. Many DARPA projects produce little or nothing and a lot of them sound very kooky; didn’t that dude with the “Golden Fleece Award” poke a lot of fun at them. And agencies like the NIH and energy have it much worse politically.
February 9th, 2009 at 3:13 pm
I see this post less as successful knock on DARPA and more as a successful account of why we need a civilian DARPA (CARPA?). ARPA-E, just on word-spec, sounds like an example of something good, but it probably doesn’t really address the robotics and AI problem. When we get around to the re-assembling robots getting smart, we’re going to want Autobots rather than Decepticons. It’s not useful to the military to create massively intelligent robots that follow Asimov’s rules of robotics (only the third rule is any good at all from a military perspective).
February 9th, 2009 at 3:40 pm
The chair breaks itself. Its clearly not that intelligent.
February 9th, 2009 at 3:42 pm
If you want to talk about misdirected resources, I don’t see any evidence that that chair would actually support anyone’s weight. In the second run in the video, look how the leg wobbles before it falls over.
What on earth is the point of a chair nobody can sit in? THAT is a waste of resources.
February 9th, 2009 at 3:47 pm
But at the end of the day, having research funded by DARPA rather than through some other mechanism tends, at the margin, to channel work to military applications rather than to civilian ones.
Your assumption seems to be that if the money weren’t being given to DARPA for military research it would be given to a civilian government agency for civilian research instead. But that isn’t necessarily true. In fact, it probably is not true. The money might be used for tax cuts instead. Or for civilian applications you oppose, like human spaceflight.
February 9th, 2009 at 3:52 pm
DARPA is probably the leading funder/promoter of R&D on robotic cars, which threaten Matt’s vision of a transit-and-density utopia. No wonder he’s against it.
February 9th, 2009 at 3:55 pm
Here, here. This is the fundamental concept of risk-to-reward ratio. I know that the American people have forgotten this little lesson because Wall Streeters kept telling them that you can get high returns for little or no risk. How did that turn out for you?
This is why we fund blue-sky research and foundational research. Because, while they have a high failure rate, the long term economic gains can be tremendous, especially as they work their way up the technology readiness ladder. It is also why other countries have been trying to create similar funding programs of their own.
Businesses are a horrid way to fund this type of investment, because their outlook is too short term to see a return on their money. The only businesses that have ever funded this type of research were effectively monopolies (AT&T, Microsoft, Google) at the time that they funded it. As they ceased to be monopolies and their business sector became competitive, money because less and less available.
February 9th, 2009 at 5:07 pm
What do you DO with robots besides either 1) have them take jobs from humans throwing ever more humans who are not intelligent enough to work in non-robot jobs on the streets and 2) equip them with swords and lasers and have them fight it out in Tokyo.
February 9th, 2009 at 5:25 pm
Well, to answer the first question, read this site:
http://www.peoplescapitalism.org/
I’ve never studied it in detail myself, so I have no idea if he’s right. But it addresses the question.
Second, consider Cameron on the “Terminator” show. One thing they ignore about that character is that she is the most valuable artifact on the planet in this time period. Her robotic endoskeleton could revolutionize industrial production, and her AI chip is worth at least six trillion dollars in the educational industry alone. Just as a functioning individual robot, she’s incredibly valuable.
You could do five seasons of a spinoff show like “The Fugitive” or “The Hulk” just having her chased by everybody all over hell trying to get their hands on her.
Yet the Connors have her doing their laundry! (When she isn’t trying to do John Connor.)
February 9th, 2009 at 5:39 pm
The world of professional wrestling will never be the same
February 9th, 2009 at 6:37 pm
I notice that at 1:20 there is an edit, following which one of the chair legs can be seen to have been moved, presumably by a person, so the chair could assemble itself.
So the chair needed human help, and that fact was concealed by the video makers to raise the WOW! factor as well as to deceive the viewer.
Color me unimpressed.
February 9th, 2009 at 8:41 pm
Erik W,
Instead of creating a civilian DARPA, which will invariably be cut out of the budget sooner or later, just give the existing DARPA more money with a mandate to fund more research with civilian uses.
.
February 9th, 2009 at 10:40 pm
The main question is whether aligning the interests of defense (which is a sacred funding cow) and research (which is a valid function of gov’t) does more harm than good.
In my view it is good to align the interests of defense and research. Even if you skew the research agenda towards defense applications, you still greatly strengthen the funding position of the research sector. Maximizing funding will maximize innovation and positive-externality spillover.
The first best solution perhaps is a populace and government that cares about funding exactly the most useful things. But pragmatically the current setup is pretty good IMO.
February 10th, 2009 at 1:47 am
I suspect that Darpa does have the disadvantage that they reward people that are willing to lie about their capabilities. I’ve heard stories about a researcher that got funded on one of their most outlandish robotic projects telling everyone at a conference that they got away with a scam.
March 11th, 2009 at 4:42 am
Excellent site. It was pleasant to me.
March 12th, 2009 at 11:30 pm
I bookmarked this site. Thank you for good job!
March 17th, 2009 at 2:27 am
It is the coolest site,keep so!
tramadol
March 22nd, 2009 at 6:16 am
tramadol
Incredible site!
March 22nd, 2009 at 10:30 am
buy viagra online
Great site. Good info
April 2nd, 2009 at 5:21 am
If you have to do it, you might as well do it right
buy cheap viagra
April 3rd, 2009 at 4:13 am
I bookmarked this site. Thank you for good job!
cheap brand pfizer viagra
April 9th, 2009 at 6:14 am
Very interesting site, Hope it will always be alive! viagra
April 19th, 2009 at 8:56 am
uQyF6K bcseccqfzwwv, [url=http://uzdswejprjvj.com/]uzdswejprjvj[/url], [link=http://fnwkcbksbunx.com/]fnwkcbksbunx[/link], http://rvwryaflzyra.com/