The military can do some repairs themselves - every system defines the Line Replacable Units (LRU) and has a detailed plan for what are reasonable field repairs, depot repairs, and what needs to be sent back to the manufacturer. Many pieces of equipment require specialized skills and tools to fix. They also are frequently subject to strict environmental requirements that necessitate good seals, shock mounts, and shielding. All of these things are easily damaged if handled improperly and would no longer meet requirements. While I appreciate the intelligence and skills of our service members, training them to properly fix every piece of gear is not cost effective or reasonable. I know it's frustrating when equipment isn't fixed overnight, but it feels like the manufacturers are getting blamed for the whole problem, including the contracting delays, shipping delays, parts availability, and myriad other issues that slow down the process.

Everyone seems to forget we've had fully autonomous lethal weapons for decades - they're called land mines and sea mines. All we've done is make them smarter and more mobile. Making them smarter should only help reduce the number of false-positive casualties. To some extent, the same basic rules apply to robots as minefields - the person culpable is the one who deploys them. We've just got more control now than we did before.

Yeah, that would be a terrible commute - >12 miles by ferry each way, every day. Admittedly, there's often less traffic to deal with, but you don't tend to get seasick on the Interstate on a regular basis. It's probably not worth doing until you can virtualize most of your workforce and make them telecommute. Even then, it leaves you reliant on a network and power bridge to the mainland, which is not cheap. It does fix your cooling problem somewhat - the ocean is a decent heat-sink.

An alternate strategy - if you choose to opt out, you are then financially liable for all measles cases in your state. Take the costs from all those cases, divide by the # of children on your "Idiot Parents" list, and send each parent a bill. Do this every year until they wise up. This prevents people from externalizing the costs of their poor decisions on everyone else.

This math error basically doesn't matter. 1.3 trillion seems like a lot for a mistake, but that's like arguing with the magician who's just accidentally dropped a card out of his sleeve. Does it really matter whether it was a card from the deck or an extra copy of a card? It's still a magic trick. The whole proposal is built on poor estimates and pipe dreams with little or no supporting evidence. Admittedly, it's still a pretty big math error. I wonder if someone forgot to renew the license on the spreadsheet software at the White House and they had to break out the pencils and slide rules?

For example, see the article http://arstechnica.com/cars/20...

This is already being solved - the Defense industry is working on autonomous offroad driving which handles all the things you've mentioned. Plus, some of the newer farm machinery can already be programmed to drive waypoints by GPS, so you can easily see combining the two in the near future.

Found the other relevant paper - "Model-assisted pattern search methods for optimizing expensive computer simulations" by Seifert, Torczon, and Trosset at The College of William and Mary in VA

Yep - it looks like they rediscovered known methods for optimization using a series of cheap approximations of the black-box. You find the optimal solution of the approximation and use that to choose one or more points to evaluate on the original black-box. Then you update your approximation with the results and repeat. I was doing this in college back in 1999. I wrote to generate pseudo-random functions to test on, and we used the same class of functions as approximations for the harder, more expensive problem.

Wrong - the CIA keeps its own records separate, for exactly this reason. FBI, DoD, and Contractors however were screwed by this.

The group that's complaining doesn't realize that we've deployed "killer robots" for a century now - they're called mines and especially naval mines. They may not be your traditional humanoid Robbie the Robot with a gun, but they are fully autonomous, capable of selecting targets on their own, and definitely capable of killing people. Many of the arguments they make in the article are bogus anyway. If you took the text and substituted "people" for "robot" it would read just as well and make just as much sense. The authors act like people don't routinely mess up, make bad decisions, follow charismatic, insane evil overlords, or do generally nasty things to each other for very little reason all the time.

There is a lot of strategy. However, since the battle happens in 3D and there's no real way to maintain formations, you tend to end up with these blobs of friendly ships and enemy ships. The strategy is in maintaining the proper range to friends and enemies. Weapons have different ranges and tracking speeds. Similarl, the repair ships (think healers) have limited range as well (roughly 50 km for the largest). In these big fights, a lot of the work is in choosing targets, trying to do enough damage to destroy the target before his repairs kick in. There's also a fair amount of complication from the electronic warfare possibilities - jamming and such.

I work in the defense industry, writing software for both fielded systems and the modeling and simulation used to define and design such systems. I constantly use algebra, trigonometry, FFTs, logarithms, complex mathematics, statistics, probability, and a bit of calculus underlying many of the models. I do a fair amount of data analysis, which has me thinking about the various ways to condense and display data. My coworkers who can't understand the math are relegated to building UIs and writing messaging interfaces. Those of us who can understand the math create all the underlying code doing the real work. The former are much more likely to be viewed as replaceable and laid off in the bad times.