Wow does that bring back memories. Back in the 60's I ordered a book of recipes for rocket fuel and explosives from an ad in the back of Popular Science (imagine trying to do that today!). Well the formula for ammonium nitrogen triiodide was one of them and one of my favorite explosives in my wayward youth. Crystal iodine was easy to get at the time as was the concentrated ammonia you need (household ammonia might work, but results would not be nearly as good). Fortunately I didn't get hurt or hurt anyone (unlike one the the disastrous recipes for a molten rocket fuel that did blow up in my face). We'd stick the AMT in emptied gel capsules with dry tapioca and let it dry in the fridge (stable while wet, and cold helps improve stability). Then you had little "firecrackers" you could toss and would make a nice little pop when they hit the ground. We did have some accidents with stuff blowing up in the fridge that had been packed with too much pressure. My poor Mom. As I say, I'm lucky I survived my childhood, but it did launch me on the start of a chemistry career.

You miss the point. if the insurgents capture the video they know (from the metadata) exactly where we are looking and therefore where we are not. Based on that information they can decide when to do what they want and when they need to be covert. Other commenters are correct. We've known about this problem for a long time and it should have been fixed. But its now gotten very complicated because we have hundreds of systems in theater that expect to receive that video feed unencrypted and ALL of those systems need to be upgraded before you can begin encrypting.

Correct: just the video feeds are unencrypted. However, those streams include metadata such as the GPS coordinates of the current view.

Ender is old tech: human (even if child) commanders (remotely) controlling human armies. The new tech is semi-autonomous: humans remotely providing guidance to semi-autonomous swarms of unmanned systems. I think Ender's Game would have made a lot more sense if they had been remotely providing commands to teams of manned and unmanned systems. The lack of robotics in Orson Scott Card's Sci-Fi series is puzzling. BTW, Ender's Game was "required reading" for a DARPA program I was on at one time.

To answer this question adequately would take a much longer lesson in systems theory than I have time for. You touch on some of the aspects of why multiple systems are better than a single system. Swarms don't just "hold formation" they self-organize in the most efficient manner to accomplish complex functions with whatever resources are available. But there is plenty of literature about why cooperating systems are always greater than the sum of its parts. May I suggest a good introductory video lecture by Scott Page on the Logic of Diversity as a taste of what is possible: http://fora.tv/2007/03/14/Power_of_Diversity

I'm afraid the Terminator hype actually detracts from an important point that you are making. If you knew anything about the state of computer science today you would feel quite secure in knowing that "step 2" ain't going to happen for a very long time (step 3 always happens even today :-)

Still there are a number of very important people who are concerned about how much autonomy we might eventually give to a robotic entity with lethal capabilities and there is legitimate scientific discussions happening today. Asimov's Robotic Laws seemed to make a lot of sense before we had armed Reapers roaming the skies. Since I work in the defense industry I am less concerned about what our own government will do in this regard than I am with rogue nations and terrorists groups. People who don't work in the defense industry have no clue how conservative the military is when it comes to giving up human control! Although reading all the FUD posts on this article give me a chance to chuckle, I recognize that this technology can be used for causing harm. The swarming technology described is not easy to replicate, but is ultimately based on simple, cheap platforms running what are basically simple programs. Thus while UBL is unlikely to ever replicate even a relatively simple Predator, he COULD develop unmanned swarming capability.

Personally, I feel that it is nearly inevitable that this technology will fall in the hands of the wrong people. When that day happens, I would much rather have a military that has experimented with, understands, and has prepared defenses against such a technology. That's why we need to approach step one and learn.

It takes three people to remotely pilot a Predator. There are never enough Predators or Global Hawks in the sky for all the intelligence we would like to gather. We don't have enough people, platforms and dollars to buy, launch, pilot, and support all the reconnaissance we would like. And while the imaging capabilities on the big unmanned platforms is impressive, they still can't see through mountain ridges or down deep urban canyons. For that you need something that can fly right overhead and get close enough without being seen or heard and that requires lots of small UASs. But the only way we can get enough of those into the air is to have some way for a single person to manage two or a hundred platforms just as easily as one.

Swarm may be an unfortunate term, since it can evoke the image of a killer swarm of bees - hence we naturally think of swarms as lethal attack technology. In fact, unmanned attack swarms are still science fiction. The swarming research that is going on (and demonstrated in the article) is all about surveillance and reconnaissance. Even if we get to the point of arming the individual swarming platforms, there will always be a human in the loop making the final decision to fire a weapon. Don't kid yourself: even with all the new technology it has only gotten more difficult to make the decision to engage not easier over time. Ask those that do this for a living about the hoops they have to run through before they can fire a weapon from a Reaper.

No, its not based on Particle Swarm Optimization (PSO). See the linked article in Defense Industry Daily. The swarm is controlled by digital pheromones, a very different technology from PSO. Nor is it centralized. In fact the algorithm is completely decentralized: no central computation, no global knowledge. Of course, the operator provides the guidance, objectives, rules of engagement, constraints and the swarm figures out the rest.