Monday, June 21, 2010

Science Fiction pisses me off

Science fiction is one of those genres where nobody ever seems to get it right. We play EVE or watch Star Wars or whatever and we don't think much of it, but goddamn I hate science fiction sometimes.

It's not that I don't like spaceships or laser swords. I really think those things are cool. It's just that there is always some colossal science blunder that makes sci-fi totally unbelievable. You guys never notice this stuff, or maybe you do, but don't think much of it. This stuff makes me ARGH though. I'm like "WHY DON'T YOU JUST USE BULLETS" and "WOULDN'T A COMPUTER MAKE THAT SO EASY?"

Transportation and communication don't generally bother me. It's usually done horribly wrong, but generally people don't put a lot of detail into it. For instance, in Macross, we never learn how "reflex" thrusters or space fold systems actually work. The technology is "magical" from the perspective of the characters, and isn't really that important for telling a romance. Also, it's pretty easy to use something like plasma thrusters and fusion reactors or solar sails for your impulse drive and some barely quantum theory transportation for your superluminal transport. Quantum theory is nice since we can't really understand it yet, so it makes as much sense as anything else.

Communications is similar, and besides, most people understand that we need "magic" communication to send messages at FTL speeds and thus make up some type of technology that's so far out of the range of real physics that nerds like me just shrug our shoulders and surmise that it could probably work. Once in a while you get stupid things like 40k and its "super fiber optics" (I shit you not) that can transmit signals at significantly faster than fiber optic speeds through a physical medium. Yeah, that only works if real life physics don't exist, and unfortunately, the writers of 40k seem to understand real things like the principles of bullet drop enough to make retarded ideas like self-propelled bullets to solve it.

So let's get into the real things that piss me off. Super fiber optics and gravity drives (making a black hole in front of you so you can fall into it is simply not a good idea) aside, transportation and communication do a good job of making up "magic" technologies that exist outside the realm of understood physics. However, other things don't.

Computers. Agh! I hate sci-fi computers. It's like when writers produce something that's sci-fi, there are 0 nerds involved. Computers are always screwed up in sci-fi, because people don't understand that computers aren't very good at some things, and are incredibly good at other things.

First, the Terminator. The idea of SkyNet is absolutely ridiculous. No matter how much processing power a computer might have, the computer's ability to be self-aware is just impossible. I guess it's not "no matter how much," but it's fairly clear that SkyNet is not that smart, because if it was, there would be no possible human resistance. There really isn't another option. If a computer can think abstractly, it could also think logically insanely better than anything ever possibly could. I'm not communicating this well, so I'll explain.

Computers suck at games. Take DarkSims from Perfect Dark. DarkSims cheat extremely badly - spawning automatically with high powered weapons and armor, where a player would have to hunt for them. DarkSims also aim perfectly and rarely miss (except when their weapon would miss through mechanical failure), often headshotting their opponent with ease. The fact that DarkSims are beatable is testament to how bad computers are at strategy. When a computer tackles a strategy problem, it has to consider all possible options and all possible responses to those options in order to accurately present a challenge. Obviously, computers in video games rely simply on following a basic algorithm and doing it as best they can. DarkSims can aim really well, so a human can rarely fight them straight-up. However, they can be baited into the simplest of traps, because they don't take enemy behavior into account.

Computers are simply not capable of strategy. Sometimes they can perform decently if they are put into a game like Chess. Chess is a very deep strategy game, but each board position might only have a few dozen possible moves (more as the game gets later, but still...). Chess computers are better than humans now, but they are not unbeatable - while Chess is still a solvable game, a computer's solving heuristic still cannot guarantee a 100% success rate against humans' abstract problem solving. This is in a game that is turn-based and has a small, finite number of decisions possible at each step.

Put a computer in a military situation where moving 13 feet before shooting is logically different than moving 12.9 feet before shooting and remember that the computer also has to make decisions in real-time, and add allies (even if they're other computers that can communicate) to the mix, and suddenly you've got a huge mess. People estimate that computers get a little better at chess every time they get twice the computing power. It's theoretically possible that computers could get smart enough to make tactical decisions, but I can only imagine the computing power needed for a computer to determine that a grenade has been thrown in its direction and the proper course of action is to run the hell away. I think terahertz aren't even in the ballpark.

Humans are incredible at strategic thinking compared to computers. We can make abstract guesses about what our opponent might do, then implement counters. We aren't perfect at this, especially compared to other humans who can predict our counters. However, being 50% or 80% effective at predicting a tactical situation is a lot better than 0%, which is about how good video game computers are right now at predicting player behavior, especially in real time.

As a side note, I'd like to point out how much I hate EVE for this reason. Combat in EVE takes away the single most important human factor - tactical positioning - and lets computers take charge. I think that no matter how advanced spaceships get, that we will be navigating them manually in combat, because little nuances of positioning can never be performed by a computer.

However, one thing computers are really good at is mechanical things. Humans suck at things like hitting the right keys when we're writing or executing combos or aiming at things. Humans are absolutely fail at things that improve with mechanical precision, while computers ROCK at it. We can train ourselves to be better but computers are perfect at it by default. Unless the task in question takes too much computing power to perform quickly, a computer will calculate and perform it flawlessly every time. It probably has to be told exactly what to do - for instance, a computer aiming a gun turret would have to be programmed how to lead targets based on the speed of the target and the velocity of the weapon's projectile - but once told, the computer will do that job flawlessly.

It's worth pointing out that aiming hitting a fast moving target with a comparatively slow unguided projectile is better done by humans, since erratic movement by the target will cause mechanical leading to always fail. However, the computer can still tell its human operator an ideal lead, giving the human information to better make critical predictions.

A good example is using artillery. Human operators might aim the gun in some improper way, or not account for some factor such as wind. A computer can make calculations so that, to the best of its ability, it will be able to put that artillery strike directly at the designated position.

Another good example is piloting from two destinations. A computer can automatically predict the optimal path using weather pattern data and such or in the case of ground vehicles, by using known roads. Actually, this is another good example - in Grand Theft Auto 4, our GPS will give our car the fastest route on the roads, which is nice - but our human ingenuity can think of alternate routes off the road or driving the wrong way on one-way streets.

Too often though, computers are retarded. I'm talking about Star Wars droids, where they have illogical emotions and fail at doing computer-optimized things. Also, at the end of Episode 4 when people are doing the bombing run, why are they using targeting computers if the computers can't hit the target? It should be common sense that if one fighter's targeting computer doesn't work, all of the fighters computers will probably be insufficient, and human ingenuity will be needed. I mean, in the end this makes sense, but when Luke is asked why his computer is off, it should be a no-brainer. If other people can't use the computer to hit the target (bad programming), a human is better off making the guesswork. 50% is better than 0%.

That was a loooong rant about computers. This isn't over yet though, I'm still pissed at sci-fi.

LASER GUNS. WHY?! No, I'm lying. Laser guns somewhat make sense, but in real life military scenarios they just don't. It's possible for a personal laser gun to someday in the future shoot a beam that could be dangerous or deadly. However, modern Kevlar vests are extremely effective against a beam weapon (of any type except particle beam). Modern kevlar is by its nature heat resistant, and this is really true of any ceramic or polymer armor. The armor will tend to heat and flake off, rather than applying heat to the rest of the armor. This limits a laser weapon's use against modern armor, and really I don't think that armor is going to become less effective against lasers or plasma guns in the future. Beam weapons are still probably decent against stuff made of metal, and probably really good against fleshy stuff like people. However, modern Chobham tank armor has ceramic plating that is extremely good at absorbing and dispersing heat so lasers wouldn't be very good against tanks.

Against aircraft, lasers have many advantages, since they travel at around the speed of light, nullifying an aircraft's speed advantage (especially with computers to aim!). The problem is that the development of laser weaponry is likely to cause aircraft manufacturers to develop anti-laser measures, including reflective or heat-resistant materials. This would counter stealth technology (which relies on absorbing radiation such as radar rather than deflecting it; a reflective stealth plane would not be very stealthy!) to a degree, but making a gun simply to shoot down stealth planes seems counterproductive.

Also, laser guns have a much more limited range than a typical projectile; rather than cut through the air, energy weapons disperse, or "bloom," losing coherency.

But the worst issue with rayguns such as lasers and plasma guns is that it simply takes too much electricity to make a powerful laser. The same amount of electricity used to make a laser that could melt metal could instead be used to throw a slug at very high velocities - and that slug would be devastatingly effective against any type of defense. Armor is simply worse at stopping projectiles, and until electromagnetic fields are created that repel metal bullets, no protective method will be better at stopping bullets than it will be at stopping ray guns. Even then, lead bullets with a brass jacket wouldn't be deflected by an electromagnetic field, as neither are ferrous metals. Also, neither are depleted uranium or tungsten.

On the same note, electricity is not very good for weapons in general. It takes a lot of electricity to throw a bullet out of a coil gun, and when you compare it to the energy in exploding propellant, electricity just isn't efficient. When we get to the point where capacitors can store electricity to fire a railgun in a vehicle-portable environment, that will be sort of scary, because that same amount of electricity can power a small city. Even still, if some super-tech is developed that makes energy mostly a non-issue, technology is always better throwing bullets with that energy than it is shooting death rays.

Don't believe me? Well, if we had a hypothetical long range death ray laser capable of zapping tanks, people could use that same amount of electricity to throw small particles at sublight speeds instead. The power of a few grams of energy flying at a tenth of the speed of light is absolutely incredible (nuclear weapon incredible). Granted, there's recoil issues obviously, but in terms of destructive power... yeah.


  1. I totally agree with your last paragraph :D The Japanese actually took the building of an actual death ray the farthest, but they halted after they realized they needed more than all the electricity the country could possibly produce. We all found better ways to use smaller amounts of electricity.

    It sounds like the levels and balance of suspension-of-disbelief erks you most(?) I think that usually comes from someone very schooled in a certain field, or someone generally above average in intelligence, combined with something they like.

    You know the guy who invented the original circuits that lead to computers since the 40's or 50's? Well, he conceived the idea of another circuit. He knew it could exist, but he never figured it out. Now, scientists have figured it out. it is a brand new major key component that does away with the way we've all come to know computer's to process information. No numbers anymore. Wired spoke about it and they tried to simplify it and say that with this new missing circuit, computers will be processing closer to human cognitive abilities. Instead of numbers, it's like billions and billions of different shades of gray.

  2. That's actually neat. I'm not sure that it'll actually be able to compute in an illogical manner, but it is an interesting thought for data storage and such, since it can remember exactly what kind of electrical current went through it first. I guess it's one of those, "we'll have to see" sort of things. It's kind of a scary technology to be honest. I hope that computers can never do things like fight wars.

  3. Reading the description, the closest practical use I can see for this new memristor is dramatically increasing available memory storage. If each 'bit' was instead an infinitely large possible set of numbers, we could be talking about increasing storage 8-16 times or much more. If it's faster too, I could see how this could lead to better brain simulation.

    As a sidenote, modeling the human brain is actually already happening.

    We're not quite there, and a supercomputer is pretty much required, but we're headed there even without this technology leap. I imagine what memristors will do is make it so that everyone can have one in the future. Which would be... interesting.