Sunday, January 31, 2010

Exploiting Multiple Universe Time Travel

Let's say you're in a nondescript room with a box that can move things backwards in time. Every hour, on the hour, it moves whatever is in it to fifteen minutes ago, but offset by a few meters so that in ends up in a different box (the boxes are labeled). You have a thousand dollars.

In the stable-time-loop model of time travel, it is impossible for you to make money in this scenario. It's also impossible for you to make any changes at all, and is therefor only interesting insofar as it allows audiences to make observations about free will.  This is the model used by The Time Traveller's Wife, Terminator, and Twelve Monkeys.

(Edit: I came across this post while searching Google today, and I feel the need to clarify that statement about stable time loops; it's very possible to make money if you have one, because you can look up stocks or lotto numbers or use what's referred to as Time Loop Logic. However, in the scenario presented, it is indeed impossible.)

In the multiverse model, each instance of travelling back in time spawns a new universe (a brief note on terminology: when I say universe, I actually mean cosmos, because the is universe by definition everything. When I say multiverse, I mean the collection of all cosmoses.) Given what you have in the world, it's technically possible to make money, but you would never do it.

Here is the stupid plan: put the money in the time travel box (from now on, T-box), and when it hits the hour, it'll be sent fifteen minutes backwards to the recipient box (R-box). But then, when the money comes into the R-box, you run back over to the T-box and pull that money out before it can be sent back, leaving you with twice as much money. This would seem to create a paradox, but since we're actually dealing with different universes, there's you-A in universe-A with no money, and you-B in universe-B with double the money.

First, this plan is stupid because it redistributes the money in a stupid way. Marginal utility says that you-A will be a lot less happy with not having the money than you-B will be with having twice the money. This is why you should never make a 50/50 bet at 2:1 payoff.

Second, it won't work because there's no incentive for you-A. He'll see that there's not any money in the R-box, so he'll pull his money out of the T-box before it disappears forever. Which means that there's never going to be a universe-B, or (depending on how the T-box works) universe-B won't be much different from universe-A.

There are a few ways to circumvent this; the first, if you're comfortable with yourself, is to just cilmb into the T-box and wind up in universe-B R-box. Of course, making a profit this way is difficult, because you'd only be able to take as much stuff with you as would fit in the box, and you would have to share an identity with the other you. The various benefits include being able to halve your rent, having someone to play co-op games on the Xbox with, and being able to work phenomenol hours in comparison with the rest of the workforce.

The second circumvention method is to send information back in time. Since information is practically free (for small amounts, you only need a pencil and paper, and you get to keep the pencil when you're done), it's something that you would do without any incentive - unlike the example wherein you waste $1000, it has some chance of actually happening. The real problem is finding some bit of information that's worth more 15 minutes ago than it is at the time the T-box activates.

And here's where we have to take the thought experiment outside the idyllic realm of nondescript rooms and boxes. The first thought that comes to mind is lottery tickets. Fifteen minutes isn't a long enough time, because a play on any of the numbered lottos rolls over to the next week more than fifteen minutes prior to the drawing (generally one to two hours before the drawing, depending on what state you live in). However, a similar effect can be achieved with the use of scratch-off games. Your course of events would look something like this:

1) Wait by the R-box to see if any tickets come back. If they do, go to step 2b.
2a) Run to the nearest gas station and buy some scratch-off tickets (if you're too far away from one, you can coordinate this by phone with another person).
3a) Record all the numbers on the cards.
4a) Have the employee run them through to see if you've won (this is faster than actually using a penny and checking for yourself).
5a) Make a note of the winning numbers.
6a) Put the sheet of paper with your records in the T-box.

2b) Take the sheet of results to the nearest gas station.
3b) Only buy the winning tickets (most clerks, if it's not busy, will let you buy scratch cards out of sequence, especially if you act really superstitious)
4b) Profit!

Of course this plan requires some initial investment money, because you have to actually play the cards to know what they contain. However, you've given yourself a huge edge, because for each time you use this system there's a 50% chance that you'll be able to skip the whole "losing" part of gambling. However, most of us would be willing to spend $3 on a making hundreds of dollars.

But we can actually increase those odds considerably. At the point you hit 4b, you should have two things: a sheet that tells you which tickets to buy and some amount of cash. One of those things is now worthless to you. Here is our replacement step that we'll do instead of screaming "Profit!":

4b) Put the sheet of paper with your records in the T-box.

What happens now is that instead of two universes, one in which you gamble and one in which you win without gambling, there are a nearly infinite number of universes. In one of those, you gamble, while in the other infinitude of universes you win without gambling. This means that for any time you try this system, it's a freak occurrence for you to actually have to gamble, because all the versions of you on the b-track outnumber the single version of you on the a-track.

(Note that step 4b should be moved down to 5b, and the new step 4b should be something like "If the piece of paper is getting unreadable, write all the numbers down on a new sheet of paper and put that one in the T-box". You'll note that if those steps were followed as is, the piece of paper would still be aging with every cycle. A million cycles of 15 minutes is about 30 years. By copying the information onto a new sheet of paper, you can ensure that information is the only thing traveling. Of course, then you get into the problem of transcription errors, but whether that would actually happen depends on what you believe about the universe as far as chaos theory and quantum mechanics goes - whether the physical laws mean that things are predetermined, and how quickly events diverge.)

So using a single player, with a highly restrictive time machine, the best method of exploit is to send information back to yourself, especially if there's minimal cost associated with that information. Time to expand the scope. Let's say someone else finds this room, works out the principles of how this mysterious box works, and turns it into just another piece of consumer electronics, which can translate anything of any size to another time and place.

Lotteries would immediately stop, as would most other games of chance, because the house edge would absolutely evaporate. The stock market would most likely stop serving any useful function, because information from the future would (in almost all realities) be flooding in. A large number of information services would fold, because it's only necessary to pay for an application once, at which point you simply send the source code back to before you paid for someone to create it. Crime would be stopped before it happened, because the date and time of nearly every murder, theft, arson, etc. would be known beforehand. There would also be a mass exodus to the past, where information about the future is even more valuable than in the present, and the further back you go the easier it is to fake an identity.

Of course, this assumes a reality which has experienced more instances of time travel than the initial reality. Arranged left to right, we could order these universes in order of causality, with the leftmost universe having been the root cause of time travel and having never actually experienced anything coming from the future, and the rightmost universe being one in which nothing has ever actually traveled to the past because every attempt has been interrupted by people from the future.

The impossibility of time travel has often been disputed with the simple fact that if it were possible, we would be knee deep in people from the future. However, in this model, we can clearly see that this doesn't prove anything; it merely means that we find ourselves in a universe further to the left.

Tuesday, January 26, 2010

The Wierd Inversions of Singularity Fiction

Alright, first a definition of terms: the Singularity is sort of the Geekpocalypse. Only instead of dying, or going to heaven, we all become super intelligent and transcend into almost unimaginable beings of immense power. Supposedly, this will happen because of a large number of advances in technology - brain uploading, artificial intelligence, nanotech, bioengineering, etc. It will be incomprehensible to us in the same way that the Pythagorean Theorm is incomprehensible to a dog.

Personally, I'm sort of a soft believer in the Singularity. This whole concept of exponential growth is all well and good, but it requires a little too much faith for my tastes. The truth is, there's so much we don't know about consciousness. There are also hard limits on technological progress, which means that what looks like an exponential curve must actually be an S-curve. There's also a question of economics; it might be that AI is possible, but that we'll never reach the point where it's actually pursued. Or AI will take a lot of time to make better AI, so instead of an intelligence spike, it's a gradual upwards slide, and there's no future shock, just slow and steady progress. Or we might blow ourselves up, or get hit by an asteroid, or exterminated by extraterrestrials. All that aside, we do seem on track to be hitting some serious boosts in technology if the next hundred years are anything like the last hundred.

Anyway, I've been reading a lot of science fiction recently, much of it focused on the Singularity. There's a weird theme running through these books; as technology progresses, civilization starts to regress.

For some reason, governments start to melt away or dissolve completely. Instead, you have the ungovs of Marooned in Realtime, the phyles of The Diamond Age, and the synthetic groupings of Accelerando. The reasons for this breakdown varies. Either the government is too big and slow, or the corporations gradually take over, or the system of taxation becomes infeasible. Some authors take it farther than others - sometimes the future is only of companies and their shifting alliances, and sometimes it's nothing more than independent agents, connected to other people only through tenuous agreements. Society becomes tribal once again, only this time humans - and our descendants - don't move through the plains or forests, but through the sea of humanity.

Another throwback is currency. Historically, we moved from barter, to backed currencies, to fiat currencies. In Singularity fiction, this also warrants a regression; typically this comes in the form of reputation markets, where people spend goodwill (basically, their upvotes). That system seems really dumb to me, primarily because it needs the same sort of belief supporting it that fiat currency does. Only instead of being under the control of government, that wonderful institution whose primary occupation is existing, it would have to be under control of a private corporation with a risk of folding. The other two dominant narratives are an information currency (in Accelerando they use computerized people) or a return to barter (on the theory that computerization can eliminate all the costs associated with that). Barter makes a lot of sense if you've already postulated that civilization is doomed - you have to trade in things with actual value, and anything that's useful can't very well be put into wide circulation, and if you're using something without intrinsic value as your currency, then you're basically using a currency founded on belief anyway.

Okay, so why do the writers choose to do it this way? First and foremost, I should point out that a writer can only be so clever. That's why if you read (bad) detective fiction, the detective solves easy problems and everyone pretends that they're hard. Most hard scifi writers aren't geniuses; they read scientific papers, look at trends, and then try to write an interesting story that's within the bounds of reality. So perhaps the reason that systems seem to be regressing is that we don't really know what form civilization might take. A monarchy is at least easy to understand.

I guess the other reason this happens is that it provides a nice sort of mirroring effect; humanity moves from tribes to kingdoms to nation-states, then back downwards until we are tribes once more. It has a certain irony to it.

Tuesday, January 12, 2010

The Future Will Be Customized

The archaic model of production was artisan based. If you wanted shoes, you went to a cobbler, who would take measurements of your feet, and a few days later you would come back to pick up your order. If you were poor, you would buy shoes secondhand, or just go barefoot.

The old model is mass production, of the "any color so long as it's black" variety. Eventually this becomes something more like "any color so long as it's black, blue, or green", which gives enough variety for possessions to be relatively distinct, especially since the added cost of something like different colors is trivial. The stuff that actually has to be engineered requires larger costs, and so won't be made unless it has a large enough market.

Eventually a variety of technologies come together to vastly expand available markets and greatly lower the cost to enter those markets, which leads to the current model - long-tail distribution. iTunes can sell not just the big hits (like a physical music store would have to) but really obscure songs that only a hundred people would want to buy. They can do this because the cost of having a song in the store (because of smart searching and digital hierarchies) is practically nothing. In the same way, anyone can start up a website and start selling something, with a cost of about $50, and the search and delivery infrastructures mean that it can actually be worth their while. This leads to things like etsy.com, where people can (for virtually no cost) put their wares online and sell them. Instead of profit being located exclusively at the head of our market size graph, it exists further and further down the tail. Note that this model works better for things that don't require huge amounts of capital to produce, like works of art or small consumer goods. Something like a car still requires a factory, so you're less likely to be able to buy something exclusively suited to you.

The long-tail distribution of goods represents the direction for the future. Production costs for nearly everything are falling, due to improvements in technology. There are two technologies coming down the pipeline, whether it be ten years from now or a hundred, that will drop those costs to practically zero. Those two are artificial intelligence and nanotechnology.

Artificial intelligence will allow the dynamic generation of pure information commodities; books, movies, music, software, and so on. Once those abilities are up to the point of being able to match human quality a large number of things will happen, but one of the biggest will be the near-infinite extension of the long tail. Right now it's not uncommon to find books that only a handful of people read. But with hard AI, there will be books only read by one person; custom crafted to that person's tastes. In the same way that Amazon takes in all of your reviews, rating, and prior purchases to suggest things that you might like, an AI would be able to take in all of that information to create an entirely new work. Alternately, large commercial works would be able to build variability into their pieces - a comedic movie would be able to respond to the audience's laughter in the way a stage production can, for example. This complete customization applies not just to information goods, but services as well. Instead of sitting in a classroom with thirty other students, you would be able to sit in front of your media center and learn from an AI program which customizes itself to your individual needs and learning style.

Nanotechnology means pretty much the same thing for physical goods. In the science fiction version, this means a powder that transforms nearly anything into nearly anything else, usually by rearranging protons, neutrons, and electrons. That's a long way away; in the short run, nanotechnology means that nearly anything will be able to be produced as a one-of. We can do some of this now with 3D printers, assuming that you want to make something out of plastic and at a fairly low resolution. As time goes on, the resolution will keep getting better, and mixed materials will become available. Since information is the only important input, the same rules that will apply to things like movies and books will start to apply to physical things like cars, dinner plates, and so one. Instead of buying a car from a dealership, you'll go to a dealership (or a large rapid production facility), have an AI figure out what your needs and desires are, and custom make a vehicle for you. The same thing will happen to larger things like houses, as nearly all aspects of home construction will be outsourced to cheap AI and robotics. Instead of finding a house suited to you, one will be custom built.

In the future, when this technology comes around - whether that's twenty or a hundred years away - nearly everything will be made on a case by case basis. Everything that you wear, watch, and use will be custom made. The complications that arise from this are numerous; there's already been talk that with the availability of the internet, people are segregating themselves into diverse groups, reducing our ability to get along as a society. When we no longer have even our mass produced goods in common, and a culture of one, what happens then?