this is a long post for someone with nothing to talk about

Believe it or not, my little BitTorrent dissertation was actually meant to be the beginning of an even longer entry. What can I say? I'm not so good at self-editing. Anyway, there was a logical end point to the last one -- namely, the realm of fact. From here on out it's just a high-concept misapplication of scientific principles.

During my, er, research for this post I discovered that suprnova.org (sic) wasn't the cutting-edge BitTorrent clearinghouse I thought. In fact, it kind of sucks.

You might recall me rambling about .torrent files and trackers in my last post. Well, those files need to be hosted somewhere. Suprnova is one of those places. But there are others -- including many that, unlike Suprnova, require a user account.

The specific one I ended up using is called EliteTorrents. The gimmick there is that your upload/download ratio is recorded on the server. When you download a torrent it's custom-tailored to your account, allowing your individual activity to be tracked. Download more than you upload and you'll start having to wait to join the torrent; keep doing it and you might get banned. It's an extra layer of quality-of-service assurance on top of the QoS stuff already built into BitTorrent. There are a lot of other sites with similar schemes.

This is nothing new. Some of this blog's readers might remember logging into the dialup BBS I ran in middle and high school. The file section there had an enforced upload/download ratio too, to ensure that you contributed roughly as many badly faked EGA pictures of topless celebrities as your friends did.

This all got me thinking about the appropriately derided hacker credo (or motto, or whatever): "information wants to be free". Well of course it doesn't really want anything (barring arguments for panpsychism, which even I know are beyond the scope of this entry). But it does seem to try to get free, in the same way that a drop of dye in a glass of water tries to diffuse until it's uniformly distributed.

The reason for that dye spreading out is the second law of thermodynamics, aka entropy. It's a nerd greatest-hit: "the unusable energy of a closed system will always increase". Put in a somewhat less formal way: things spread out. On ignition, chemical energy in a molecule of propane spreads out into lower-energy bonds in water and CO₂, plus surplus heat. Electrical energy spreads out from places of high concentration, through whatever appliances you put in its way, and into the earth where its charge can be most evenly distributed. The mechanical energy of a football flying through the air is transferred to the air and the person catching it; eventually it ends up as friction, diffusing that mechanical energy down to the level of random molecular motion. In fact, theoretical physicists think that as the universe winds down, protons themselves will fall apart. Space will be filled with nothing but a soup of elementary partles and heat -- enough of it, anyway, to keep things a fraction of a degree above absolute zero. Nature seems to want very badly to turn itself into a big, messy, undifferentiated soup.

The practical interpretation of entropy is that systems tend to disorder. Jeff is fond of explaining entropy as the reason Nintendo controller cables get tangled. That isn't entropy, strictly speaking -- the energy of the system is probably actually higher when the cords are twisted. But, like entropy, the tendency to tangle is an artifact of the fact that very few potential outcomes of a probabilistic circumstance correspond with what humans think of as "order". The odds of the controller cords ending up in an untangled configuration are almost as bad as the odds of those dye particles randomly arranging themselves into a drop-sized space inside a larger volume of water.

My point is that there are entropic-looking processes that don't have anything directly to do with entropy, but which echo the principle, and can be thought of in similar terms in an instructive way. We can sometimes employ entropy as useful shorthand even at scales or in situations where it may not be strictly applicable.

Is there an actual correlary for information? Probably not in a scientifically rigorous way. The type of information diffusion I'm talking about involves human actors, which obviously muddies everything up. Constructing parallels between physics and sociology is an undertaking that’s unlikely to produce much testable theory.

But there are some connecting strands here. I’m playing pretty fast and loose with the term “information” here (and entropy, for that matter), but Claude Shannon and his intellectual descendents did a much better job – down to the level of defining just how many bits you can get per photon. (If you’re interested, I highly recommend this book as an accessible introduction to information theory.)

There are other explicit ties between information and the physical world. As processors become more powerful, they’re also getting hotter in terms of watts per square centimeter¹. In a couple of years, don’t be surprised to find that your Dell desktop PC comes equipped with a water-based cooling system.

One solution to the problem is something called reversible computing. It turns out that a lot of processor heat comes from the destruction of information. If you do your computing in a way that can be reversed – that is, one can take the result and work backward to the input – you waste much less energy. It’s a hard thing to understand, and an even harder thing to design for; I don’t mean to imply that I can do either. But it’s the weird truth. The intersection between physics and information is real.

So I think that information's tendency to diffuse may be usefully thought of in entropic terms. And from that premise, I think that the following two observations can be made.

First, I find it interesting that when enough order is injected into either type of system -- usable energy, or unique information -- broadly "anti-entropic" mechanisms arise. It would be much simpler for sunlight to diffuse its energy into heat directly, but life arose, keeping the energy ordered for many more phase changes, and slowing the entropic process to a relative trickle. Similarly, it would be a lot simpler for file transfers to be direct, leech-style transactions or simple broadcasts. Instead, distribution mechanisms have arisen, in parallel, that ensure the orderly dispersal of the data. Perhaps both of these are actually disorder-maximizing processes – in the absence of plant life, a lot of sunlight would radiate off into space, still in a usable form. And the point of EliteTorrents and its ilk is to ensure a steady, optimal flow of data, not to keep the data closer to its original, non-distributed state. Still – that’s an awful lot of apparent order that has spontaneously arisen.

Second, when you have a sufficiently large natural imbalance, you can use it to drive an engine. High charge can run a vacuum cleaner as it seeks ground; compressed, energetic gas can push a piston head; even the difference between hot and cold water can be used to do useful things as it seeks to equalize temperature. In the same way, the dispersal of information might be harnessed to power what I’ll reluctantly call a social engine. You could make the argument that publishing houses are such engines; or that EliteTorrents is. But I’m not convinced these efforts rise much above the comparative level of hot air balloons in terms of technological sophistication. It may be that such processes can’t be effectively harnessed. It may be that social engines self-organize into webs of vast complexity, but generate only informational waste heat and no useful work.

Either way, I think it’s at least an interesting paradigm to apply to the internet, piracy, and Shawn Fanning’s various failed business ventures.

¹ Yes, Jeff, I know. Don’t look at me; I didn’t choose the units used for this.