The Phases of Titan

Note: This is a very old post! Most of the links are broken!

Of all the worlds in the solar system, I personally believe Titan to be the one most likely to host life -- even complex life.

I don't mean in the water of some subsurface ocean, though that's not impossible. I mean in the sea, on the land, and maybe even in the air.

This is why.

It's a paper entitled Computation at the Edge of Chaos: Phase Transitions and Emergent Computation by Christopher Langton.

Langton's paper is in my opinion one of the more important papers I've read. The fact that it's not more widely known and discussed probably stems from its origin in an obscure area known as Artifical Life, a kind of ad-hoc interdisciplinary intersection of computer science, biology, physics, math, and philosophy. I first enountered it there as a result of my own interest in evolutionary computation and machine learning. This field was among my "first loves," and I still get e-mails about a neat little hack I wrote while exploring it.

It's a pretty easy paper to follow. Langton surveys the space of cellular automata, simple computer programs with fascinating emergent properties, by assigning them a computable statistical parameter λ (lambda) and exploring how they behave when sorted along that axis.

In slightly more layman's terms than those presented in the paper, λ measures something analogous to temperature and phase. At the 0.0 end are cellular automata that don't do anything and remain zero, while at 1.0 are those that just turn all-one. It's the behavior in between that is interesting. Rule sets near 0.0 exhibit very static behaviors; using the termperature and phase analogy we would call them "frozen." Those near 1.0 exhibit chaotic, random behaviors that we might call "gaseous" and "hot."

The most interesting of all are those near λ=0.5.

(All that is a bit of an oversimplification. Read the actual paper for the real deal.)

Here, Langton finds, cellular automata seem to exhibit universal computation. They exhibit behaviors like those of universal Turing machines.

At a phase boundary between order and chaos -- between ice and fire -- these idealized systems seem to take on the ability to store and process information. They become computers.

Life is all about information processing.

Another favorite researcher of mine from this field is Christoph Adami. He works at a lab at MSU where I very nearly did a Ph.D, had I decided to go that route, so I've met him a few times.

Here's Dr. Adami giving a TED talk on exactly the topic I'm writing about. I must absolutely credit him -- and Langton -- with the opinion I'm expressing here. While I haven't heard him talk about Titan specifically, for all I know I'm echoing something he might argue.

Adami entertains a sort of algorithmic information-theroetic definition of life that I find fascinating and absolutely compelling. I really hope he'll forgive me if I'm butchering it in my attempt to both paraphrase and render it more generally understandable. Adami defines life as "a phase of matter in which the dynamics of information processing come to dominate the ordinary dynamics of matter and energy."

Life, according to this definition, is a phase of matter in which computation dominates behavior down to the molecular level. We might call this phase "Turium," after Turing universality.

... and if Langton's thesis is basically correct, Turium occurs in the vicinity of phase boundaries in nature.

So where do we see that?

Earth is one such place. In this photo you can see water in all three of its phases: solid, liquid, and gas. Water is essential to life on Earth, and if the line of reasoning I'm following here is correct it may be precisely because of its effects on the informatics of organic matter that this is the case. At Earth's temperature and pressure, water inhabits a behavioral state space where λ=~0.5. It's life's universal solvent because in the vicinity of that phase boundary matter dissolved in water can exhibit universal computation. It can encode, transcribe, and process information.

Titan is another. It has rivers, lakes, clouds, and apparently rain and snow.

Solid, liquid, and gas equal phase boundaries. But the working material is different. On Titan the phase-transitioning material is hydrocarbon, not water. Its oceans are of methane and ethane.

But if life is fundamentally informatic in nature, maybe that doesn't matter. It's not the medium but the message -- or the ability of the medium to compute messages.

So I'm going to go out on a limb here and make a prediction: there's a high probability that we'll find life in Titan's cryotropical seas. I wouldn't even be that shocked if there were life on land. If it turned out to be complex, multicellular life -- if there were stuff swimming or running around -- I still wouldn't be that surprised.

It would be radically different life from a physical and chemical point of view, but it would be the same from an informatic point of view.

If there were intelligent life in Titan, we would appear to it as burning hell-beasts with blood of molten water. To us it might appear slow and strangely static, possibly running on a different sort of kinetic time scale and only apparing alive under time lapse photography. We would never be able to meet. On Titan we would freeze solid, and on Earth they would flash vaporize. But we could talk, and even land probes or robots. We wouldn't have to worry about contaiminating Titan since nothing from Earth would metabolize at those temperatures, and vice versa.


I never did end up doing that Ph.D, so this counts as a bit of amateur science pontificating. I decided to write it up after waiting years and never seeing anyone else make this exact point, at least not in print. If some future NASA mission proves this hypothesis out, Langton and Adami deserve credit for doing all the real work behind this speculation.

Instead of going the academic route, I decided to repeatedly try to ride the tech startup mechanical bull. Here's my current attempt. I'll leave it to you to decide whether this makes me smart, dumb, or just crazy. :)


One final P.S. that some readers might be wondering about: what about computers? They're Turing-universal, but they seem pretty (literally) solid state.

I think Langton's thesis still holds. CPUs send electrons through cascades of switches, and in so doing dissipate energy as heat. Electricity is a highly "organized," profoundly low-entropy form of energy. Heat is high-entropy energy. CPUs are devices where electromagnetic energy seems to be undergoing a thermodynamic phase transition from low to high entropy. I'm not knowledgeable enough about quantum or statistical mechanics to take this any further, but it seems like the same principles may be at work but in a radically different sort of medium.

When I was playing with artificial life systems like my own Nanopond, I once did a casual experiment where I attached a thermometer to my CPU's heat sink and then ran the simulator. The simulator begins by randomly generating genomes until one of them can self-reproduce. When this occurs, these self-replicating "bugs" then proceed to explode and take over and evolve. When that happened I measured (with no other tasks running) a tiny increase in the CPU's temperature. Since self-replication decreases the overall entropy of the executing bitcode in the "pond," perhaps this resulted in fewer missed branches in the pipeline and thus more work for the core.

Perhaps it's not just a simulation.