The life of a sofa

Someone once said - I think it was Arthur C. Clarke - that they couldn't think of a good definition of science fiction, only that they'd know it when they saw it. Personally I think hard sci-fi features the consequence of some scientific idea as integral to the plot. If the same story could be told without needing to explain any of the science, then it's not science fiction.

But I digress. They say philosophy isn't so concerned with definitions these days, but some of them definitely still cause headaches. Like, for example, life. That one certainly has everyone jolly confused. The school definition of MRS GREN is clearly unsatisfactory because it denies that mules are alive, not to mention (arguably) old people, which is silly. And I think this Aeon article makes things unnecessarily complicated :

At this point, it is instructive to distinguish two concepts pertaining to acquisition and use of information: we’ll refer to them as life and alive. Alive refers to the continual use of negative entropy. It is the opposite of dead. While alive, a cat can use the negative entropy it acquired in the form of food to generate order in its cells, to construct itself. Things that are alive are constructors. A dead cat is not able to do that anymore. Being alive requires a process of maintaining homeostasis, that is, overcoming perturbations and maintaining the balance of the organism, overall. 

Life, on the other hand, refers to the process that generates the required information. It also generates things that are alive and the information required to produce them. Also included in life are cases where information is acquired to generate things that aren’t alive, such as sofas...  Alive things are the generating systems for expanding this process in space and time by constructing new possibilities. Thus, the cat and sofa are both life, but only the cat is alive.

Sometimes I think there's a good case to be made for rationalising. There are some truths we already know but find it difficult to articulate, so what we're doing in our struggle for definitions is a sort of self-discovery, to put into words what we already know. Thus when someone says something that's obviously wrong, we can justifiably say, "that's stupid", even though we can't actually come up with anything better. And it seems pretty obvious that sofas are not life, they are a signature of life. I don't see any point in omitting that prefix and redefinining life in this even more confusing way.

What I do like about the article, however, is its notion that life has something to do with how information is processed.

Constructors are more sophisticated than things such as sofas. Sofas can be generated only by information, but constructors can also generate new things by processing information (this includes mutant variants of themselves, which allows evolution to work in the first place) – that is, they can use information. A cat is a programmable constructor in this sense and so are you, but the sofa is not. In all these cases however (sofas and cats etc), a process of evolution is necessary to generate them. 

A theory of information that could explain living systems will thus have to account for two aspects of information – how the information is acquired, and how it is used. This information was acquired over evolutionary time, for example through selection, via survival of the fittest. The use of the information is, as Schrödinger pointed out, accomplished by siphoning off negative entropy to pay for the increase in organisation that organisms need to survive, which they can do because they are constructors with information about how to produce themselves.

Of course, defining the precise kind of information processing system that is life is complicated.

Imagine you have built a sophisticated 3D printer called Alice, the first to be able to print itself. As with von Neumann’s constructor, you supply it with information specifying its own plan, and a mechanism for copying that information: Alice is now a complete von Neumann constructor. Have you created new life on Earth?

Presumably not. A robot that can copy itself is just a mechanical device. But...

Suppose you then rig up Alice so it acquires (through your design) more information: it can use rocks and the minerals derived from them as raw materials to make new 3D printers. Are Alice and her offspring (Bob, Charley, Daisy and Eve) now life? 

Well, no, but...

Getting annoyed with continually having to find raw materials for all the little 3D printers running around, you decide to equip one offspring, Eve, with even more acquired information: solar panels for energy that enable Eve to go out by itself and use that acquired information to hunt for minerals. Is Eve now life?

Umm...

...you figure out a way to get your little autonomous 3D printer sent on the next mission to Mars as a stowaway. Imagine Eve has a happy existence in a hidden valley on Mars, and goes on to produce many copies of itself. Humanity discovers the valley a few million years later to find the process of evolution on 3D printers generated a wide variety of them that are quite different from your original design – small ones, big ones, blue ones, red ones, ones that hunt other 3D printers for resources, and so on.

Haaah. Yes. Awkward, isn't it ? At some point in the sequence, we can plausibly say that we have silicon-based life. But where ? What's the critical difference ? There doesn't really seem to be one. And yet it's not clear if we did end up with life in final iteration after all. We might have done, but we might equally still have only very elaborate robots. It's debatable whether that's really all life is, of course, but it's all very messy.

On the other hand, why should reproduction be counted as so fundamental ? If I stow a mule on board the Mars-bound spacecraft, there'll be life on Mars until the mule dies. It has no chance of ever reproducing. Similarly, we can confidently say that plants are alive but we'd also have few problems of declaring a true AI to be alive. These (assuming plants are not conscious) are extraordinarily different entities. What's the common factor ? And as far as reproduction goes, we can currently have only one known method for that. We can't yet create an AI by writing the correct code, but if we could, there'd be no reason that it couldn't simply copy itself by copying files. Our reproduction methods would be fundamentally different : we don't need to know exactly how the standard method works, but there would (presumably) be nothing to stop the AI from having full knowledge of itself and control of the process.

In short, we know life can be unconscious, but can't adequately define what that sort of life is, and we know life can be defined by consciousness, but we have even more difficulty defining that. So we're trying to come up with a definition that unifies at least two radically different states and we don't understand either of them very well. No wonder this is hard.

What can Schrödinger's cat say about 3D printers on Mars? - Michael Lachmann & Sara Walker | Aeon Essays

On a sofa in the corner of the room, a cat is purring. It seems obvious that the cat is an example of life, whereas the sofa itself is not. But should we trust our intuition? Consider this: Isaac Newton assumed a universal time flowing without external influence, and relative time measured by clocks - just as our perception tells us.