Giving Up on Life.

The ‘crawl’s been kinda quiet lately, mainly because I am (for the second time in as many years) on the last lap of this dumb novel. I am, in fact, committed to delivering the damn thing to  Tor before I leave for FinnCon — and I’m on track to do that, if I don’t let myself get distracted. As a result, any occasional posts you might read here over the next month or so will most likely be limited to Echopraxian fiblets.

(That said, I am typing this while sitting on our front porch, twilight deepening around me. I have fond hopes for a replay of last night’s three-way dust-up between rival gangs of cats, possums, and raccoons, all of whom converged pretty much simultaneously on the kibble we leave out here as an offering to the local wildlife. If that happens — and if I escape with all my toes and some decent pictures — I might post those too. Although I don’t know how long it might take to upload a toe using my Telus account.)

Anyway. While this project sprints for the finish line, others unwind around the world. One such is a little e-collection which is about to come out from Fata Libelli, in Spain— and over the past week or so I’ve been sneaking away from Echopraxia now and then to answer some questions they e-mailed me in hopes of spurring interest amongst their base.

Tonight’s impoverished offering, therefore, is an prexcerpt from that interview, with a bit of a semantic bent. Because we all know that if there’s anything more fascinating than watching a bunch of panelists sitting around arguing about the definition of “science fiction”, it’s gotta be watching a bunch of people sitting around arguing about the definition of “life”:

FL:

Your aliens have been widely praised for looking genuinely different, not just like green humanoids. Sometimes, those aliens are so weird (no genes, no cephalisation, hive minds instead of individual selves) that even the main characters have trouble identifying them as living beings. Is there a basic definition of ‘life’ suitable for humans, aliens and IAs?

PW:

Up until recently, Dawkins’s definition would have done just fine: Life is information, shaped by natural selection. Of course, that means that computer viruses have the potential to qualify as life forms, not just metaphorically but literally. I can live with that. A-life can meet Darwin’s criteria as well as any other kind.

The problem now is that we’re actually in the process of creating synthetic life — squishy bugs with real genes and metabolic processes — pretty much from scratch. Those things are undeniably alive, yet were not shaped by natural selection. You could make an analogous case for any conscious AIs not derived via genetic algorithm.

Darwin coined the term “natural selection” to distinguish it from the “artificial selection” that characterizes things like the selective breeding of dogs and pigeons. So perhaps tweaking Dawkins definition to “Information, shaped by natural or artificial selection” might be enough to cover the synthetics coming up through the ranks.

Or maybe it’s time to give up on defining “life” in terms of the way it was derived, or what it’s made of, and to concentrate instead on what it does. So: How about defining life as any complex of structured energy pathways that restricts entropy increase below some threshold rate?

What do you guys think? Anybody have any thoughts on where that threshold might lie?  Hell, judging by your past comments it’s pretty obvious that  most of you know how to define “entropy”, and I bet a few of you even know what units it goes by. Which, offhand, is more than I can claim right now.

Shhh. Something rustling under the porch…

This entry was written by Peter Watts , posted on Wednesday May 15 2013at 03:05 pm , filed under biology, interviews, just putting it out there..., public interface . Bookmark the permalink . Post a comment below or leave a trackback: Trackback URL.

34 Responses to “Giving Up on Life.”

  1. It’s interesting to think about. If we create a synthetic critter that can only be ‘grown’ or decanted or hatched in a factory but isn’t capable of reproducing would that qualify?

    I tend to think of it in terms of biological machines and non-biological machines. We think of life as biological and evolved, we think of AI as mechanical and artificial. But there has to be a spectrum between the two.

    I’m reminded of the line in blind sight about meat being reinforce-able but at some point is it really meat anymore? I would also say at some point we could make a molecule for molecule copy of a microbe, extant or extinct, that would be capable of biological reproduction and that would be very much alive.

  2. Come on now Peter, I can tell you’re just begging for some microbe to evolve that violates the laws of thermodynamics.

  3. I started writing this reply sure we could salvage the old definition, just by refactoring it slightly. I’ve given up on the idea of “life” as a property possessed by discrete objects anyway. Life is the vast four-billion year old self-perpetuating chemical reaction with the characteristic feature of undergoing natural selection (i.e. descent/modification/differential survival). That’s what it does as well as why it is. Synthetic life from scratch is just the instantiation of another one of those chemical reactions. No philosophical upheaval needed.

    But the awkwardness comes when someone makes something like Seth mentions, a from-scratch squishy critter that can’t reproduce but can, e.g. eat, move, plead for the continuation of its existence. Is there a distinction between a (spayed) trad-“living” cat and a perfect sterile factory-produced organic simulacrum? If that distinction is merely that one is part of Life (the reaction) and one isn’t, then it kinda looks like “life” isn’t such an interesting property, and really metabolism is the important thing after all.

    If there are two properties being conflated, I’d still maintain that “being part of a Life chain” is the one that deserves the term “life”, and prefer if the thermodynamics-related property got some other term. But I suspect the language will (hah) evolve the other way under the twin selection pressures of attention-bait headlines and toy/pet marketing.

  4. Isn’t one of the classic definitions that it has an energy cycle of some sort (feeds and builds/maintains itself), defends its own existence, and reproduces? Add in any kind of ability to adapt — whether in the energy-budget, defense, or reproduction — and I think that’s life, regardless of how it became what it is.

    Cheers,

  5. Why do we need a definition? It’s pretty clear what all these things are, we understand them all at a molecular level. We know a synthetic organism is a DNA/RNA/protein machine, we know what every bit of it does. We know in rough outline the working of naturally evolved organisms. We know exactly how computer programs work. Why not let language evovle naturally, and “life” take on whatever patchwork definition is most convenient for the moment?

    Except… there’s a moral question hiding behind this definitional one. There are *shoulds*. Life is good, there *should be* life. If “synthetic life” is “better”, should it replace naturally evolved life?

    I don’t know what life is, but I know it when I see it.

  6. Your definition of life as any complex of structured energy pathways that restricts entropy increase below some threshold rate could maybe be tweaked a little further. After all, entropy always increases. What life does is constrain the flow of entropy so that it may be put into use as free energy. Life then is self-perpetuating constraints on entropy maintaining a far-from-equilibrium state.

  7. @Paul Harrison, who wrote, in-part: [T]here’s a moral question hiding behind this definitional one. There are *shoulds*. Life is good, there *should be* life. If “synthetic life” is “better”, should it replace naturally evolved life?

    I think that the morality of life — if you want to hinge it on “what’s better” or isn’t as “good” — becomes highly subjective. If we’re discussing prokaryotes, it’s fairly simple and emotionally/personally abstract to suggest that one organism is “better” than another, has some particular efficiency or somesuch, and thus that relative to the other organism, it is thus “good”. But that’s really more far from the idea of morality that you had in mind, perhaps?

    If you’re the organism that it drives out of some environment (or utterly consumes from the environment), if you were capable of thinking, you might think “damn that’s effective” even as the pseudopods engulf. I might point out that in general, mice seem to me to look generally just nervous, at least while alive, but the ones the cat drops off in the morning tend to look more, well, amazed. The tomcat, of course, is notoriously not notable for its morals, unless of course you are the farmer whose grain the mice were eating. In that case, the tomcat is admirable. Laudable, even. 😉 I doubt that the tomcat gives it a lot of thought; after all, he’s just a highly tuned killing machine that pisses out a microorganism that infects mouse brains and makes them attracted to the urine spots where the tomcat has marked his territory. And can one honestly suggest that it’s morally laudable for an organism to rewire a host’s neurology to make it attracted to a place where the host’s predator is sure to return?

    Sure, it’s morally laudable if you’re the parasite that lives part of the life-cycle in the mouse, and another part in the tomcat. Other than that, the tomcat benefits from zombie mice that suddenly get horny smelling cat piss once their brains have been rewired… but there’s nothing in it for the mouse that is now compelled to (probably quite cheerfully) await its doom.

    Of course, if life didn’t exist at all, the question would be moot, so perhaps you’re right. Life itself is “good”, even if only because it gives us something to discuss.

    I might point out on a tangent that one of Peter Watts’s characters might argue that life is good, but geometry is not, at least not those pesky 90-degree angles…

  8. Entropons?

    Ergs per square yard?

    errr…

    Can I phone a friend?

  9. Consider the following entities:

    1. a steam engine, which might pass a definition of life as “constraining the flow of entropy” if that definition were loosely drafted enough, or could be considered as part of the “extended phenotype” of the humans who made it (pretty sure Dawkins discussed this in the book of the same name)

    2. an obligate parasite bacterium with a minimal gene set, like Mycoplasma or somethin’, incapable of independent life

    3. a biotech bacterium assembled from scratch that can’t reproduce without very very specific conditions carefully maintained by humans

    4. an artificial von Neumann machine based on nanotechnology that uses different building blocks and pathways from Life As We Know It

    5. an obligate parasite organism designed from scratch using mostly known biochemistry but 20% completely new pathways (or 30%? or 50%?)

    6. an organism designed from scratch to look as if it evolved by natural selection (complete with parasitic genetic elements, ‘dead’ genes, inelegant redundancies etc) – say a bioweapon aimed at crops

    7. an organism from a biosphere that originated & evolved completely separately from Earth’s

    8. a chemical soup where RNA strands or inorganic crystals are just beginning to undergo some form of selection

    9. a “Boltzmann human” that is exactly the same as your body and brain right this instant, formed by random entropic fluctuations a trillion trillion years in the future

    10. an exact copy of yourself artificially assembled using a replicator or whatever

    etc. etc.

    Which of these entities you consider living has to be a matter of taste and personal preference, to some extent. A historical definition (“the result of natural selection”) founders when it comes to clones or fakes, and might also struggle when it comes to very early stages of life.

    If we include products of artificial selection or design, an assembled bacterium that needs constant human attention to reproduce seems like it should be classed as alive, but then should we include a mass-produced steam engine design that has a ‘metabolism’ and can ‘reproduce’ itself with human help?

    I think the whole question is ambijective: http://slatestarcodex.com/2013/05/05/ambijectivity/

    A rock is definitely not alive, a beetle is definitely alive, but there is a large and group of objects which could reasonably be classed as living or non-living.

    tl;dr: Life is like pornography. I can’t define it, but I know it when I see it (because it gives me an erection).

  10. Maybe some of the imprecision comes from ‘life’ being both a singular and compound noun in English?

    On the singular scale, I remember from a biologist friend that the reproducing / entropy reversing type classification as to whether something is alive or not starts getting tricky when you get to the level of a virus or prion.

    As for the compound sense of information that’s evolved by selection, if artificial selection doesn’t count, what are Brussels Sprouts or Labradoodles?

    I would interpret ‘natural selection’ as meaning variations get selected depending on whether they work in some way or not, as opposed to variations being selected by the mystical divine process of some deity.

    Aren’t a lot of our current cloning and artificial organism creation processes still very much trial and error? We’re still not able to engineer genes as precisely as we’d like, it’s more try a few hundred times and keep the survivors.

  11. Ah, well, damn near everyone beat me to asking the apparently obvious. If AIs cannot procreate, are they “alive”? If it can (I suppose they will except they’ll likely need storage space, etc. from the baselines).

    And mightn’t the proposed new definition leave out the likely admittedly rare hardcore nihilist? Philosophy raises its ugly head and R.S. Bakker hatches an aphorism.

  12. @Hugh, who wrote, in-part: Aren’t a lot of our current cloning and artificial organism creation processes still very much trial and error? We’re still not able to engineer genes as precisely as we’d like, it’s more try a few hundred times and keep the survivors.

    I am assuming you were not intentionally being subtly ironic, because after all, that is about how things work in natural selection. Mutation occurs, and generally it’s detrimental. As in your statement, try a few hundred times and keep the survivors. Most that survive do so because their mutation isn’t beneficial but at least in the current environment it isn’t detrimental or fatal.

    We are improving, remember how pharmaceutical companies used to pick a compound known to be useful and then make all of the variants they could muster? For example, the amphetamines. Somewhere under erowid.org is the diary of one of their chemists, who made a very large number of variants on the general theme of amphetamine and then tested them on himself. Progress, in the modern day, is seen by us modeling various compounds computationally, and picking for trials only those conformant to some notion of significantly enhanced effectiveness. The previous trial-and-error method might be thought to be much more like evolution, and the modern method more like intelligent design.

    I still think that a good definition is that it has an energy budget (or call it “anti-entropic” or “entropy-resisting”), builds/maintains itself, resists damage or destruction of itself, and reproduces. The latter two are going to vary in an uneasy balancing act, selectors of quantity versus selectors for quality in reproduction, or massive reproduction for those things that are easily destroyed or eaten, possibly less of that for things not easily damaged.

    So, is the global economic system a living thing? I don’t see it reproducing, but it definitely seems to be anti-entropic more or less, builds itself, and through the interventions of the humans that created it, it resists and/or heals from damage.

    I should be quiet for a while, I think. 😉

  13. Let me attempt a criticism of your sketch of definition of life.

    In many cases, organisms do not even try to maintain their entropy below a threshold. As a first example, take some mushroom that just grows as much as possible. It tends to increase its size and its entropy increases with it. (Ok, I am not very charitable in my interpretation of your definition…) A second example is given by organisms that sacrifice themselves for their offsprings.

    Moreover, too little entropy is just as lethal as too much. We can freeze to death!

    Another possible problem is that you might not even be able to apply concepts like energy or entropy to a computer virus or any software life.

    To construct a definition of life, I prefer the following idea.

    Before Darwin, life and organisms were defined by their goal directedness. Being alive is “trying” to reproduce, to eat, to maintain a certain level of entropy and so on. Maybe we should go back to this earlier view. Understanding the process of life over time as evolution is wonderful, but as far as definitions are concerned it might be a dangerous temptation.

  14. “What do you guys think? Anybody have any thoughts on where that threshold might lie? Hell, judging by your past comments it’s pretty obvious that most of you know how to define “entropy”, and I bet a few of you even know what units it goes by. Which, offhand, is more than I can claim right now.”

    Probably best to stay out of the units discussion, as entropy should not be considered in terms of energy., S being defined as Joules per Kelvin be damned. That is a shallow truth, like when you were told that there were three states of matter–solid, liquid, and gas. Whuuut? No plasma? It’s a pedagogy thing, and Wikipedia is not half so good a reference as many people believe.

    To approach entropy, we don’t need to know anything about energy; just the probabilities of possible states.

    Entropy is a measure of what we don’t know about a system. We can’t even think of it as a measure of disorder. If we can accurately measure even an arbitrarily large disorder, the entropy of the system is 0.

  15. A bit of explanation about that last post. I come at this from a physics orientation. I’m not sniping at people who who have a biology approach. In fact, I recently discovered the field of computational population biology. In retrospect, I guess I thought that would be stale by now–you can hang ‘computational’ on anything these days, and it’s new. Surely computational population biology holds no great insights into something that must have always been based around serious math?

    A quick review of only *some* of the literature tells me that that would have been an unwarranted assumption. So, no. I’m not sniping at biologists.

  16. Stuart Kauffman has a whole book on this, called _Investigations_. Here’s the short summary of what “life” is (might have to tweak it still further to accomadate digital life):

    “An autonomous agent is something that can both reproduce itself and do at least one thermodynamic work cycle.” (pull quote from http://www.edge.org/3rd_culture/kauffman03/kauffman_index.html)

    That takes care of any synthetic critters who bypassed “natural” selection (or selection of any sort).

  17. Something that’s probably been pointed out before about Dawkins’ definition is that it begs for further elaboration that can easily result in circular formulation. Photons from a remote source, for example, surely constitute information that is “shaped” by stochastic influences such as gravitational lensing, objects that block light etc. The exclusion of these photons from Dawkins’ definition seems to hinge on our understanding of “natural selection”, a concept that seems hard to define without resorting to terms like “life” or “organism” (already the word “natural” appears somewhat loaded).

  18. I don’t think we can come up with a working definition of life that isn’t going to be pretty fuzzy already. Especially since we’re press-ganging the same word into working for ‘at a species level’, ‘at an individual level’, and ‘at a cellular level’.

    I mean, take ‘reproduction’, an element of a lot of these definitions. There are plenty of people who can’t reproduce, but obviously, they’re alive. And cellularly, they still reproduce. But we say they’re alive because they’re part of the human species. If a species is composed of a large number of non-reproducing bodies, without which the species can’t continue, do the drones no longer count as alive?

    An organism doesn’t “reproduce itself” so much as ‘have a design (not necessarily intelligent design) where it’s easy to reproduce. That is, there are plenty of things which may, say, grow spores or plant cuttings, there’s no intention behind it, and, if a completely inorganic system was cut out of the equation (like the wind spreading the seeds to barren areas) the reproduction might not happen. It’s not reproducing itself, it’s just being of a ‘type’ that allows outside forces to reproduce it. Similarly, you could consider any software object to qualify under those grounds… merely by BEING code, something that anybody (and even automated processes) CAN replicate, it can qualify, even if the program itself doesn’t take active measures to replicate itself. Because in our ecology, backups will tend to happen even by accident.

    And of course, until we’re prepared to either consider fire ‘alive’, any definition we use must find a way to explicitly rule it out.

    Personally, I say, screw life. As a category, I mean. It’s a term that started out being fairly useful, but the usefulness is decreasing all the time, the more we know and the more we create our own tools and agents that blur the categories. Or, I mean, we can use it, in the same way we might talk about “art” or “good”, a term we can throw around loosely in conversation, but abandon any hope that we should have a universal definition everybody should agree on scientifically. I think if you ask if something on the borderline is ‘alive’, you’re focusing on the wrong question.

  19. I like the general spirit of the definition. Its good. Just be careful with that word ‘entropy’, its a tricky one. So, for example: ecologists & etc. use the principle of maximum entropy to explain why life arises, e.g. A. Kleidon, Phys. Life Rev. 7, 424 (2010). among many others.

  20. Sorry, here’s the URL for above reference; its an open-access journal piece: Kleidon, “Life, hierarchy, and the thermodynamic machinery of planet Earth.” http://www.sciencedirect.com/science/article/pii/S1571064510001107

  21. Hugh wrote: “Aren’t a lot of our current cloning and artificial organism creation processes still very much trial and error? We’re still not able to engineer genes as precisely as we’d like, it’s more try a few hundred times and keep the survivors.”

    No. We can assemble any sequence we want, pretty much. The trick is to start with oligonucleotides (short DNA fragments), the synthesis of which is completely routine, then use DNA repair enzymes to assemble them into larger and larger strands. This is then inserted into a hollowed out cell (RNA and protein machinery without and chromosomes). Craig Venter was recently enthusiastic over doing this to create an organism with a wholly synthetic chromosome. He has a Java app…

    However, the protein folding problem is only partially solved. Designing a novel protein… we could do better than evolution, but not hugely better, and evolution has been going on for a long long time. For the present, it’s far more productive to appropriate existing evolved proteins (like DNA replication and repair proteins!).

    So it’s like a computer, you can tell it to do exactly what you want it to, and it will do precisely that…

  22. Zed: Is there a distinction between a (spayed) trad-”living” cat and a perfect sterile factory-produced organic simulacrum? If that distinction is merely that one is part of Life (the reaction) and one isn’t, then it kinda looks like “life” isn’t such an interesting property, and really metabolism is the important thing after all.

    I prefer the term “persistence” over reproduction, for pretty much exactly that reason. Not get all teleological or anything, but the point of the exercise is to extend your personal information forward in time; replication is generally the way we do this because our chassis wear out, but an immortal chassis would get the specs into the future even better (without all that nasty mixing). Lamarckian evolution might be doing the job somewhere, somehow, I suppose.

    Paul Harrison: I don’t know what life is, but I know it when I see it.

    Yeah, but look at how well that worked for “porn”. You know it when you see it because it gets you off; but different strokes and all that. Sure, porn involves naked bodies for a lot of folks; but it involves shoes and vacuum cleaners for someone out there. There’s a whole class of fetishism involving inanimate objects (there’s a woman in Germany who has a sexual relationship with the full-scale guillotine she keeps in her living room, and who married the Berlin Wall).

    Given the philosophical value we attribute to “life”, I don’t know know if such flexibility is a good thing when it comes to, for example, deciding whether the Europan subsurface ecosystem should be preserved. Whose “knowing it” do we go with?

    Daniel Brooks: Life then is self-perpetuating constraints on entropy maintaining a far-from-equilibrium state.

    Yes! That was the concept I was groping for: active divergence from equilibrium. In which case life can be assessed not as a binary state, but along a gradient: the further from equilibrium a system maintains itself, the more “alive” it is.

  23. Mr Non-Entity: I doubt that the tomcat gives it a lot of thought; after all, he’s just a highly tuned killing machine that pisses out a microorganism that infects mouse brains and makes them attracted to the urine spots where the tomcat has marked his territory. And can one honestly suggest that it’s morally laudable for an organism to rewire a host’s neurology to make it attracted to a place where the host’s predator is sure to return?

    Sure, it’s morally laudable if you’re the parasite that lives part of the life-cycle in the mouse, and another part in the tomcat.

    I think what you’re getting at here is that you may not be able to define morality, but you know it when you see it…

    Ben: Consider the following entities:

    All of which are cool but 6 and 9 of which are especially so…

    You’re talking gradients again, which I like a lot because it eschews arbitrary yes/no boundaries while retaining a certain quantitative rigor lacking from the porn perspective. (I seem to recall Dawkins once described the dynamics of clay particles in a river bed — certain sizes differentially created/accumulated by physical forces — as a kind of proto-natural-selection. Which would be maybe a little bit above the steam-engine end of your scale.)

    Hugh: On the singular scale, I remember from a biologist friend that the reproducing / entropy reversing type classification as to whether something is alive or not starts getting tricky when you get to the level of a virus or prion.

    Well, yeah, in the sense that they can’t replicate on their own, and are utterly dependent on some kind of host machinery to do a lot of the heavy lifting. Then again, you could draw the same kind of line around human beings, who might not be truly “alive” because we can’t do many of the things essential to our continued existence — gas exchange, energy acquisition, and so on — without relying on an external environment to help us out. Where do you draw the line?

    Again, I think a gradient in which departure from equilibrium state quantifies “degree of aliveness” is a great way to resolve this.

    Also “entropy reversing” might get you into trouble, given the fundamental problems with the very term “negentropy” (which, I’m given to understand, does not actually exist. Life doesn’t reverse entropy so much as simply slow down its increase compared to the rates observed in nonliving systems. Mr. Non-Entity uses the term “Entropy resisting” just downstream. I like that better.

    As for the compound sense of information that’s evolved by selection, if artificial selection doesn’t count, what are Brussels Sprouts or Labradoodles?

    Horrible. Both of them are horrible.

    Mr Non-Entity: So, is the global economic system a living thing? I don’t see it reproducing, but it definitely seems to be anti-entropic more or less, builds itself, and through the interventions of the humans that created it, it resists and/or heals from damage.

    I, for one, wouldn’t rule it out. Although it might not be the theoretical construct itself that resists entropy so much as all the physical infrastructure necessary to keep the house of cards away from its equilibrium state…

    Tickli: Another possible problem is that you might not even be able to apply concepts like energy or entropy to a computer virus or any software life.

    Hmmm. But surely the software manifests through the operation of physical structures and systems? E.g., a flipped bit resulting from quantum tunnelling at wee scales would be a mutation of sorts, and a teensy increase in entropy.

    Before Darwin, life and organisms were defined by their goal directedness. Being alive is “trying” to reproduce, to eat, to maintain a certain level of entropy and so on. Maybe we should go back to this earlier view.

    Kind of amazing how much Darwin got right back then, isn’t it?

    What about the suicidal (prior to suicide) or any life form whose survival instincts had been eliminated in some fashion? Is a metabolizing system that starves itself because it is no longer “trying” dead bey this definition?

  24. Joe Ardent: “An autonomous agent is something that can both reproduce itself and do at least one thermodynamic work cycle.” (pull quote from http://www.edge.org/3rd_culture/kauffman03/kauffman_index.html)

    That takes care of any synthetic critters who bypassed “natural” selection (or selection of any sort).

    But not those of us who’ve had vasectomies…

    gregm: Entropy is a measure of what we don’t know about a system. We can’t even think of it as a measure of disorder. If we can accurately measure even an arbitrarily large disorder, the entropy of the system is 0.

    I’m not sure I’m getting this. My understanding is that entropy basically comes down to the amount of information necessary to completely describe a given system; a totally random system has high entropy because you need to describe the specs on every particle independently (i.e., there are no rules or symmetries that allow you to predict the position of any elements short of brute-force mapping). But your definition seems to imply almost a ind of observer effect, in which the entropy of the system is somehow related to how much data we have in-hand about it.

    What am I missing?

    Corwex: Something that’s probably been pointed out before about Dawkins’ definition is that it begs for further elaboration that can easily result in circular formulation.

    Fair enough — then again, Dawkins himself seems to recognize as much with his river-delta-as-natural-selector analogy.

    Either way, gradients to the rescue.

    Peter D: And of course, until we’re prepared to either consider fire ‘alive’, any definition we use must find a way to explicitly rule it out.

    I actually wrote a story once — the only story directed at a primary-school audience I ever wrote — in which a UN orbital computer network, charged with protecting the biosphere, decided that fire met all the metabolic criteria necessary to qualify as life forms and consequently set out to ignite firestorms across the planet.

    Not one of my more successful stories. The sixth-graders hated it because none of the characters were likeable and the world blew up at the end.

    (Oh, and good point about cellular reproduction in non-reproducing bodies.)

    Linas Vepstas: Kleidon, “Life, hierarchy, and the thermodynamic machinery of planet Earth.” http://www.sciencedirect.com/science/article/pii/S1571064510001107

    Thanks, Linas. I look forward to reading this.

  25. Defining life sounds about as straightforward as defining intelligence or pornography. They seem to be relative values based on the observer, not the observed.

  26. …I wonder if I can I make this all hang together….

    Say I’m online and chatting up a woman. How can I know it is truly a woman, and not the famous computer scientist Alan Turing? And should we not therefore conclude that Alan Turing should for moral and legal purposes be considered a woman? This is the famous Turing Test, which the aliens of Blindsight angrily failed and the English legal system chose to disregard. This test of life is whether an entity can interest me in “knowing” it in the Biblical sense. Yes, there will be disagreements, and we’re going to have to live with that somehow, hopefully peacefully, possibly violently. I don’t mind if you think some particular vacuum cleaner is alive, but I might have a problem with it if you want to fill the planet with them. Perhaps a peerage of lifeforms that all more or less agree that they are all alive and, with sufficient technological help, potentially worth shagging. Or at least chatting with, which is kind of like sex but for brains.

    To put it in exceedingly Dawkinsian terms, genes want to survive, and one excellent strategy for this is seeking out strange new genes to hang around with. Genes are life, everything else — cells, clusters of cells with germ lines, Captain Kirk — is just politics.

    Re entropy based test: With the rate we’re burning coal and oil, I think we fail. And Shannon’s box that turns itself off whenever it is turned on would count as alive, which is awesome.

  27. Peter Watts:
    none of the characters were likeable and the world blew up at the end.

    this makes a fun blurb.

  28. Peter Watts asked: “My understanding is that entropy basically comes down to the amount of information necessary to completely describe a given system; a totally random system has high entropy because you need to describe the specs on every particle independently (i.e., there are no rules or symmetries that allow you to predict the position of any elements short of brute-force mapping). But your definition seems to imply almost a ind of observer effect, in which the entropy of the system is somehow related to how much data we have in-hand about it.

    What am I missing?”

    In Boltzmann’s definition of entropy, the entropy of a system is the (logarithm of the) “number” of different completely specified states (microstates) that have the same macroscopic properties as your system.

    (The word “number” should be replaced by an appropriate volume in the space of possible configurations of your system if you work with a continuum of microstates.)

    This definition is somewhat subjective, since the choice of macroscopic properties is not clearly determined.

    Also, there are many inequivalent possible definitions of entropy…

    “Hmmm. But surely the software manifests through the operation of physical structures and systems? E.g., a flipped bit resulting from quantum tunnelling at wee scales would be a mutation of sorts, and a teensy increase in entropy. ”

    This is very ingenuous, but there is no reason to think that this method of extending the definition of entropy really works. It seems natural to think that artificial life in simulated worlds should rather be defined according to the entropy according to the laws of the simulation. However, this will not in general correspond with the entropy of the system running the simulation. It is also doubtful that physical laws that can support something we would all recognize as life must necessarily support notions of entropy or energy.

    “What about the suicidal (prior to suicide) or any life form whose survival instincts had been eliminated in some fashion? Is a metabolizing system that starves itself because it is no longer “trying” dead by this definition?”

    If a man wants to commit suicide, his heart still pumps the blood to feed his cells, his immune system is still protecting him, his lungs are still oxygenating the blood and so on… Most of the organism is trying very hard to make the person survive. We should not reduce someone to his neocortex 😉

  29. Peter Watts:
    Not one of my more successful stories.The sixth-graders hated it because none of the characters were likeable and the world blew up at the end.

    Had to target a few grades higher to break into the angst-ridden blogging teenage girl market, eh?

  30. @Peter Watts, who said in part after quoting me, in part:


    Sure, it’s morally laudable if you’re the parasite that lives part of the life-cycle in the mouse, and another part in the tomcat.

    I think what you’re getting at here is that you may not be able to define morality, but you know it when you see it…

    It’s not quite that simple. I know my morality when I see it, and you know yours, even if we both might have some difficulty nailing it down and articulating it satisfactorily to all concerned. However, any given other entity asked the same question is likely to give a quite different answer. Morals are highly subjective for the individual and each culture has its own notions. Each species that is “social” might have a share set of moral principles, but of course it’s not just subjective, it’s “relative”, and a lot of it boils down to selfishness, if not always for one’s self, for one’s genes. “Moral is what’s good for me, what’s good for me is moral.”

    This whole question is probably most easily resolved by pointing out that because morals are so subjective and relative and often vary by situation, they might as well be left out of the discussion and rather we need an ethical system, which generally means a legal system. Something codified and laid down as law, often attributed to divine authority, may contain provisions intolerable to some individuals but the society can at least point to tradition or to codex, and decide whether or not to act (or not) in conformity thus evade sanction. Of course this line of reasoning easily degenerates into debate on whether it’s moral to abide by immoral statute, and there we are back at the beginning. To add even more recursion and pretzel-logic, the legal maxim is “moral action is the action which would be chosen by ‘a good man’.” Whatever that might be, especially if you are a giant purple space squid which has never even heard of a human being. 😉

    @Paul Harrison, who wrote in-part: However, the protein folding problem is only partially solved. Designing a novel protein… we could do better than evolution, but not hugely better, and evolution has been going on for a long long time. For the present, it’s far more productive to appropriate existing evolved proteins (like DNA replication and repair proteins!).

    So it’s like a computer, you can tell it to do exactly what you want it to, and it will do precisely that…

    Heh, so I am not so far off when I write using the neologism “organonanotech”. BTW I’ve been running folding@home throughout its existence and now I’m farming out spare clock cycles to BOINC.berkeley.edu — and thanks for the clarification about capability to build in DNA exactly what you want rather than having to slop at it a couple hundred times before you got what you wanted.

    That and 3D printing in tissues looks promising indeed! Head-cheese on rye, anyone?

    @Alexey, who wrote: Had to target a few grades higher to break into the angst-ridden blogging teenage girl market, eh?

    Oh gods, no, please warn the poor poor man, he really doesn’t want that market. Though I think he covered a while back some encounters with that ilk when remarking on a colleague’s online defamation as a raging misogynist by roving packs of the PC Brigades. Basically that market is writing for, to, and about each other and no others need apply. The only good thing about it is that eventually they encounter their adult phase which generally lurks quietly between eating their young. That is generally one of those “blink and you missed it” sort of things but well worth witnessing, if you manage to catch it before the tattered and mortified victim erases all online evidence that they ever existed. Though sometimes you can catch traces of it on 4chan screen caps under /b/ (which is definitely not safe for work).

    Cheers,

  31. @Peter Watts again:


    Mr Non-Entity: So, is the global economic system a living thing? I don’t see it reproducing, but it definitely seems to be anti-entropic more or less, builds itself, and through the interventions of the humans that created it, it resists and/or heals from damage.

    I, for one, wouldn’t rule it out. Although it might not be the theoretical construct itself that resists entropy so much as all the physical infrastructure necessary to keep the house of cards away from its equilibrium state…

    I think you touched on the matter of all of the automated trading that goes on at really astonishing speeds, to the point where even a smidgen of lag time due to telecom distances is a make-or-break issue. One might wonder if various traders share adjacent slots in co-location facilities quite near the actual trading exchange machines. One might also wonder if that much complexity in such proximity might in some ways approach the interconnectivity of, for example, various centers or lobes in a smallish mammalian brain. If that’s the level of complexity and interconnectivity, we probably wouldn’t see it as much of a functional whole, due to the fact that the “lobes” are all in competition with each other. And what exactly are any mechanisms promoting homeostasis or even feedback loops that aren’t positive feedback death spirals or widening polar extreme swings? There have definitely been “oh shit” moments, as you point out, where people pretty much went scrambling to pull the power cords.

    But start to look at the wider systems, the just-in-time-delivery nature of so much modern manufacture and then distribution of (partial or finished) product. And we long since have passed the time when living humans could possibly manage the logistics of our world. So which is the servant here, and which the master? Are we (so to speak) turning into little more than worker drones for the system? This is one subtext I took away from the Scramblers, whether you intended it or not: if you’re a perfectly good (however complex and capable) worker drone, do you need to be wasting potential efficiency on wondering whether or not you like being a worker drone more than you might like being an independent “self-actualizing” independent entity that just happens to be an essential part of a social-technological-economic system outside of which one cannot independently survive? (Subtle or moot questions, perhaps…) But I seem to recall that the Scramblers were thought to have always been non-sentient, though we humans might ultimately wind up just as non-sentient in the name of efficiency.

    Do folks here think we’re already committed to that path? Could we survive if we were to try to slowly deconstruct the machine to move back into a more “sustainable” way of life?

  32. When people talk about life preventing the entropy increase, they forget (or leave outside the definition of the system) one glaring thing: the energy source. (In the case of Earth, that’s obviously the Sun radiation). If we include the entropy of the incoming energy, the result is opposite: system with life in it produces greater entropy increase than similar system without life. That’s what justifies the existence of life from thermodynamical point of view: it’s the most efficient process of increasing the entropy of the system as a whole.

    Regarding the definition of life, I like David Deutsch’s take (he whom some call the Father of Quantum Computing). (*) Knowledge is structure across (quantum) universes. Life is a process of making many universes converge (and thus create knowledge). A living thing embodies knowledge about the niche it exists in, in one form or another (for a virus, this knowledge is implicit in its genetic material; for a human, there is in addition explicit knowledge in its brain).

    Unfortunately this approach won’t make much sense for those who don’t accept the Many Worlds Interpretation of quantum mechanics. Another disadvantage is that it’s too vague for practical applications, but I believe that as our knowledge of the world grows, we’ll be able to make it more quantitative.

    (*) I can’t wouch that I correctly represent D.Deutsch’s point of view here; these are merely my thoughts, in many ways inspired by D.Deutsch’s The Fabric of Reality.

  33. Funny that you should mention Entropy, I believe that a working definition of life should involve it.

    A living organism strives to keep an energetic and chemical balance separated from the surrounding environment, and tend to keep its entropy low and dump it in the ecosystem.

    Add the ability to reproduce and you may get a reasonable working definition.

    My statistical mechanics is a bit rusty, so bear with me if I can’t give you formulas.

  34. Just throwing it here because it’s pretty neat.

    http://bytesizebio.net/index.php/2012/02/11/life-is-short/