Hacker News new | past | comments | ask | show | jobs | submit login
Life (stanford.edu)
133 points by lainon on July 9, 2017 | hide | past | favorite | 60 comments

I find autopoiesis ("self-constructing, self-maintaining, energy-transducing autocatalytic entities") to be the most satisfying definition of Life. But I find it weird that they would throw that term out there without citing Humberto Maturana or Francisco Varela, who came up with the concept.

They wrote mostly long essays in obscure philosopher's jargon, but one day decided to go ahead and explain everything very simply; they wrote The Tree of Knowledge, which summarizes their philosophy in the style of a children's book. Very worth reading if you're into these kinds of questions. I found a PDF here:


I'm put off the idea of investing much time in reading that because of the blurb by Fritjof Capra.

Should I be?

To me, there seems to be a pretty straightforward line to draw between life and non-life: Life is that which undergoes Darwinian evolution.

If (a) it makes copies of itself, (b) the copies inherit traits from the original, (c) the traits are subject to variation/mutation, and (d) the success of the copies at making further copies depends on those traits which are inherited, then evolution must occur.

Every other definition seems to me over-complex and ad-hoc; like specific things we have noticed which tend to be true about most life on Earth, but which might not be true for special cases, or for some other hypothetical extraterrestrial life which we might discover someday.

This definition is an unambiguous binary classifier, and it seems to correctly categorize everything we would intuitively call "alive". (Also, under this definition, viruses are unambiguously life).

typically i'd ignore semantic arguments but since this is in reference to philosophical consideration i'll assume you are someone who actually enjoys getting into the weeds of semantics.. if i am wrong in that assumption please feel free to ignore the superfluous following

your definition loses your purported binary classification when examined through rigor

> (a) it makes copies of itself

what about those forms of life that are unable to procreate, or choose to abstain? also 'copies' is ambiguous: see (b);

> (b) the copies inherit traits from the original

seems an unecessary condition.. if one is to accept the first than this is redundant, also it is confusing because you say 'inherit traits' and 'copies', a copy should inherit all traits, so which are we talking about? copies or trait inheriting progeny? if 'copies' is intentional what about forms of life unable to make exact copies of itself, like humanity?

> (c) the traits are subject to variation/mutation

if 'variation/mutation' is random one possibility would need to be completely foregoing the 'variation/mutation'

if 'variation/mutation' is deterministic then are there deterministic models of 'variation/mutation' that you would leave out of your definition for life?

(d) the success of the copies at making further copies depends on those traits which are inherited, then evolution must occur

what about life that fails in its own ability to continue to evolve but heavily affords other life to evolve further.. i'm thinking something like wheat that is scientifically controlled in its evolution because we forced its evolution to a sweet spot that for years helped humanity to evolve as we wanted

Right - your objection under A includes mules, as an example. (Normally they're sterile.)

surely we must include mules in life.

I think they could be included by allowing for evolutionary dead ends. That is, if mutations are random, not all will be successful long term (though pecularities in their specific environment may enable their survival for some time).

You are missing a ridiculous large space of trait possibilities. But then, nobody ever remembers that.

Always test your definition against fire and crystals.

Fire: Does not mutate.

Crystals: Do not mutate, or if by some twisting of definition you say they do (e.g. defects are propagated through the lattice), the success of propagation doesn't depend on the inherited traits.

Hmm. If Darwinian evolution is based on factors currently present around the species, wouldn't it basically be optimizing for short-term factors? So you could call Darwinian evolution a greedy algorithm, that optimizes for a local maxima. What about the global maxima, how can life evolve to that state?

If I understand your question, the answer might be something to do with what happened in humans; we developed brains large enough to be able to recognise patterns in our own behaviour and abstract those across all aspects of life.

Having said that, the long term stability of that path is definitely debatable!

> What about the global maxima, how can life evolve to that state?

It doesn't. Evolution is indeed a greedy algorithm, and cannot plan.

As long as we're ok with the fact that by this definition, it's easy to make artificial life and we already have. (Take the entities from any sort of evolution simulator.) I guess you would call that "digital life", and contrast it with "physical life".

EDIT: I'm confused about the downvotes. To be really concrete, the swimmers in Darwin Pond[1] seem to me to satisfy all of OP's criteria for life. Do they not?

[1] http://www.ventrella.com/Darwin/darwin.html

I would say digital patterns which follow the rules above could be called "life", albeit artificial and exceedingly simple compared to biological life.

Is a computer virus that mutates itself via an evolutionary algorithm considered life?

I would say yes. It's just not biological

What? That's like saying "this drug causes sleep, it's just not soporific".

What makes you think life can only be biological? What about artificial life and digital organisms?

The word biological just refers to life or living organisms. So anything that's alive is also biological, by definition.

My definition of life: A forkbomb implemented in hardware.

Recently I got interested in the origin of life. I found this 3 part talk which discusses how the simplest possible life form might have looked like.

My feeling after watching this is that we are slowly getting there (which will also mean jump starting in a dish a self replicating thing from inorganic matter).




I think we now know enough about the mechanics of life (molecular biology in particular) to conclude that we are simply fantastically complex clockwork automatons.

It's awe-inspiring and impressive that it works, and that it likely evolved from goo, but I am legitimately curious if anyone actually disagrees with the statement above in a justifiable way.

The issue is with your word "simply". On one level, it's true, but on another level it's not.

The Eiffel Tower is "simply" a bunch of puddled iron. On an even more reductionist level, it's simply a bunch of quarks, as are you.

While "Both you and the Eiffel Tower are simply bunches of quarks." is a true description (according to modern physics), it's not a very useful one. It says nothing about why the Eiffel Tower is different from other bunches of quarks, and definitely says nothing about (e.g.) what kind of reaction you might have to visiting Paris and climbing the Eiffel Tower. Would you be scared (acrophobic)? Exhilarated? Bored? The "bunch of quarks" description is of absolutely no help there.

Reductionist models have their uses, but attempting to use them to explain everything is a grievous mistake.

> While "Both you and the Eiffel Tower are simply bunches of quarks." is a true description (according to modern physics), it's not a very useful one.

Except that in this case, the molecular details of life are useful. If you want to know why people need salt you need to know about ion channels. (This is in contrast to your Eiffel Tower example, where you don't gain much insight by looking at molecules than you knew at the material-science level.)

But ion channels (and their function) are on a much higher level than quarks. Looking at things at the quark level doesn't tell you anything about ions, or salt, or, for that matter, people. :-)

The question is are we simply automatons?

Yes we are automatons ("a really smart meat" per Ready Player One) but is that all there is to human life?

1. To me the mystery lies in the awareness.

Modern neuroscience/psychology/cognitive science/ are aware that subconscious accounts for most of brain's activity.

The aware part is such a small part that it is reasonable to ask whether it is needed at all.

Anecdotally we all have heard stories of people doing things on auto-pilot unaware of doing them(driving, blacked out shenanigans, etc).

Thus I can very well imagine a Chalmer's zombie which functions identical to most humans.

2. The biggest problem with simple reductionist strategy into automatons is what it means for morals/ethics.

It's not only that everything becomes "vae victis" or localized majority consensus. No globals left not even the golden rule.

Even worse everything becomes almost nonsensical.

Nihilist ISIS suicide bombers trying to escape the matrix then make as much sense as Elon Musk trying to expand humanity to Mars.

Just some automatons trying to make sense of it all.

> The aware part is such a small part that it is reasonable to ask whether it is needed at all. Anecdotally we all have heard stories of people doing things on auto-pilot unaware of doing them(driving, blacked out shenanigans, etc).

We know what someone without awareness looks like, we call them coma patients.

The anecdotal stories of people doing things and not remembering them does not mean they were unaware at the moment, it just points to the glitches in the memory processes. Alcohol interferes with ability to record short term to long term memory, among other things. As for driving on “auto-pilot”, it is your brain tuning out all-too-familiar sights and not recording them as not containing any new information.

Your brain is very good at optimizing things, and the more routine-bound your life becomes, the less new information comes in that it considers worthy of recording - if it sees an instance of a pattern it has seen a thousand times, it categorizes that commute to work as that pattern and does not bother recording it unless something out of ordinary happens.

That also explains why time seems to pass faster as you get older - there are less and less genuinely new experiences and encounters coming in for your brain to record and remember as interesting, so time flies by faster and faster.

Coma patients are just a subset of people without awareness.

Awareness is not binary. My friend going into diabetic coma exhibits a degree of awareness until you realize her answers to my questions have stopped and you have to call the ambulance.

I do agree that the brain optimizes routine things, thus my question is it not possible to function without awareness (ie sleepwalking through life).

I am going from books like these: https://www.amazon.com/Philosophy-Flesh-Embodied-Challenge-W...

Here's a premise: Most(90+%) of our thought occurs subconsciously . Is it not possible function 100% subconsiously?

Think about sleeping on a math problem and the solution coming to you suddenly.

I think your #2 is why I bring this up. It's interesting that our in-depth understanding of the specific mechanics is fairly recent; before, it was easy to say that there may be some essential quality of life that made it unique and valuable, simply because it wasn't fully understood.

The implications start getting dangerously close to http://www.paulgraham.com/say.html

"keep track of opinions that get people in trouble, and start asking, could this be true? Ok, it may be heretical (or whatever modern equivalent), but might it also be true?"

It changes nothing for ethics. It's a discipline shaped by the human emotions, and those don't change depending on our free-will (whatever meaning that word even has).

It changes the practical side.

As more and more psychopaths learn that there is no global ruleset the more entropy we get.

If everyone learns that there is no difference between suicide and drinking coffee(via Sartre), then the number of suicides will inevitably rise.

There are people who supported certain presidential candidate(s) just for the lolz. They were proud of their nihilism.

Yes, but clearly computers are also fantastically complex clockwork automatons, and they aren't alive. So it's not a complete definition.

Right! I suspect our intuitive definition of life includes a biological substrate. If I were somehow able to to engineer a new species of viable biological organism, possibly even not using any existing DNA building blocks, I think most people would consider the result "alive". When really I just created a new program for molecular machinery to run.

Clockwork is not a model that immediately springs to mind when I consider the complex chemical interactions and environmental sensitivity of a eukaryotic cell.

I disagree with your use of the word "simply".

"Automaton" is another potentially problematic word. Wiki tells me it means, literally, "acting of one's own will" - which we do - but most people would use it to refer to something that lacks some essential component of being human, which definitionally we don't. It is not a compliment to be called an automaton.

I submit that if you would make an exact clone of me and subject both of us to the exact same condition that we would behave differently.

I suspect it is because our two brains (while identical physically) would have differing quantum states such that the superstructure of active connections would be impossible to measure perfectly - and therefore impossible to duplicate perfectly.

The "exact" clone that you propose is fundamentally impossible. What you're looking for is the no-cloning theorem:


Doesn't "clockwork" imply that true randomness can't exist and quantum physics tell us that it does?

I don't think clockwork implies determinism, only that rational actors can understand cause and effect in the system. Much like a window into clockwork.

That said, i do appreciate the horror of being a wound up automaton, but i think this is a mistake to think of humans as primarily rational. Hubris, even, might be a good name for it. We are influenced by thousands of arbitrary signals every day and react in mostly normal ways; this doesn't seem to scare people at all. That is: even if we have a full mechanism of free will, we don't have functional free will, as humans are clearly influenced by many unconscious factors.

I'd have to look at Jefferson's writings again to confirm, but i thought "god" in a deism sense was portrayed, in many senses, as a scientist himself watching the progress of the universe, not an author laying out a grand, anthropomorphic, meaning-laden story. That is: even the clockwork deists were not explicitly deterministic!

> true randomness can't exist and quantum physics tell us that it does?

Not necessarily. Quantum physics says that certain things are unpredictable. I consider indeterminism separate from the "truly random", with the former being generated via an algorithm and the latter being produced via no algorithm whatsoever. It may be the case that since the measuring system (the experimenter) is necessarily entangled with the measured system, the information needed to perfectly predict eigenstate selection is inaccessible, and thus appears perfectly random, since it is formally unpredictable. What would be interesting though is if someone devised an experiment to ascertain whether eigenstate selection is predictable in a universal sense although not necessarily predictable to us since we're part of the system that is being predicted. How such an experiment would be devised is beyond me though.

Well, here's an argument in favor of true randomess:

If there was a PRNG deciding the outcome of measurements, it would have to have the mathematical moving parts exist somewhere as hidden variables. Hidden variables are outlawed by Bell's theorem and a couple of other things. Therefore, the randomness must be stateless and truly random.

> Hidden variables are outlawed by Bell's theorem

Local hidden variables. I see nothing wrong with non-locality; it's basically been confirmed beyond a shadow of a doubt by now. That "closeness" occurs within Hilbert space rather than its projection into position space is ultimately no problem for a PRNG.

Let's talk about something simpler first - about movies. We know enough about chemistry of photographic film to conclude that movies are simply fantastically complex clockwork automatons, correct?

Incorrect, movies are not automations, they are simple data sets. Movies+players combination is, yes. But this definition from the parent comment is an oversimplification. This is how the article framed it and its TL;DR:

"living systems may be defined as open systems maintained in steady-states, far-from-equilibrium, due to matter-energy flows in which informed (genetically) autocatalytic cycles extract energy, build complex internal structures, allowing growth even as they create greater entropy in their environments, and capable, over multigenerational time, of evolution."

So Yojimbo and Wild strawberries are simple data sets. Where does inspiration come into play?

I think the difference between life and machines is in how entropy is shifted. Life goes on while, all machines degrade over time.

Agreed. A film and its corresponding image management toolchain is a fantastically complex automaton.

Having made or worked on quite a few films, I can assure you that they don't watch themselves, and that if one has done one's job correctly they just repeat the exact same behavior every time. I would be quite interested to see a movie in which one of the characters decided to do something different from the previous viewing.

Pierre Menard could have made a movie. But even leaving that aside, and assuming for a moment that replays are indistinguishable (and one can enter the same river), would you accept that a movie is reducible to a photographic film chemistry?

Oh sure. I'm disagreeing with the idea of it being an automaton qua having some ability to adjust to interaction rather than being static like a book.

I don't know why you leapt to this thing about the film watching itself. The author did not ask me to assume this.

On an evolutionary scale, life exploits the boundary between signal and noise to migrate across its manifold substrate. There is no reason to presume that statistical interference does not contribute to outcomes more locally.

Are you saying that, while it appears that we understand the basic mechanics involved, there may be non-obvious interactions that have additional explanatory power?

It is common for people to assume that chemistry works mechanically, and we have, in fact, been very successful at reducing a great number of biochemical interactions to pretty much raw mechanics.

Also, life has a stake in reducing a stream of inputs into 'gated' processes, because life exists by providing a stable interface for its replicator core.

However, this doesn't mean that the stream of inputs in all cases reduces to 'gated' processes.

I agree that most of what makes up life is a grand, replicating automaton, but I think there is a soft core to it that retains the controlled unpredictability of a solution in brownian motion or a population in genetic drift.

Please don't post nonsense on HN.

I've replied elsewhere. This comment was extremely rude.

I did not find it nonsensical in the least.

It looks like postmodernist fluff to me. I assume from the downvotes it received that I'm not the only one. I accept the possibility that we all missed some great insight - would you care to give us a hint?

I started writing an explanation but it got really wordy and I don't want it to sound patronizing. Better, have a look into Catastrophe theory, and consider lifeforms (and species) as coordinates in higher-dimensional parameter space, wherein informational asymmetries are equivalent to energy discontinuities.

A simple practical example of this would be bacteria around a hydrothermal vent; it's the boundary between the high energy source (signal) and the low-energy empty space around it where life thrives.


Although I am not familiar with catastrophe theory by name, I can provide a specific example of a 'controlled randomness' that governs genetic drift. It's a particularly nice example because we use it to seed random pools of our own now, too: cosmic radiation.

TL;DR: living systems may be defined as open systems maintained in steady-states, far-from-equilibrium, due to matter-energy flows in which informed (genetically) autocatalytic cycles extract energy, build complex internal structures, allowing growth even as they create greater entropy in their environments, and capable, over multigenerational time, of evolution.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact