What Is Life? (1944) [pdf] | Hacker News Reader

What Is Life? (1944) [pdf]

dlab.clemson.edu

109 pointsMichaelAO24 comments

context: https://en.wikipedia.org/wiki/What_Is_Life%3F

"A New Physics Theory of Life"

https://www.quantamagazine.org/a-new-thermodynamics-theory-o...

HN: https://news.ycombinator.com/item?id=13103215

"An MIT physicist has proposed the provocative idea that life exists because the law of increasing entropy drives matter to acquire lifelike physical properties."

"Life does not violate the second law of thermodynamics, but until recently, physicists were unable to use thermodynamics to explain why it should arise in the first place. In Schrödinger’s day, they could solve the equations of thermodynamics only for closed systems in equilibrium. In the 1960s, the Belgian physicist Ilya Prigogine made progress on predicting the behavior of open systems weakly driven by external energy sources (for which he won the 1977 Nobel Prize in chemistry). But the behavior of systems that are far from equilibrium, which are connected to the outside environment and strongly driven by external sources of energy, could not be predicted."

or: http://nautil.us/issue/34/adaptation/how-do-you-say-life-in-...

I flirted with this idea for a long time after hearing it and wanted to like it but in hindsight it's pretty easy to prove it wrong.

Yes entropy creates pressure on life and has a hand in pretty much everything, but Earth has an abundance of free (Gibbs) energy that insulates it and creates a new kind of environment and conditions.

It's like saying that hunger drives social complexity. It's part of it but it doesn't drive it.

Yes you could apply Newtonian physics to everyday social encounters and structures but they're way too broad and generic to derive meaning conclusions at the interpersonal level.

After reading the Nautilus article, my understanding of England's 'dissipation-driven adaptation' is: a physical account of why life-like structures would likely occur (and persist) even at levels far more basic than what we would call organisms. The basic structure (of his system) follows that of natural selection and evolution in that there is an element of randomness constructing forms, and something environmental which determines fitness of competing forms—but it differs in what constitutes fitness. In England's theory, we should expect irreversible, dissipative systems to persist simply because of their irreversibility; if they could be reversed, they would return to an indistinguishable part of the background from which they emerged.

If his theory is true, there is a natural 'inclination' at a more basic material level for life-like structures to emerge, so the probability of life evolving 'randomly' is more likely than we had reason to believe before.

Furthermore, it gives us more general criteria for attempting to manually 'evolve' life-like systems, since we mimic life a more basic level than genetics (aiming to produce things with high likelihood of being irreversible).

(I'd appreciate any correction on parts I may have gotten wrong!)

I've not studied this, but doesn't that theory seems to conflate "likely to persist" with "likely to emerge"?

For example, the US prison population tends to be far more violent than you would expect - in my state's prisons more than one out of four prisoners are there for either homicide or rape. This is because the longer sentences for those crimes. Because of their "persistence", there are sixty times more people in prison for thirty year crimes than for six month crimes, compared to the relative conviction rate for those crimes.

However(!), this persistence tells us nothing about the base occurance rate. There aren't sixty times more murders and rapes than six month sentence crimes.

So it also seems that the persistence of life does not affect the probability of life's arising in the first place.

Or am I missing something?

I think it follows the same basic pattern as biological evolution: there is a semi-random process that introduces new forms, but when you see a preponderance of some particular form (or attribute) it's simply due to the fact it wasn't eliminated like competing forms/attributes were.

I think what's missing from my description though was an account of how the dissipative efficacy of the non-reversible formations compounds in time. I re-checked the article and it is IMO vague on the subject aside from saying that that does in fact happen. In any case here's a relevant quote by Jeremy England given in the article:

While any given change in shape for the system is mostly random, the most durable and irreversible of these shifts in configuration occur when the system happens to be momentarily better at absorbing and dissipating work. With the passage of time, the “memory” of these less erasable changes accumulates preferentially, and the system increasingly adopts shapes that resemble those in its history where dissipation occurred. Looking backward at the likely history of a product of this non-equilibrium process, the structure will appear to us like it has self-organized into a state that is “well adapted” to the environmental conditions. This is the phenomenon of dissipative adaptation.

Doesn't that just say roughly that "a new form that is more likely to persist in a given environment came from a previous form that was more likely to persist in that given environment"? That almost boils down the same thing as before. The persistence of life still seems entirely orthogonal to the odds of a self replicating life arising in the first place, right?

I once heard someone say something like "a molecule is 'biological' if it describes the environment [for its organism]." The key takeaway from my crude recollection of the quote is that "information" is a kind of master theory (which, tangentially, makes it a meaningless theory - the hypothesis "this event occurs because of information" is true 100% of the time).

Let's suppose for the purpose of a thought experiment that abiogenesis occurred because of specific conditions during or shortly after the Big Bang. Increasing entropy is the direction of the arrow of time, but there would be no such direction of time if the universe were at its theoretical maximum entropy, heat death - all would be indistinguishable. So, the Big Bang must have had lower entropy than the cosmological inflation that followed it, which was a stage of hot gas that was actually pretty much at maximum entropy for its volume. The entropy of the early universe only increased by increasing its volume into an open space.

But then.. - time moves in the direction of increased entropy.. - the universe started in a highly ordered state.. - ..so the uniform ball of gas became highly varied as a result of time-symmetric physical processes..? - ..so whatever non-symmetric, entropy-increasing process made the differentiation of regions of the universe likely would also gave rise to an astonishingly rare process that appears to achieve local entropy decrease by dissipating it into its surroundings?

The chance anything as intricate as life (or any type of 'information', really) would result from total uniformity is perplexing. Let alone that biology can't exist if environments were not yet a part of nature, and environments only came to exist as result of universal spontaneous symmetry breaking.. For all we know, the laws of physics were themselves also subject to cosmological natural selection before or near the Big Bang [1]. The laws we know now would just be the ones that lasted - maybe because the others underwent some kind of self-annihilation, or maybe because the 'successful' laws we know now in some way reconstituted the cosmos, so that the others were no longer propagable and vanished. Apparently wave function collapse is a time-irreversible process, though it happens because of the 2nd law of thermodynamics rather than causing it.

So, what if self-replicating life seems transcendental because of our assumptions about the distribution from which the parameters that we call "the laws of physics" come from?

In my opinion, we typically assume the laws of physics come from a stationary ergodic process, which means

- the properties of the probability distribution of the process generating the laws of physics don't change over time

- the process producing the laws of physics is ergodic, roughly meaning there'd be no disagreement between a time-averaged (from before the big bang up to now) estimate of physical law from an individual observer/experiment/sensor, and a group-averaged estimate of observers/experiments/sensors at one particular time

I have no real knowledge of wave function collapse, but time-irreversibility, I assume, would imply non-ergodicity. There are various types of non-ergodicity, though. I guess I also could have eliminated much of what I've written by just relying upon the 2nd law of thermo already being suggestive of non-ergodicity.

All that said when you place a bet on a small probability event, every source (e.g. cosmological natural selection) of higher-order uncertainty (your uncertainty about your uncertainty, ... etc.)) tends to increase the payoff of the bet [2]. Not because the uncertainty causes the event to happen more or changes the bet's reward, but because the observed probability in the past is less likely to overestimate something in the future that is rare, than it is likely to underestimate it. And the more unclear the relationship between past factors and future sources of uncertainty, the less underestimated the future event. So to consider sources of higher-order uncertainty that predate abiogenesis is crucial to correcting estimates of life's prevalence at all.

[1] http://bigthink.com/in-their-own-words/cosmological-natural-...

[2] https://arxiv.org/abs/1209.2298

I've long thought "life is inevitable". Why? In a universe where every possibility plays out, some outcomes are bound to lead to life. [Diverse] Chaos => universal turing machines occurring somewhere + random codes being executed + sufficient computation => life.

I'd love to see these notions made rigorous. Somehow they appear obvious to me.

Order (i.e. life) is necessary to reach the ultimate unordered state (i.e. a cosmic soup of energy)? Now that's a pretty non-trivial statement.

Video by the author: https://www.youtube.com/watch?v=e91D5UAz-f4

This is a beautiful, romantic view, though I don't see how it can me made all that physically rigorous interpretation.

Here are some thoughts on the philosophy of what you say.

> Life is inevitable

On one level, this is a tautology: the physics of the universe are such that life has formed therefore life occurs with 100% frequency in all known universes.

On a different level, it's not too hard to build "physics" that are too trivial to even encode Peano Arithmetic much less "life" as we would identify it. Thus life isn't obviously inevitable in all physical models of the universe, so we can take the emergence of life as an "experimental constraint".

On a personal level, I too feel the importance of life. However, I don't think any of that special feeling is compromised by taking life to not be somehow "universally fundamental" in some way.

> Life is a fractal

The mathematician in me kind of flinches at this a bit. :p Fractals are super cool though! But, heck, the eastern coast of Cambodia exhibits fractal-like qualities too. As do coupled-pendulum setups and the x86 execution pipeline.

This line of thinking feels to me like putting The Mysterious into some spiritually special place in our thinking.

For myself, when I think something is "obviously true" I usually find that I just don't know enough to appreciate the intricacy of what's really going on. Dunning-Kruger and all that.

Anyway, just some 2 cents.

> In a universe where every possibility plays out

Why would we assume this is the case? Certainly an infinite number of possibilities have already been ruled out, like the possibility that the entire universe would be totally devoid of all matter.

> In a universe where every possibility plays out

I think that's an assumption.

When you have insights that solve no problems, they are not actually insights worth having.

This is a book full of good questions and poor answers.

Or, as the biologist Max Perutz more scathingly put it, "Sadly, a close study of [Schrodinger's] book and of the related literature has shown me that what was true in his book was not original, and most of what was original was known not to be true even when the book was written."

I think HN should have a rule that no one makes a general critique without including some specifics.

Here's a reference I just found without any effort, from an occurrence of the above quote in Freeman Dyson's Origins of Life.

GP comment does not deserve to be downvoted: a solid searchable reference was provided.

https://books.google.ie/books?id=1Fqsd-v4LcwC&pg=PA3&lpg=PA3...

I'm not the person you're replying to, but I also didn't think much of the book. I read it 13 years ago, so it's hard for me to remember specifics of my criticisms.

I'm cool with that critique of my answer and will be more specific next time I post such a critique.

Interesting. He argues his way to this principle of genetics:

> And the gene is most certainly not just a homogeneous drop of liquid. It is probably a large protein molecule, in which every atom, every radical, every heterocyclic ring plays an individual role, more or less different from that played by any of the other similar atoms, radicals, or rings.

He seems not to have known of (or at least does not cite) some of the points listed in the Wiki summary of the history of DNA research[1] which would have been extant in 1944. Watson/Crick/Franklin published in 1953, and Schrödinger (d. 1961) must have been delighted.

[1] https://en.wikipedia.org/wiki/DNA#History_of_DNA_research

Most of this book should be read from a historical point of view. It is also great to read thoughts written by the scientific minds of the 20th century, whose thought processes often went much deeper than today's scientists, trapped in a miscalibrated incentive system can go.

However, if you read just a little of "What is Life?", read chapter VI on Schrödingers idea how living matter is related to negative entropy / information. This is said to contain remarkable thoughts until today, and I met two professors in computational biology by now who told me how this chapter inspired them at the beginning of their career.

To those who enjoyed this one I strongly recommend Heisenbergs "Der Teil und das Ganze".

This is a brilliant book, clear, succinct and insightful. Part of the canon on our understanding of what life is / what it means to be animate instead of inanimate.

Other books on the intangible quality of being alive:

- At Home in the Universe by Stuart Kauffman

- The Timeless Way of Building by Christopher Alexander

- Gödel, Escher, Bach by Douglas Hofstadter

Slight tangent. Some of the comments here made me think of a great video on entropy vs. complexity by minutephysics:

https://youtu.be/MTFY0H4EZx4

It's actually part 3 of 5 from a short miniseries with Sean Carroll. I recommend the whole thing if you are interested, it is quite well done.

Wow! That's a great series. Thanks for sharing it. It really changed how I view the world.

My pleasure!

Data processing inequality would suggest a breakdown in complex structure as entropy increases.

Reply on news.ycombinator.com