Constructor Theory Solves the Riddle of Life
aeon.co
aeon.co
From my point of view, chemistry more or less explains how cells work, and we don't need a quasi-platonic explanation of how the universe is designed to allow for a grass-eating goat machine to explain this at all.
The author appears to be one of the primary proponents of so called 'constructor theory'. I dug a bit into their research and it's setting off all of my skeptic alarms. Their primary research papers appear to be typeset in Word for christ sake.
To nitpick some particular aspects of the theory and article:
From the wikipedia article on constructor theory:
"Current theories of physics based on quantum mechanics do not adequately explain why some transformations between states of being are possible and some are not. For example a drop of dye can dissolve in water but thermodynamics shows that the reverse transformation, of the dye clumping back together, is not possible. We do not know at a quantum level why this should be so" [1]
An egregiously incorrect statement. This is basic statistical mechanics. The dye dropped into a solution diffuses and doesn't return back into the original drop is very simple: there are very few configurations of the dye/solution system where the dye is in a concentrated drop and a mind bogglingly large number of configurations where the dye is diffused throughout the solution. Nothing in physical law forbids the formation of a concentrated drop from a diffused solution - it's just ridiculously implausibly unlikely.
In fact, due to time reversal symmetry we know that the formation of the concentrated drop is explicitly allowed. Here's a little gedanken experiment: put a drop of dye in the solution and let it diffuse. After some time t, reverse all of the momenta of all of the atoms in the solution (you are God). At time 2t the dye will reform in the original concentrated droplet. But here's the zinger: keep watching and that droplet will again diffuse across the solution! Keep letting the clock run backwards and you'll probably never see the drop reform again.
One more note on that quote: the last sentence says that this is somehow forbidden by quantum mechanics. Again, this is bogus. Ensembles of large number of particles is the domain of statistical mechanics! Long discussion in of itself but the (semi)classical view is a very useful and valid perspective for many problems!
Now a comment about the article:
"Here’s where the puzzle arises. Biological replication and self‑reproduction are in fact such stupendously well‑orchestrated physical transformations that one must explain how they are possible under the simple, no‑design laws of physics such as ours. This additional explanation, which was not included in the theory of evolution, is essential for that theory to properly explain how living things arise without intentional design – to close the explanatory gap"
1. Absolute bullshit that one needs to 'justify' how certain physical states that _seem_ improbable are realized. Total garbage. Another core tenet of statistical mechanics is ergodocity [2] which states that ALL states of a system are will eventually be visited (if you have infinite time).
2. My major gripe: (non-equilibrium) statistical mechanics accounts can explain the formation of complexity (such as life)!! It is not mysterious or outside of the scope of our current framework!! By analogy consider a pot of water being heated by a hotplate from underneath: if the hot plate is on low the heat will diffuse up through the water in a disorganized manner; but, at some critical temperature convection rolls will form spontaneously in the water. They form because the increase in energy dissipation wins out over the reduction in entropy induced by their formation.
This is exactly the mechanism driving the formation of life! The existence of life is not mysterious, it's inevitable!
This article is complete and utter bullshit.
[1] https://en.wikipedia.org/wiki/Constructor_theory#Motivations
To take this further: you can physically do this with a similar system (replace diffusion with stirring) for very low Reynolds number flows, which are time-reversible. It's called the Taylor-Couette experiment, you can see it in action here: https://youtu.be/k7ZZtxdtmeQ
Statistical mechanics does not explain the existence of the life that currently exists. At most it proves complexity can come into existence. But not every state of the system has been visited and we don't have infinite time (and whether a billion years is sufficiently 'infinite' for some purpose first needs to be proven). The question is: is this life likely?
I don't remember the name, but there was a computer architecture journal which had papers printed with dot-matrix printers (this was in the 80s). I would not discount the value of work due to this.
to me it became much clearer when i read somewhere (just dont' remember where) along these lines - imagine a "stick" shaped pretty simple organic molecules with one end being hydrophobic and the other hydrophilic (there are a bunch of such molecules known in organic chemistry). Having a soup of those molecules in water, some of the molecules would naturally clump together with hydrophilic ends outside and hydrophobic ends inside - kind of a small bubble. Now if on contact of a bubble with another molecule, the molecule is added into the bubble (similar like a bunch of magnets clumped together "assimilate" another magnet piece correctly orienting it), the bubble gets bigger and at some point it would naturally splits into 2 smaller bubbles (surface tension, etc...) and each bubble would continue the "assimilation"/growth and division in their turn as long as there are such molecules around... Of course it isn't even "pre-cell", it is just showing how natural physics laws can drive the initial pre-life stages.
There will be a continual cascade of chemical changes at the vent, and if some random chemicals appear that increase the chances that each other are formed in that cascade, a self-reinforcing metabolic loop could occur.
The metabolic loop could optimize itself as it occasionally produced slightly more efficient self-reinforcing chemistry. Eventually this optimization might take advantage of self-organizing membranes to compress the metabolism into individual units, i.e. the first cells. At that point we would have "life".
Maybe that happened. Maybe not. But it is an alternate way for life to start in which distributed replication occurs before single replicators.
I consider this issue pretty much answered by Chapter 2 of the Selfish Gene, "the replicators". Link: http://www.evolbiol.ru/gene/index_2en.html#02
No one has ever presented an actual theory of abiogenesis, the current article notwithstanding, as opposed to some intriguing but speculative hypotheses that might (or might not) turn out to be part of an eventual theory (RNA world, clay catalysts, deep sea smokers, lipid membranes, etc. etc.)
Considering a totally open subject to actually be answered and finished is just saying you personally don't care to pursue the topic in more depth.
And thing is that laws of physics don't have it in themselves to give birth to processes when by 'process' we mean something more in line of a function, something that would not allow us to say that the Sun is a living organism, although there's processes happening inside a star.
From that point of view, there's no need to invoke processes, nor entities like organisms or cells: what's being reproduced are the genes to reproduce, and the laws governing it are just mathematics, like with a prion or a virus.
Looking forward to reading comments that disagree with this disagreement.
I think what it's trying to say is that constructors are resilient (i.e. long lived, through virtue of replication) 'structures' expressed in matter capable of stably holding and repeatably acting on digital information. And once you have a basic constructor, assemblages of these can provide successively higher levels of organizational power. Their digitally error-corrected resilience differentiates them from flash-in-the-pan analog chemistry, and once you have that longevity of structure you can start to have life. I think.
I don't know. I mean, yes, intuitively, life as an emergent property of recursive layers of self-organizing structures makes more sense than life as the continuation of a random walk through chemistry, but whatever constructor theory is, it's not deeply explained here. What are its implications? How does it actually compare to alternative theories?
Constructor Theory: All things not explicitly forbidden by the laws of physics can exist. The answer to "Why does life exist?" is: "because the laws of physics do not explicitly prevent it."
Seems like an interesting take. I'm not convinced that it's at all useful, but it's not obviously flawed or vacuous.
Sidenote: I've never felt that the existence of life was as mystifying a physical problem as the existence of black holes, or expanding space time, or stars. Why people are mystified by life but blithely accept the existence of distant fusion reactors the size of our solar system is beyond me.
Yeah, I don't see this as all that useful, either. It also still doesn't in any way advance discovering what process brought things into being, it only accepts that it happened, which everybody already accepted.
Constructor theory says, in short: "Yep. We can also accept the existence of space unicorns, for a physically possible definition thereof. Probably not on the far side of the moon though, since we've been there and would've seen them."
That sentence just skipped over the most important part of the whole question! :)
Either you accept that the initial state that lead to life was constructed, or you accept that it was not constructed and are faced with the problem that constructor theory intends to address. These theories attempt to explain why the initial state for the universe was conducive to life/stars/black holes/whatever - everything I described is long after the fact.
I'll assume it means this:
//Old theory
u = new Universe();
//TODO: populate...
//New theory
r1 = Replicator.getInstance();
r2 = Replicator.getInstance();
r3 = Replicator.getInstance();
u = new Universe(r1,r2,r3);
>In constructor theory, physical laws are formulated only in terms of which tasks are possible (with arbitrarily high accuracy, reliability, and repeatability), and which are impossible, and why – as opposed to what happens, and what does not happen, given dynamical laws and initial conditions. A task is impossible if there is a law of physics that forbids it. Otherwise, it is possible – which means that a constructor for that task – an object that causes the task to occur and retains the ability to cause it again – can be approximated arbitrarily well in reality.
This seems to be false, as there is no physical law preventing entropy from decreasing but there is a physical law which prevents the existence of a 'constructor' capable of decreasing entropy.
>Moreover, it is a fundamental idea of constructor theory that any transformation that is not forbidden by the laws of physics can be achieved given the requisite knowledge. There is no third possibility: either the laws of physics forbid it, or it is achievable.
Either they are saying something is achievable if and only if there is a constructor for it, which is false (see previous point), or they are claiming something is achievable if and only if the laws of physics don't forbid it, which makes this 'fundamental idea' vacuous.
But where it sounds like this is going is: If you don't know it's impossible, then it must be possible to build a constructor for it. That assumes that everything that we don't currently know is impossible must be possible. That's... quite an assumption.
Worse, that assumption is swept under the rug rather than explicitly stated.
Or I could be misjudging their position...
I also suspect you'll encounter something like the halting problem when trying to figure out which constructors aren't constructable. At worst you'll encounter constructors where this question is unanswerable (e.g. can you construct a constructor, which constructs all constructable constructors that can't construct themselves?).
A constructor is like a program. In this case, the rules of the language are physics itself. We're familiar with writing simple programs in C. We can imagine programs, like the halting problem solver. It turns out that program can't exist, but you kind of have to think about it for a while to understand why.
A constructor (as i understand it) is sort of hand rolled assembly (matter) running right on physics. Perhaps there are space unicorns on the dark side of the moon. We can imagine that. It's not really clear if physics actually allows that though. My gut says no, but the rules of the game, much like the rules of the language may allow such a collection of matter to exist.
I mean, quines are pretty weird. I wouldn't have thought that was even possible.
But I for one look forward to worshiping the Great Constructor.
But fortunately, as you point out, not everyone stops asking questions.
Fundamentally the issue is that constructor theory seems to require us to write down every possible constructor and gives us no way to validate whether they will work, not particularly useful from a theoretical perspective since it is essentially equivalent to having to do an exhaustive search of the space. True, but tautological and rather useless since it doesn't provide any insight into the relative importance of initial conditions (or previous history) and the fundamental forces in delimiting what is possible in any given universe.
We want to be able to describe the probabilities of any particular configuration of matter coming to be at some point in time in the universe as a function of the initial conditions and the forces that govern it (statmech can sort of do this for simple systems). This simply assigns probability 1 to everything and leaves as an exercises for the reader how to check whether such assignments are consistent with the laws of nature.
this statement is strictly untrue. it seems to assume that this behaviour is not going to be selected for, when very trivially it is...
natural selection isn't special to life... it applies to everything, again for obvious reasons.
its a very, very simple consequence of existence. those things that survive longest survive longest. its so simple its hard to explain. its no more complicated than the action of a sieve, which stops clumps from going through due to physical constraint - a form of natural selection.
As I understand it from this article, it is saying: "Once you invent the universe, then an apple pie will exist... given enough time." ?
An interesting conjecture but not exactly provable! I assert that a universe is a necessary but not sufficient condition for apple pies.
Given a universe, in which the exact arrangement of particles that make up an apple pie is not physically impossible, it is physically possible for an apple pie to arise through the evolution of the universe, without requiring 'outside input'. (i.e., design or influence by some entity outside the universe).
Furthermore, while it would be stupendously unlikely for an apple pie to arise completely by chance (imagine the odds of all the right atoms just happening to collide in the right way at the right time), it's not nearly as unlikely that some kind of self-replicator should arise, and this self-replicator could eventually generate an apple pie, or something of similar complexity.