Multi-Cellular Evolution in a Petri Dish...?

Lee Cronin on Assembly Theory

What Assembly Theory does it allows you to trace back all the interactions between objects that give rise to other events. Right, at least probabilistically you can say right if I've got this molecule, and I chop it up on the shortest path, and I want to take the shortest route to make this molecule with these bits, what do I do including reuse? I add this bit to this bit, I reuse it okay because I've got memory, and therefore what Assembly Theory does it says what is the minimum amount of memory I need to have in the universe to make this complex object? And if that memory exists for more time than to make one of them, I can make many copy numbers. So if that memory exists for more than making one object, the more complex the object, the more interesting that memory is, because that memory has to really exist. Whereas if the memory only exists to make one object, and then the memory is gone, then that's what you call a random ensemble. In fact, it would be the same. So what Assembly Theory allows you to do is to show how you can build things up by understanding that the causal interactions or the "contingency". So then this is kind of like a very almost Evolution essay: For this to happen, this needs to happen, this needs to happen, this needs to happen, this needs to happen... in this sequence of events. And so the fact you're able to go through those sequence of events and get your thing, you can think about how weird that thing is the more steps you've got because it could have gone off in one of a zillion different directions. And so it allows you to appreciate how unique some things are, and whereas entropy just says "Ah this is the average, this is my Ensemble.

So with and that's classified as the Assembly Index?

The Assembly Index is classified at the moment as the number the number of, um, the shortest the number of steps on the shortest path to construct the object from the basic building blocks.

Just a technical question so I understand that. Why is it the shortest path, so I understand? That that's a lower bound.

Now everyone asked me, that yes, you know it was coming, well actually, I mean this is, it's really funny when you invent a theory because like remember Assembly Theory could be complete nonsense. And so, you know, I don't think it is, but it's like I don't know, it seemed good. It's just like you know did ever, no there's actually more fundamental reason. So my reasoning would be, if you take an object, what it does, and you basically reduce it to the shortest path, first of all, that is a that is a finite quantity. That is a quantity every object has a shortest path to make it. So what that means is, if I come across that object I know the minimum number of steps I must make to make it. That doesn't say, yes only that steps, because I'm sure things are made in Long objects, like how many people do we see being really inefficient? Well we can say don't do that do that right, but what I'm saying is like, so for literally you know, for me to be able to. Maybe I'm typing, type a sentence right, for me to type a sentence, or come up the word "Abracadabra" or "banana" or something, I say right, this is the minimum thing they need to do to make that string, or that thing, and and I can count those up. And the larger that number, the more improbable it is. And so, what, by having the short by going for the the shortest route, it gives me a nice Baseline to say, Look shorter, you can't go shorter than shorter. So you and so you'd be you'll then have some confidence that something is odd, if it's over your threshold. But of course the ways that the objects get assembled here you know, it was probably an average. There's probably other things there. You can spend a lot of energy to go beyond that, but really it seems the shortest path is incredibly significant because of the statistical meaning of those objects coming together in the universe...

I think the shortest path is about, really, the minimum amount of memory the universe needs to have to make that object.

...Intelligence is the way in which selection "wants to do its' thing"!

I think it's like it's so funny, that it's like the answer to everything, "a selection". It has to be, there is no other reason. So like the selection, why do that selection? You know? But I think as a phenomena,  human beings display an incredible amount of cognition and intelligence. They're able to build machines, do mathematics, write poetry, communicate to one another, cheat one another, you know, build fantastic machines.

----

Chemputation - A kind of a concept where we have an abstraction, a kind of prescription ontology for chemistry, and then a practical ontology. Where we're able to basically take that prescription and then program a robot to then make that chemistry happen. So make a drug, make a series of compounds, make a formulation. And really chemputation is about taking a physical input, being chemicals or some other stuff, some code, and then producing an output that's reliable every time.

And actually I wanted to build a thing called the chemputer, which does chemputation. Or arrays of them, to like ,help crack the origin of life. So I can literally imagine like trillions of chemputers all doing slightly different experiments, with different selections. No one would fund that right? 

So I do it, basically, and said, "okay I'll make drugs and robots for drugs, and all that. What can go wrong? And chemputation, actually was the verb I gave to the word, because as one organization decided to trademark the word "chemputer". And I was like, Ugh, you know, and then so they trademarked it, and every time I used it they started telling me off. I was gonna be using it, they owned it. I was like, okay, but then I thought, okay I'll just... and then that inspired me to say what do chemputers do? Oh, they do chemputation like computers can do computation. Okay, what is that really? Let's, here's a chemputer, just a gimmick, a brand name that deserves to be nothing other than trademarked, you know, like whatever. You know, like some kind of commodity. 

And I was like, well, actually no, a chemputer is a generic chemical... it's a chemputer is a chemical engine which runs on a programming language, and chemicals, to make desired molecule outputs. And so, and then, chemputation is the process of doing that. So I thought it was quite good. 

And when I started figuring this out, they want to make a programmatical chemistry. My colleagues just told me I was kind of bananas, right? And it's never going to work, and so the more they told me it wasn't going to work, the more I just raised more money to do it. And yeah, we have a a bunch of compute.. chemputers or "chemputational systems" I should say in my laboratory. And they all run the programming language, they all have been used to do organic chemistry and organic chemistry formulation science, and what we're busy trying to do is to unify some of the code base to kind of make it easier. 

And where are we right now? Well we're able to make quite complex molecules. We're able to do Discovery. We're able to look at unknown things. We're able to look at inorganic materials and we're able to make, I don't call you know, the Holy Grail, which is like a closed loop lab where you just put code and molecules in, and really cool molecules and the code has come out and the code is consumed. But that's chemputation in the nutshell. 

The thing is, right, it's, I don't know, I think it's remarkable. I think it's kind of.. it's hard because chemistry is hard, and it's dangerous, and things fail. And I don't know, people would say, you were never going to get the, you're never going to get things working the way you want because it's just too unreliable. So we work super hard to try and create the correct process description that would do most of chemistry, and I'm glad to say we've got there. We've done hundreds and hundreds of reactions now, we can reproduce work, we can basically make complex molecules, the chemputation code is now used on the origin of Life rigs which I've built it for, I've built, build Dynamic logic in, and also, I use assembly Theory to measure. So kind of all these profound little projects are all working together, they're all trying to solve one problem we should try, and make an alien. Yeah well I mean also the drug development applications are pretty interesting, yeah? 

So I mean I started a company and the company's dream is like, it's like, the company's name is Chemify. It's like the AWS for chemistry I've kind of seen, you know, Amazon were doing all this stuff. But the thing that just grew and grew at Amazon was just all this compute and what they did with it. I mean it's fascinating. And so Chemify is doing the same thing, AWS chemistry... make the infrastructure for chemistry so everyone gives each other a code that runs in a robot, or very, very minimal robot-human being working on the bench, and you can just interchange molecules and make, discover, new drugs, make new materials, sell code I guess, and just really speed up the process of innovation at that boundary, because chemistry and Material Science takes ages.