Chris Kempes
speaker
209 appearances
1 recordings
1 series
first heard Jan 2025
last heard Jan 2025
Chris Kempes’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
So that's the sort of total energy budget the cell is the metabolic power. And then scaling is just this idea that if you look across orders of magnitude in one feature, you get a linear relationship in orders of magnitude of another feature. And those could have different exponents, which would be the slopes of those different curves in that log-log, order of magnitude, order of magnitude space.
Mm-hmm. And so when we say there's a shift in a scaling relationship of metabolic power, what I'm really saying is that as you go across this evolutionary divide, you see a change in the exponent from one class to the other. Specifically, in prokaryotes, the exponent is greater than one, meaning... if you double in cell volume, you more than double the total metabolic power.
If you go up by a power of 10 in cell volume, you go up by more than a power of 10 in metabolic power. In Unisar eukaryotes, the exponent is sublinear, meaning that if you go up by an order of magnitude in cell volume, you go up by less than an order of magnitude in in metabolic power. So the power per unit volume is increasing in prokaryotes and decreasing in unicellular eukaryotes.
And that's a really fundamental difference that drives all sorts of downstream things that we can observe, like the growth rates, like the requirements for the number of ribosomes in the cell, all sorts of different features.
Exactly, exactly. You still need more power, but in a sort of gained efficiency way.
Yeah. So it looks like the nucleus is a more complicated story because I think the nucleus is mostly about regulation and separating out certain parts of the genome from each other and protecting the genome from reactive parts of the cell. But the mitochondria seem to just come along for the ride of that. So there, I think there's a big debate about causality. People would say,
Well, the number of mitochondria have the same scaling as the metabolic power. So are they driving the metabolic power? And I would say we don't quite know yet. We don't know what's being optimized because it could be that the number of mitochondria you have are just optimized according to some other constraint.
And that constraint is what's driving metabolic power and mitochondria come along for the ride. Or it could be some fundamental constraint about the mitochondria that require the metabolic scaling to be what we observe. But yeah, I mean, in general, that's exactly the sort of connection we try to draw. We say we have a physical constraint. Maybe it predicts something like total metabolism.
And then once we know total metabolism, we can write down a cell model. You know, we can write down a model of cell physiology and ask what other components should scale in what way according to this metabolism. And in bacteria, that gives us a whole host of predictions and the same in unicellular eukaryotes. Yeah.
Absolutely. Yeah. And I think, you know, I think that's something we're starting to run up against more and more recently. I mean, it's interesting that, you know, in the early days of computer science, a lot of the considerations were really about, well, how much resources will this need as we scale up? So how much time, how many CPUs, how will this scale up in a resource constraint?
And then we got to a point where we were mostly interested in, well, just artificial life in the computer. We had enough resources. So then it's all just about how do you get the right sort of evolutionary dynamic? How do you get open-ended evolution? How do you get things that build up higher layers of structure and function?
And there's a really wonderful bit of artificial life that was done there. And now with the emergence of AI and LLMs, I think we're back to really wondering about scale and constraints, right? I mean, there's a huge debate there about what will happen if we make the neural nets 10 times as big and how many data centers does that take?
And so I think we're back to thinking very clearly about scale and about ultimately the physical resources that you need to support artificial intelligence or artificial life or really sophisticated computer programs.
It's a great question. So recently they've been called for some of those reasons that you're mentioning, Sean, in terms of the problems that raises. Recently, these transitions have started to become major transitions in individuality. rather than major evolutionary transitions. I haven't heard that one. That's good.
Yeah. Now, I think individuality is also a complicated concept. And so then we're into talking about, well, what is an individual and how do you define that? And people like David Krakauer at SFI work on entire theories of individuality. And that's a frontier area in and of itself. So I think that reframing doesn't quite solve all the problems with how do you know one? How do you spot one?
What's the right metric? Right. I think generally what people are thinking about is you have extra layers of architecture or hierarchy where you've packaged together more lower level things into some higher level thing. And where that higher level thing in an evolutionary sense operates as a whole. So the selection that matters to it, the evolution that matters to it is really on the whole.
Now, many of the transitions are really murky where we have for multicellular organisms, for example, we have lots of organisms that happily live as a single cell and then come together in multicellular assemblages and even have their own physiology and set of responses and behaviors in that multicellular assemblage. And then when life shifts, they go back into the unicellular level.
That even happens in bacteria. Bacteria are going in and out of these biofilms and making complicated choices about whether to form a biofilm and stay in a biofilm. And so multicellularity is a great example of a really murky idea.
I think many people would point to a major evolutionary transition in individuality as multicellular where it's not so easy to go back and forth between the unicellular and the multicellular phase. And where the multicellular really is being selected, the genomes of the multicellulars are really being selected as a whole.
So in our case, all of the features that we do, everything we do in our life, if we're going to have children... results in a single copy, half a copy for each human being combined with another half a copy to form a new organism. So eventually everything goes through one copy of the genome to go through development and form an entire other multicellular organism. So it doesn't matter that
Showing 121–140 of 209 · page 7 of 11
← Previous
Next →