Fiction

Other Projects

You can comment on everything. Or give me (harsh) personal feedback.

Made with 💙 by me (how to).

Dedicated to my family.

I have previously written about what I think about current approaches to aging science - I am now finally collapsing some of my models into an actionable shape. This specifically means that I am building my first company / research lab, more on that soon. I am convinced that all problems are solvable if you turn them from a modeling problem to an engineering problem: if you have sufficiently reduced the space of all possible solutions to a “shape” that follows a clear “ask the universe the right questions and improve based on answer” loop. If you want to be on the moon, you first have to observe where it is and how it moves. From that alone you can speculate that it’s high up, fast, and surrounded by nothing. The rest becomes a question of how to jump high and reach sufficient speed in a vacuum - the constraints.

Aging is a change in how the body responds to entropy, it is inelasticity of the repair machinery. This is similar to a spring getting brittle over time.

But only somewhat: living systems can precisely fine-tune how much and when to export entropy, they are self-annealing, self-tuning springs.

This is insane given that they need to repair and rebuild more. Why are they not retuning themselves to be less brittle?

Independent of what causes the increase in inelasticity, we need to restore it. This will look similar to how you retune a spring:

The solution must add energy to the entropy export machinery, even if it doesn’t “want” it. Re-forge the spring with fire.

Biological systems are hard to understand and made up of a very large set of things. Nevertheless, a lot of things are copies - both in space and time: your hair all looks the same and if you pluck some it regrows.

Of course biological systems are compressible; not all things can matter. This manifests in the “multi-scale competency hierarchy” of life: cells distrust their genomes, so they have transcriptomics to ignore details, tissues distrust cells, etc.

(phrasing from José)

When you’ve made it to 20, you couldn’t have had a genetic defect, otherwise you’d be dead. Most people make it to 20, so one can assume that your current genome is perfectly capable of doing homeostasis. Your genome also can’t vary widely, as you have a “difference detector”: the immune system.

You can’t fix what already works.

Cells live both shorter (blood cells) and longer (neurons) than their organism. Reprogramming them into a youthful state is not sufficient. Single-celled life is immortal, including your germline.

The hard problem of You is that you’re made of 30 trillion.

Genes determine expression, determine protein levels, determine how the ECM looks, determines which cell survives, which determines genes. Ecosystems do not have single causal levers.

The most relevant level is the agentic You, but the abstractions it has access to are lossy. This is good to some point (micromanaging your immune system would probably be dumb), but it limits what can be done.

The solution will have to solve coordination problems for the massive, warm, and mobile environment that is you. Tissues are the interface between your mind and your cells, therefore manipulating them should have the greatest effect size per unit effort.

We do not know what we are, but we know what we are not: skin cells, hair, gut microbes, and even individual neurons.

People paralyzed from the neck down seem to be sufficiently themselves, therefore a full rejuvenation of the liver / thymus / gut, etc. could only be instrumentally useful, not terminally.

The only relevant endpoint is “are you still you”, which is only not the case when your brain has become compost. The solution must keep the dynamics (not the specifics) at the level of your 20 y/o self. Keeping the rest of your nervous system intact is a very nice to have, keeping your digestion or movement system intact is pointless.

Never confuse a metric with your target, never Goodheart, never use KPI’s other than “looks correct”.

Scarring, immunity, energy & signaling, ECM & vasculature all look different there.

What will work in the body is not guaranteed to work in the brain.

On that point:

So fuck them. Never build models that don’t capture what matters.

Exporting entropy is simpler for less massive beings, so evolution has found a way to reverse all entropy in humans: the germline. If you compress You into a single cell that has diluted away all damage, than immortality is easy. But it compressed the wrong you (your genes, which are not you).

It can’t look like fasting, comparative biology or tweaking metabolism. It has to look like surgery or Neuralink.

All these constraints lead to one meaningful reduction of the space of all possible solutions: