On Evolution

[Primordial Pouch, Wild Cat Simulator, Simulations Have Preferences, Earth: The Oldest Computer]

2025.04.06

XCIV

Thesis: The most impressive program ever written is called Evolution, and it’s running on a pretty big computer right now, called Earth. 

Primordial Pouch

Have you noticed the cute & adorable flap of stomach fat cats tend to have near their hind legs? It’s called a “primordial pouch,” and there’s actually some very good reasons for the fat to be there.

If you play with a cat, you’ll notice that sometimes they “bunny kick.” When they bite you, they start bringing their hind legs up towards their mouth and kick down. This can help disembowel prey and also rake the body of an attacker (like another cat, or your hand).

So, one reason that has been posited for the primordial pouch is that it protects the cat’s organs from another cat’s bunny kicks. 

Intuitively, that makes a lot of sense. Of course a cat would “want” extra fat where their play partners and enemies naturally strike… but it’s crazy to think that a cat would actually end up with a protective pouch there. 

I “want” wings, but that desire has no weight on the evolution of our species. For all of evolution, organisms did not get to request convenient features for the next iteration of their species, no matter how much they might’ve desired it.

Still, we’ve ended up with cats with very intelligently placed protective primordial pouches… but how? 

The most convincing answer is evolution.*

*While I tend towards more atheistic views, this post is not intended as combative towards religion–my understanding is that evolution as presented here is roughly compatible with many modern religious stances, even if a sprinkle of divine agency is thought to influence the process.

Wild Cat Simulator

To understand evolution, you might view it as a game. Let’s say we have a Wild Cat Simulator, WCS for short.

At the start of turn one of the WCS, we have thirty randomly generated cats, all with slightly different “features,” including pouch size & claw sharpness on a scale of one to ten, as well as hair color.

Throughout turn one, the simulated cats navigate some simulated wooded area and simulate eating. Each cat gets a score at the end of turn one based on how much it was able to eat. This is a simple representation of it’s fitness and how likely it is to reproduce. And, of course, if the simulated cat passed away in turn one, it gets a score of 0. 

Here are six cats & their scores from turn one:

Now, before the WCS starts turn two, it makes a new “litter” of cats based on the most successful cats from turn one. 

The characteristics of the highest scoring cats have a higher chance of appearing in the new litter, while the characteristics of the lower cats have a much lower chance of appearing in the litter.

Then, turn two will proceed much like turn one, with the Simulated Cats simulated eating. Once again, at the end of turn two, the WCS will review the scores and weigh the characteristics of the highest scoring cats more highly than the characteristics of the low scoring cats to produce litter three. 

Simulations Have Preferences

As the WCS goes on, the cats in each litter will converge on a “solution” for scoring high in a simulated environment. 

Even if we didn’t know anything about the inner workings of the WCS itself, we could infer a lot by:

• Watching the litters evolve through many turns in one game

• Comparing the end results of different games

Maybe we find that all games tend to converge on cats with a pouch size of 6 after 100 turns. This tells us that likely both a larger pouch and a smaller pouch are punished.

Afterall, there is some sort of scoring system inside of the simulation, meaning some characteristics are more likely to fare better than others. In a way, the results help us understand that scoring system.

It also helps us understand how real cats could develop pouches. Even if we started with cats with very small pouch sizes, if even a slightly bigger pouch size helped a few cats win a few fights, on average across A LOT of cats, and a lot of time, these cats would get to reproduce more (in the most simple sense, they live longer and can mate more). 

In other words, an animal’s ability to reproduce is the real world’s scoring system, and the characteristics of the animals that score more highly are more likely to make it to the next round.

Earth: The Oldest Computer

Following this process, we end up with cats getting a very convenient feature—primordial pouches. Evolution is so unbelievably incredible at giving species desirable features like that that it is very easy to see why you might want to apply agency to it or give credit to a god.

To me, it also almost feels like the most incredible program ever that's been running for 4B years on the computer we call earth. It can generate mind-blowing organisms (like humans) without itself “knowing” anything or even having any real agency.

Religion & scientific mysticism aside, understanding evolution has been a very powerful frame for me as I think about getting results elsewhere.

Whether it be testing products, sales strategies, or even heuristics for life in general, if you have something that you can…

1. Score variations of

2. Create new variations based on the last round’s scores

…then you can do a lot of crazy things.

Of course, the WCS is a vast simplification of a very real evolutionary process. Highlighting some of the model’s deficiencies:

• Uncountable Features: The WCS only has three qualities for cats (claws, pouches, fur color). In reality, there are uncountably many factors that would go into calculating a cat’s success.

• Unseen Features: The cats need to be able to develop previously unseen features to get to real evolution*.

• Scale of Experiment: We need way more than 30 individual specimens and 100 “turns” to get anything noticeable. There have been ~117B humans ever over the last 190,000 years, and even that scale hasn’t gotten us much physical change but smaller bodies & brains.**

• Dynamic Environment: The WCS is static. In reality, the environment is changing and so are the other species in the environment.

Still, none of these draw backs are damning enough to move with the spirt of rapid iteration and hypothesis testing.

Live Deeply,

*Extreme example: If you’re starting with a blind amoeba and you want to eventually have a human with eyes, you need to be able to randomly get an amoeba that has at least some light sensitivity even if no amoeba with light sensitivity existed before it.

**Looking at culture in terms of memetics implies that we’ve leveraged (intentionally or not) the mechanics of biological evolution to accelerate cultural evolution. You could apply the same algo to our technological development, too.