I should have loved biology

Human civilization is severely time-limited (or just time-pressed). We can't wait millions of years running a simulation to optimize a little widget. We need to rely very much on high level design and comparatively little on efficiency and optimization. On the other hand, life cannot afford huge DNA (very costly), or energy waste (generally disfavorable from evolution). So human built systems tend to be of a low "Compute complexity": the computational complexity of obtaining solutions and solving problems themselves (like civil engineering structure problems, or design of objects) must be fairly low. For life, systems can be amazingly intricate, every tiniest cell a wonder that would probably take thousands of years for civilization to maybe be able to replicate. But it all ranges from about 130kbp to 8Mbp[1], which would be around 16Mbit/2Mbytes at most. So it fits (uncompressed) in a diskette (floppy).

Even now with powerful computers, we're still mostly constrained by cognition (specially human), you see simplicity all around you.

So if you look at the human world, you see (computational) simplicity everywhere, but the natural world has undergone trillions of generations of optimization to arrive at almost perfect (in an almost literal way) little machines, complicated but with a hidden amazing (size) simplicity.

I think there's a connection to be made to algorithmic inference as well. Originally we came up with ideas for Universal Inference (from ideas from by Solomonoff, Kolmogorov among others) [2][3], the most glaring candidate was the "size prior": evidence explainable by the least algorithmic information ought to be most likely (Solomonoff inference). Later, there were promissing ideas around an additional term: the "speed prior" (from Schmidhuber[4]) -- the biological word is one where the "size prior" (simplicity is most likely) works almost perfectly, and human civilization is one where the "speed prior" (computationally easy is most likely) is helpful.

[1] https://en.wikipedia.org/wiki/Bacterial_genome

[2] https://en.wikipedia.org/wiki/Kolmogorov_complexity

[3] https://en.wikipedia.org/wiki/Algorithmic_probability

[4] https://en.wikipedia.org/wiki/Speed_prior

Side note: I think intellectually one of the ways we're really far behind is recognizing Algorithmic Information theory as a foundation for statistics and metaphysics. We're very stuck making little progress on the metaphysical realm (which physics is advancing more into) because of a lack of widespread acceptance of those advanced tools for science. Algorithmic inference gives a solid basis for comparing metaphysical models and deep questions about the cosmos.

It did seem like just a bunch of memorization to me then. Much, much later, I took an Extension course in Molecular Biology, and what a difference! I still think the diagrams they draw for biological processes in Molecular Biology of the Cell are just stunningly beautiful, and completely blow away anything you'll ever see in a CS text.

But expecting to have those revelations when you don't know how anything works yet is foolish. You have to pay your dues.

Bonus points for pictures!

Or is it something a little more profound?