> Imagine a 4 billion year old codebase, written by dice throws and unit testing.
That's a great analogy, except it doesn't go quite far enough. The code for the toolchain that compiles the codebase (not to mention the unit testing framework) is just as old, and written the same way.
Note to self: while the code (including the toolchain code etc.) is in a monorepo, it is actually surprisingly modular. OTOH, much of the functionality was implemented by copy-and-paste of existing code elsewhere in the repo, and the copies then diverged with subsequent modification.
His point was more along the lines of it being poorly designed.
Evolution isn't going to nicely separate functionality into clean, single-purpose biochemical pathways.
It's going to overload something that it already has to support a new feature, leading to a rat's nest of intertwined effects.
The hard part in drug development isn't affecting your target: that's pretty easy to guarantee with modern tools. It's finding out that that target also governs 10 other, completely unrelated bodily functions.
It sounds harmless because it omits the incredibly complex process of developing immunity.
Your body basically is trying to do a sort of brute force/random combination of immune cells on the retrieved virus spike proteins until it comes up with the right combination.
This process can result in B cells that produce antibodies which stick to virus proteins but also stick to your own cells. The B cells are essentially permanent, that’s how you develop lasting immunity, that unfortunately also means you have lasting immunity on one of your own cell types — autoimmunity https://en.wikipedia.org/wiki/Autoimmune_disease
A 50000 people study is likely to see adverse events completely unrelated to the vaccine. Halting the trial is not a sign that they think it is related either, since it is protocol to rule things out before continuing.
