This is a special post for quick takes by ABishop. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
I made a simple game. It is built on simple rules that become clear through gameplay, and the opponent was trained using an NNUE-based neural network. Since it is a perfect-information game, a winning strategy is presumed to exist, though I have not sought it out. I also wonder if this type of game has ever been developed before.
Imagine we could grow eggs and sperm in a lab from ordinary skin cells, so couples could have far more embryos to choose from. How much smarter could picking the best embryo make a child? A 2019 study estimated that choosing among 10 embryos would add roughly 3 IQ points, and that going from 10 embryos all the way to 1,000 would raise that gain only about 1.7 times (Karavani et al.). So the big lever isn't how many embryos we can make. It's how well DNA can predict a trait in the first place.
How well can DNA predict which of two siblings will do better? One large study found that a DNA score explaining 12 to 16 percent of differences in how far people get in school predicts roughly half as strongly once you compare siblings (Okbay et al.). Which number should a forecast use?
Thank you. I’ve always been curious about the actual methods of intervention—beyond mere speculation—but my lack of biological knowledge is really holding me back.