@NateSilver538 That has been the best use case for LLMs for me -- writing short functions from a specific description. Fails occasionally, but usually in a way that's still a time saver when including time for troubleshooting. Longer code snippets are still very hit or miss, though.
@mred53 Everything in this thread matches what my simulations say, which gives me some added confidence that I implemented this year's tiebreakers correctly.