Many of them are very aware of how LLMs work, they regularly interact with context limits and there have been threads about thoughtfully pruning context vs letting the LLM compact, making backups, etc.

Their hobby is... weird, but they're not stupid.