Reply to "Why not just go all in on points":

I really treat the testnet as a practice field now, and I don't take the "possible airdrops" as an advance salary... Once you start calculating expectations mentally, it's easy to get more and more involved, with machines over-allocating, scripts becoming more wild, and in the end, it's not that you didn't get the points, but that the nodes got shut down first, or even security holes you dug yourself.

My own stop-loss is quite simple: at most one hour of tinkering each day, and I shut down after that; for costs, I use old machines/small cloud services, and stop once bills hit the threshold; for risks, I avoid unknown clients/scripts, preferring to earn fewer points. I don't know if the mainnet will issue tokens, but I want to make sure I don't mess up the main environment just because of "guesswork," since I still need to feed the cat.