4 ms·
I wonder if the model here is missing something important - it assumes there is no survival penalty for a false positive attack. In real life this isn’t true. A
by bb123 6y ago
I wonder if the model here is missing something important - it assumes there is no survival penalty for a false positive attack. In real life this isn’t true. Activating a fight or flight response in humans and animals has an energy (survival) penalty.
- zomglings 6y agoYou are right. It would work well if the models included a utility function and, instead of measuring likelihood of survival, the measurement were expected utility.
- goldfish 6y agoIt does include a survival penalty for a false attack. I laid out the assumed payoff matrix here: https://www.adamjuliangoldstein.com/blog/paranoia-parameter/#responding-under-uncertainty https://www.adamjuliangoldstein.com/blog/paranoia-parameter/.... The penalty for false positives is also baked into the simulations (see first few lines of code here: https://github.com/adamjuliangoldstein/anxiety_algorithms/blob/master/reaper.py https://github.com/adamjuliangoldstein/anxiety_algorithms/bl...).