So this is an experiment -- I have been using AI to write papers I had ideas for years in the past month+. For this just as an additional test, I had GrokBot upload the paper to crimrxiv as well. (Which takes signing in and doing some edits to a few of the fields and checking the HTML -- it did a great job.) So these are ideas I came up with. One of the critiques of surveillance cameras is they disproportionately target minority residents. Here I created an linear program to site cameras that incorporate racial fairness constraints -- so it sets the exposure minority residents have on cameras. It barely reduces the efficacy of cameras (in terms of total outdoor crimes captured within 1000 feet). The future of writing social science papers with AI is here. It can be done in an effective and safe way. No hallucinated citations, because I gave it an outline of what to cite for example (and reviewed the citations I was unfamiliar with afterwards). So this was an idea I had for years, and decided one morning to have AI help me write it up (I am pretty sure it took in total less than 3 hours for Opus 5.5 to churn it out). I have 9 total papers I have had AI churn out just in the past few weeks. Welcome critiques, but besides the AI writing being hamfisted (which I think will likely continue to improve with future models) this is higher quality than the majority of social science research (it has code to replicate and is a topic that actually makes sense to examine).