Fields medallist Terence Tao wrote this week that good open problems are a finite resource. He compared automated solving to digging up an archaeological site with an excavator. He suggested some problems may need to be declared off-limits to machines.