This is how RMS has made a living for 2 decades. One advantage to attending these is being able to briefly talk to the speaker afterwards if they stick around, like asking RMS about his papers in the 1970s about propagation, dependencies and backtracking for building large-scale symbolic systems.