| Your account is new and has no profile and for all we know could be a bot. However, in adhering to the HN guidelines [0] I must "assume good faith". And indeed, you provided an excellent, intelligent reference! Very interesting paper and debate. In the paper the authors review the debate on early stopping, and then created a model/simulation to examine what one might expect. I should note that the paper came out in 2016, which was before the December 2020 early stopping of the COVID vaccine trials, a real world example of an early stopping debacle, where trials were stopped and then efficacy turned out to be vastly less than originally reported (even worse than the "29%
exaggeration of effect" Bassler et al originally reported). I think your argument has convinced me that my position is not correct. However, it has not convinced me that early stopping is correct either. I think the obviously dumb thing is having these rigid trials, and a far better idea is to have real-time adjustable ongoing data collection and experimentation that never stops. Thanks for the article. Good read! [0] https://news.ycombinator.com/newsguidelines.html |