I don't think people are taking threat of manipulation, in a broad sense, seriously enough.

programming challenge seems much simpler than the other three challenges, seems like you'd want to have a challenge of building a fairly complex piece of software in order to match the complexity of building a house and passing the turing test. perhaps some fully functional ui with complex behavior like a networked game of some kind or something. not sure how to specify that ahead of time off the top of my head though. also, logarithmic time?

I will not change my prediction on this question under any circumstances.

— edited by lahwran

@Jotto it's already passed the trials I expected it to fail hard, on early safety; it's fairly well known to be effective, and this trial is to establish sufficiently high quality evidence that the FDA becomes happy to approve it. I would be surprised if they find that it's not effective, but not so surprised that I'd put a ton of points on it. doubling the baserate seems about right to me given that context, but I'll be watching for more news.

— edited by lahwran

@jacob.pfau I like this approach. Breaking it into more precise questions about individual components allows much more precise claims. I think it would be worthwhile to try to find objective metrics for relative system complexity as well - relative computation performed in different algorithm phases, or similar.

more ideas for searching under the general category of highly infectious disease: how many cases of measles will there be in the united states in 2020 and 2021, respectively?