Exception Handling, it seems, has utterly failed to prepare a failover version of their secret medical market with weaker medical trading algorithms that only use human-legible statistical methods.
But Exception Handling, contrary to what some angry voices are already saying, is not made up of complete incompetents! Just moderate incompetents! It's not the first time that any secret prediction market has failed for any reason, and Exception Handling does have a pretested secret version of the failover software and protocol that regular hospitals use, when some (rather less liquid) regular open prediction market seems to be failing for whatever reason, but the hospital still has Network access and the case still has interested human participants.
It involves human experts producing the same judgments and other human experts integrating those to arrive at final recommendations in place of machine algorithms. It's not particularly complicated. When a complicated machine system just failed, it is a great time to failover to a simple human system. They aren't missing the software for that. Failure on that level really would get people fired, and not just in Exception Handling's branch of Governance either.
It's going to take five seconds for the failover manual protocol to start booting up on the monitors, and twenty seconds after that to actually get the human expert judgments routed to a human integrator, and the first non-market recommendations to start appearing on the monitors. Though, obviously, they're going to have a few more seconds latency compared to when algorithms were doing half of this work.
Oh, and standard protocol in these failover cases - developed for markets that are usually a lot less liquid when they fail - does of course state that the current chief operator in the patient theatre is the one person responsible for making the final call on which recommendations if any should be followed. Just like in the sims!