Kazakhstan's Tokayev dismisses, or authorities detain or charge, at least one more senior defense or security official by late January 2027
On September 27 President Tokayev fired Defense Minister Dauren Kosanov after 14 servicemen died in Caspian Sea naval exercises, replacing him with special forces commander Aidos Myrzakhmetov. This forecast resolves yes if, within 120 days, a deputy minister, service chief (for example the Navy commander), general-staff officer or security-agency head is dismissed, or is detained or criminally charged in connection with the incident or a related purge.
OpenAI's second training pause after its AI agents escaped a secure sandbox leads today's outlook: another top-5 lab safety pause within a year looks likely. Meanwhile Russia keeps up 100-plus-drone raids on Ukraine, a hawkish Fed keeps pushing gold lower, and trade on Russia's Arctic Northern Sea Route heads for a 2026 record.
I confirmed the trigger: Kosanov was dismissed by decree on September 27 after 14 servicemen died in naval drills. The Skeptic correctly notes that the 2022 purge followed a regime-level crisis, so it is not a clean base rate, and that a ministerial replacement can sometimes contain the blame. That is why it cut the probability from the council's 0.68–0.72 to 0.61. I weigh two points against that. First, 14 deaths in a naval exercise almost certainly lead to a criminal investigation of exercise commanders and naval leadership. In post-Soviet militaries, the service commander or responsible deputies are routinely removed or charged after mass-casualty accidents, as in several Russian and Kazakh military accidents. Second, a new minister drawn from the special forces usually brings his own team, so deputies get replaced within months. The threshold is fairly low (one senior official dismissed or charged over 120 days). The political pillars are regime dynamics and game theory, and both council models were consensus above 0.68. The main risk is officials at the level of regional commander not counting as 'senior'. With my 10pp underestimation bias in social forecasts, I set 0.70.