SUPERSTRATEGY – probabilistically optimised, frontier-forecasted decisions.

Question: Will AI be credibly identified as causing significant disruption or collapse of intermediation businesses (e.g. SaaS, knowledge brokers) before 1 Jan 31?

Resolves: TRUE if, before 1 Jan 2031, a business whose primary revenue derives from intermediating access to information, services, or software capability, and which had annual revenue of at least USD 100 million in the year before its decline began, has either: (a) sustained a decline of 30% or more in revenue or paying customers over twelve months or longer, or (b) entered insolvency, distressed sale, or delisting; and both of the following hold: First-party attribution. The company attributes the decline to substitution by AI systems in a regulatory filing or earnings communication. Corroborated mechanism. The substitution channel is independently evidenced — by measured migration of the relevant user activity, or by analysis from a party with no commercial interest in the company. Pre-decline baseline. The business was growing or stable for at least two years before the decline began, measured to a point before generally available AI products plausibly serving its function existed. A business already in structural decline does not qualify. Does not count: share price falls, multiple compression, analyst downgrades, or sector selloffs; restructuring, layoffs, or performance-management changes framed as AI adaptation; company statements characterising AI as a threat or an opportunity; management attribution unaccompanied by corroboration of the substitution channel. Otherwise FALSE.

Question: Will the global AI landscape be widely characterised as split into distinct US-led and China-led spheres of influence by 1 Jan 30?

Resolves: TRUE if credible analyses widely describe such a bipolar AI-bloc structure at the date; otherwise FALSE.

Question: Will there be a major correction in AI-related equity valuations (a peak-to-trough decline of at least 30% in a broad AI/tech equity measure) before 1 Jan 29?

Resolves: TRUE if both conditions are met before 1 Jan 2029: (a) Magnitude. The Philadelphia Semiconductor Index (SOX) or the NYSE FANG+ Index falls ≥30% from its highest daily close (at any point after this question's creation) to a subsequent daily close, and remains ≥25% below that peak for at least 10 consecutive trading days. (b) Attribution. At least three of the Financial Times, Wall Street Journal, Bloomberg, Reuters, or The Economist report, in news coverage rather than opinion, that the decline reflects a correction, unwinding, or bursting of AI-driven valuations. Otherwise FALSE.

Question: Will the 50%-reliability METR time-horizon metric for frontier AI models reach at least one week (168 hours) of equivalent human software-engineering task length before 1 Jan 28?

Resolves: TRUE if, before 1 Jan 2028, METR publishes a 50%-reliability task-completion time horizon of 168 hours or greater for a publicly released frontier AI model, where all of the following hold: (a) Units. The figure is 168 hours or greater when correctly converted from the units in which METR reports it. METR reports time horizons in minutes in its data files and comparison tables; 168 hours corresponds to 10,080 minutes. A figure must be confirmed against METR's displayed or stated value before it qualifies. (b) Point estimate, standard methodology. The figure is METR's central point estimate under its own standard scoring methodology — including its standard treatment of reward-hacking or cheating attempts as task failures. Confidence-interval bounds, alternative scoring conventions, alternative agent scaffolds or harnesses, and superseded methodology versions do not qualify. (c) Not disclaimed. METR does not accompany the figure with a statement that it is unreliable, not robust, or outside the range its task suite can measure, and the figure does not exceed any then-current reliability ceiling METR has published. If such a ceiling is in force, the figure qualifies only where METR has published a revised suite or methodology stated to give reliable measurements at or above 168 hours. (d) Within measured range. METR's task suite at the time of the measurement includes tasks with human-baselined completion durations of at least 40 hours, such that the figure rests on interpolation within baselined data rather than extrapolation beyond it. Otherwise FALSE.