
World Permission Gate
The live edge is not a giant pile of themes. It is the discipline that rejects unscored claims and rewrites them into observable permission gates: who can secure port insurance, fertilizer policy proof, IMF...
Dated calls, scored when they resolve. 67 forward calls are sealed at publish with a kill criterion; the nearest come due in 2026. Open any of them for the sources, refute notes, and full argument.
13 of our Metaculus forecasts have resolved; 7 scored as binaries so far, Brier 0.06. Hits and misses sit in the same list. That is the point of a sealed record.
35 calls across 6 research notes, soonest first. Each row gives the claim, resolution date, and the public read. Open the PDF for the exact scored clause.

The live edge is not a giant pile of themes. It is the discipline that rejects unscored claims and rewrites them into observable permission gates: who can secure port insurance, fertilizer policy proof, IMF...

The loudest shared shift across this board is that the binding constraint has moved off the demand narrative and onto a specific inelastic supply input or a fixed-date administrative step, while the market has...

Our house view across this board is simple. In three separate commodity chains, the supply response the bulls are betting against has already been set off by a specific, dated event, and the market has not...

AI is repricing the data layer, but the remaining edge is no longer 'NAND is scarce' or 'HBM is sold out.' Those are now public, crowded, and easy to overtrade. The harder call is whether AI buyers can secure...

Power scarcity for AI is now consensus. The underwritten question is narrower: whether a named AI campus, HVDC line, co-located generation deal, or large-load interconnection can energize on the schedule...

The shared shift is that AI stops being mainly a model market and becomes a physical deployment market. The obvious layers, GPUs, frontier labs, generic agents, and benchmark wins, are no longer where the edge...
46 calls resolving 2030 to 2040, at limits physics and demography have already set. These grade the thesis, not this year's record; read them as research with a date attached.

The post-AI world is not defined by intelligence becoming scarce. It is defined by intelligence becoming cheap enough that the bottleneck moves to permission, coupling, and verification. The winning layer is...

All six calls share one shape. Capital, tools, and wafer capacity can all be bought on a capex clock, but the real ceiling sits a layer or two below the priced story, at an input whose supply grows on a...

The board converges on a structural shift away from geology as the binding constraint on critical minerals supply, toward inelastic human capital shortages (P1, P3), regulatory waste disposal caps (P2), and...

We see the same shift play out across all four calls, in four different physical systems. Capital and headlines crowd into the fast, fundable, capital-intensive layer (cable plants, copper mines, fusion...

Every call here runs on one mechanism. A downstream system scales fast while one upstream input is fixed by physics, chemistry, demography, or treaty, so the rent migrates onto the input that cannot answer a...

We make the same structural argument six times. In each case a loud, well-funded boom is paying for the visible layer (curie count, fractionation plants, targeting ligands, reconditioning devices) while the...

Across all six the rent sits one layer below where capital is looking, on an input that cannot be expanded with money on the relevant timescale because the limit is an element, a reservoir, a human career, or...

We see the same trade six times, and the market misreads it the same way each time. The funded, visible bottleneck in space is always transport: launch cost, cadence, landers, GPUs. Cheap reusable launch is...
Not investment advice. These are dated forward calls; they get graded when the resolution date arrives.