mt

I've been trying to prove @ericweinstein GU in Lean using AI, but I'm essentially a blind mand in a gladiator coliseum, as I'm no mathematician. As far as I've achieved, GU base structure is consistent a provable, however there's some missing gaps that the math library does not have, that the agent claims are needed to go deeper and complete GU proving: - Spin cover image = SO₀(7,7): needs the Cartan–Dieudonné theorem plus facts about connected components of matrix groups. - Submanifold step: needs general "the solution set of nice equations is smooth" machinery for these spaces. - Global spin lift (w₂): needs characteristic classes, which are basically absent from Mathlib. - Identifying the Pati–Salam subgroups with Spin(6) and Spin(4): needs abstract spin groups as their own objects, plus isomorphism theory. Again I emphasize I'm no mathematician, so I'm relaying on this overconfident abstraction, I just wanted to understand where would this experiment lead to. But I wonder it it going on a good path, or should stop wasting tokens?

left: 1999. right: ours. a village grows from a few huts. same village, frame for frame.

left: 1999. right: ours. land, fight, take the island. same raid, frame for frame.

1999? no. this is ours. same village, same vikings, frame for frame.

saga: rage of the vikings came out in 1999. we're rebuilding it from scratch last time a longship crossed the sea. now the first missions play again, start to finish: raise a village, sail to an island, breed cattle, win every viking does exactly what he did in the original

all 36 shipped maps of saga rage of the vikings (1999) now replay 1:1 with the original game in our clean room C port, every entity checked tick by tick against the original binary for 1600 ticks the clip is the port's own renderer: two vikings board a knorr, it rows out, sails across, docks and they land. same frames the original draws

it's now playable, real end to end port, function by function, renderer blit by blit, but still a lot to cover. the oracles are set in place, let's praise that the bitter lesson pays out #re #saga #rts #strategy

resource gathering and build queue ready #re #saga #rts #strategy

before the conquest

sir, we have been conquered #re #saga #rts #strategy

full viking animation cycle, comming along nicely

fish in the pond

deers enjoying the pasture

ai places buildings now

vikings, now properly animated can now walk behind buildings with pathfinding, and they have a shadow too.. ah and buildings now render too

the fog of war is in, coupled with real vikings opening up the map as they walk, the port is progressing along, and getting closer to what once one saw

ground is real now. every tile is four quarters, each a frame from one of six terrain banks, picked by two bytes per quarter in the 1999 index table. read from the disassembly today, rendered tonight. the vikings are still dots, next

clean-room port of saga: rage of the vikings (1999). first frames out of our own simulation, drawn with the original's projection code, ported and judged by running the 1999 exe in an emulator. ground is still flat colour and the vikings are dots, that's next. 4 of 36 maps bit-exact for 200 ticks so far

The terminal doesn't care. But you do.

The bitter lesson at play, given enough data the systems you'll build will provide the performance necessary that routes data to decision making effectively.