# CSM_RH Paper 18 v0.1

Campaign 17 corrects the mechanism interpretation of Paper 17.

Main results:

- the complete triangular shift aggregate is exactly a positive Fejer-weighted spectral energy;
- there is no independent "preserve h-signs and gain a power" bypass;
- MLEPG necessarily requires fixed-power local L2 improvement on the principal q=1 arc;
- MRT already average over shifts and use Plancherel, so their early averaging step is not where shift signs are lost;
- strongly minor-arc Type-d3/d4 improvements cannot by themselves close MLEPG;
- current 2026 short-interval technology moves the almost-all PNT threshold to alpha>2/15;
- the quantitative precision is still arbitrary log-power, not a fixed X-power;
- exceptional-set power alone would not suffice unless the good-interval error also shrinks at fixed-power L2 scale;
- MLEPG remains open;
- Campaign 18 attacks the principal Fejer arc directly.

No proof or disproof of RH is claimed.
No live GLM provider run is claimed.