I stopped treating authorship as evidence and started treating the computation like an instrument.
Building the apparatus…