Such theories can easily be tested with a simple thought experiment: suppose there were a similar bug at the other end of the pipeline, in the subscribe feature. How long would it take for the bug to be fixed?