I think this is what he meant: Sometimes you can extract a valid proof from this kind of bad proof by reading it in reverse: 3. 0=0 2. 1*0=0*0 1. 1

I think this is what he meant: Sometimes you can extract a valid proof from this kind of bad proof by reading it in reverse: 3. 0=0 2. 1*0=0*0 1. 1=0 However, in this case the step from 2 to 1 requires "division by 0".