Imagine you have a machine that rolls a d20 and lies if the die comes up 1-19, and tells the truth on a 20. Should you trust this machine usually? No. But if you can _see that the die comes up 20_ then you should trust it. The fact that it sometimes might lie doesn't mean that you should distrust the machine if you can see that in this case it's telling the truth.

> Coincidentally in the example the results are the same , but they need not be.

The questions is whether we should draw different conclusions when the results are the same. I don’t think that anyone has any issues with drawing different conclusions when the results are different!