[torstenvl]

There is absolutely, 100%, without qualification, no circumstance whatsoever where a regression is good for most users.

The fact that you think so makes it hard to take anything else you've said seriously.

[joshuamorton]

Do you mean that this change is a regression for most users? In which case do you have data to back that up?

Or do you mean that a change which regresses for a small number of users while improving for most is bad? In which case this is not correct and I'm happy to elaborate on why.