Earlier quoted context omitted.
I think there's value in that debate (within reason) even as a developer. If Alice thinks a task is worth 5, and Bob thinks it's worth 8, then there's a good chance one of them knows something the other does not. Is Bob aware of some hidden complexity that bumps it up 3 points? Or is Alice familiar with a convenient library that solves exactly that complexity? Planning poker is a convenient time to get that knowledge…
> there's a good chance one of them knows something the other does not Or that the assignment of points is arbitrary and imprecise and that different people have different ways of making up numbers.
As for its usefulness: You of course don't have to necessarily poker to get these discrepancies, but it's a pretty effective way of going about it, I'd say.