We do it because it's practical and because it's scalable. We do it because we're a society that loves standards but doesn't really understand how they work. So

We do it because it's practical and because it's scalable. We do it because we're a society that loves standards but doesn't really understand how they work. So we're comfortable painting everything as a bell curve. The outcome is something resembling a lowest-common-denominator education for all. Outstanding students don't pull a class upward, but unproductive or struggling students do slow a class down. That sucks for the student who's ahead of the class, because her potential is being wasted. It also sucks for the student at the bottom of the class, because he shouldn't be there. (That sounds harsh, but it's true). Solving this problem will require implementing a scalable way to aggregate students by some level or set of criteria other than age. We need to organize them somehow, because it's currently impractical to teach them all individually. And it's tricky to organize them by any "potential"-based criterion at an arbitrary age, because the brain is changing almost continuously throughout childhood. Plucking a group of "gifted" students apart from the pack at, say, age 2 is poorly predictive of who actually turns out to have high potential later on. So we need to do better than that. These are nontrivial challenges, but the solution would represent a giant leap forward in our society.