I suspect they're referring to VPRI's STEPS research, not Smalltalk. (And it was a target of 20k, not 10k.)