I'm supposing that nobody who has a job is producing abstractions that are always novel, but there may be people who find abstractions that are novel for their particular field because it is something most people in that field are not familiar with, or that come up with novel abstractions (infrequently) that improve on existing ones.