The article presupposes the 10x engineer is a common thing. It's about as true as Linus Pauling is true. It did happen. It happens a French metric tonne less than people think, and typically is somebody applying what I would call "computing tropes" to existing inefficiently deployed solutions in barely above hypothesis state, to scaleable solutions. Carmack was given a problem well suited to his temperament and circumstances. Given another problem in different qualities and dimensions, how well he would have performed is unknown.
We also underplay how often the "hacks" are well known, such as loop unrolling, or exploiting chip specific behaviours, or simply picking the O(rightsize) algorithm to replace the ones in use.
Biochemists very rarely have any of these luxuries. You can't often hack how cells work to make 10x happen. I'd say finding the rotting cantaloupe with a 1000x more productive penicillin mould, that might be a time. Or Du Pont scaling the British techniques for penicillin production using China pans to industrial scale: even they were leveraging Chaim Weizmanns approach to ABE in WW1. Russians doing phage therapy, now that's a hack. What a hack, mind you.
We also underplay how often the "hacks" are well known, such as loop unrolling, or exploiting chip specific behaviours, or simply picking the O(rightsize) algorithm to replace the ones in use.
Biochemists very rarely have any of these luxuries. You can't often hack how cells work to make 10x happen. I'd say finding the rotting cantaloupe with a 1000x more productive penicillin mould, that might be a time. Or Du Pont scaling the British techniques for penicillin production using China pans to industrial scale: even they were leveraging Chaim Weizmanns approach to ABE in WW1. Russians doing phage therapy, now that's a hack. What a hack, mind you.
reply