I'm generally supportive of HFT, I think the massive reduction in total profits to market makers relative to 20 years ago is proof of a positive-sum efficiency

I'm generally supportive of HFT, I think the massive reduction in total profits to market makers relative to 20 years ago is proof of a positive-sum efficiency gain. But I'm not sure I buy your argument for why the nanosecond arms race is good. Computers getting faster has allowed us to do more and different stuff with computers. What are the actual benefits of HFT "efficiency"? If stocks had an auction every millisecond wouldn't that still be plenty fast enough for any practical purpose? Only other HFTs will notice or care that AAPL's price got updated in 1 nanosecond instead of 2. Or to go the other way, how would we be better off if stocks ticked every femtosecond? Every attosecond? Is there some limit where the benefits of this "efficiency" aren't worth the resources to pursue it?