This is essentially the same problem schedulers run into. It's a latency vs throughput trade off. The article is underlining that blocking is bad, & lots of tas

This is essentially the same problem schedulers run into. It's a latency vs throughput trade off. The article is underlining that blocking is bad, & lots of tasks may receive cancellation signals I've worked in low latency environments, but I prefer high throughput. The latter is more conducive to flow state. It also means people aren't relying on my consistency, where some days my brain turns on & some days it doesn't