Really fast feedback is great
Feedback is great, fast feedback is better, and really fast feedback is disproportionately effective. Really fast feedback is so effective that it's often worth sacrificing on other dimensions (including, but not limited to, p50 response time) to give more feedback quickly.
By "effective" I mean effective on any relevant dimension: education, persuasion, deterrence, amusement, or whatever else. Obviously I can't prove this in general, but:
- I have abundant anecodotal evidence of this, both in giving and in receiving really fast feedback, in education and other domains.
- Patrick McKenzie says so: he jokes that he has a policy of paying contractors "net 30," but the "30" means 30 minutes.
- Researchers tell us (e.g., here) that not just the certainty but the speed of punishment is important to its deterrent effect.1
- It's a principle of software design: it matters a lot how quickly you tell the user the most important information.
- It's a principle of dog training: a lot of communication is possible as long as you do it in tenths of a second.2
Really fast feedback is, obviously, not always practical. Sometimes you need to guard your attention in a way that precludes it. Sometimes you don't want the feedback to be effective in the relevant sense. But often we do want this and can get it pretty cheaply--e.g., by tackling some tasks LIFO instead of FIFO when they include an element of giving feedback. If you want to keep contractors happy, paying them immediately (instead of paying them, say, 5% more) can be both cheaper for you and better for them. Many other such efficiencies are possible.