Here's the distribution of time people spend reviewing cards at quantum.country. The median time is about 6 seconds. The _average_ time is almost twice as long, at 11 seconds!
Conversation
I find this fascinating. It suggests people mostly fly through at a steady clip, about 5-8 secs per card, but they're occasionally getting stuck, taking 10-20 secs for an outlier card.
1
6
One model I like for thinking about spaced repetition review is that it should be almost meditative: you get into a zone. Outlier cards break that zone.
1
8
I'm really not sure of that model, though. Another model is that it's _good_ to occasionally get stuck, to have things broken up. People often dislike this feeling, & feel that they're learning less, but there's quite a bit of research suggesting they're just wrong.
2
5
So I'm kind of split about this. Should we try to eliminate outlier cards? Or are they really a feature? Not sure!
6
4
Caveat: it's tricky to interpret such data.
Maybe people are getting distracted, & that's why some review times are long - "oops, need to pay for my coffee!" Much beyond 20s, & I'll bet that's what is going on.
But there's a lot 10-20s too, so I don't entirely buy that.
5
10
You’re unable to view this Tweet because this account owner limits who can view their Tweets. Learn more
Interesting! What are the units? Seconds? Minutes? And how is thinking_time defined?
1
You’re unable to view this Tweet because this account owner limits who can view their Tweets. Learn more
If it is minutes, that would be pretty consistent! It’s hard for me to imagine that the overwhelming majority of your reviews would be completed in less than one second. I wonder if it also tracks seconds, if my guess is right.
2
I should also mention that we dropped all durations greater than 50 seconds in that analysis, to try to remove obvious idling outliers. Those were <2% of samples, iirc

