A modern take on Matthias Wandel's classic [0], which has you guess a variety of geometric attributes (e.g. angle bisection, centroid locating, shape regularization), not just simple partitioning of a line.
The fact that the numbers are in a brighter color than the end marks, and that the numbers go inwards, makes it slightly more difficult than it would otherwise be, because the eye is biased by the more prominent space between the numbers being different from the line between the marks.
It would be great to have a 'training' mode, where you get to repeat ones you miss. This would increase the learning speed.
Easy training- repeat the one you just borked
Medium training- cycles through say 5 examples until you get all five within your target range (1%, 0.1%, whatever)
Great idea! Have you considered storing triplets <range, correct number, selected number> for each try and making image plots of these (x/y coordinates are correct/selected numbers, color of each pixel represents frequency) for multiple users for each range? I think the image might reveal interesting properties of human eyeballing, like near-perfect accuracy around 50%, but with less obvious correlations.
[0] https://woodgears.ca/eyeball/index.html
It would be great to have a 'training' mode, where you get to repeat ones you miss. This would increase the learning speed.
Easy training- repeat the one you just borked Medium training- cycles through say 5 examples until you get all five within your target range (1%, 0.1%, whatever)
This is fun!
Also, I tried this on laptop as well as my phone, I liked it more on my phone (I know the whole point is about precision though)
*my old pal Claude
A time limit would make sense imho. For extra challenge, add diagonal or curved lines.
0 out of 1,600
I still missed. Even when there was centered text.
Maybe the human is the weakest link
...
handleClick({clientX: els.bar.getBoundingClientRect().left + els.bar.getBoundingClientRect().width / state.n * state.target })
(It was pure luck)