Guessing on the low end is very difficult, I would make a terrible statistician (or politician?).

Agreed. I took one game to calibrate (I've just never worked with scatter plots much and had no idea what various P-levels "looked like") and got ~30 on my second game. What finally killed me was a streak of low correlations - Eyeballing gives me no idea if it's a 0.05 or a 0.35. Anything 0.45 - 0.8 seemed pretty easy. I didn't really get anything over 0.85.

"you appear to be offline. please reconnect and try again"

I'm pretty sure I'm online. Anyone else getting this error?

I'm on iOS and its suggesting I try Firefox or Chrome. Terrible user experience.

You're using a private browser window. It probably needs local storage. Disable private browsing and it works fine.

Fantastic... I teach quant research at the grad level. Game is simple and fun-ish. Learning to visualize and think about correlations for the first time can be tough—game is perfect to engage students and supplement textbooks, lecture.

The higher the correlation, the harder I find it to guess. (r = ... umm... :P )

My physics prof was fond of saying "Your intuition sucks."

I played by following my intuition and did quite well in the higher end but found the lower end much harder. I'm guessing because I was feeling my way through it, my brain was trying to see patterns even where there weren't any, pushing my guesses higher than the plot showed.

> found the lower end much harder

Same, although I think I overcompensated for the tendency to see patterns everywhere and generally guessed too low for R < 0.5.

Intuition doesn't suck, it just needs to be trained. That seems to be the purpose of this game.

No negative correlation? Also, it seems the human mind is estimating the robust correlation, but the game is estimating the classical one. This can explain the discrepancy at low and high values.

I love it, but I think it should get progressively harder? It's too easy to keep going far longer than seems reasonable to spend on the game.

Leaderboard should have a points per guess score (perhaps for a minimum # of guesses) as well as total points, which only values persistence.

Guess a username that isn't in use might be a more appropriate name for the game. And why do I care what you think I should be called anyway?

Bonus "this site is showing frequent pop-ups; do you wish to disable further alerts from this site?" Damn right I do.

Perhaps if the author had spent a little more on the UX and a little less on the pixelated UI then it would have been something I would have vested more time in.

Guess R; increase your P.

ten well spent minutes thank you

They need to add a leaderboard.

Why? What would be the benefit of demonstrating you're better or worse than other players? There's nothing to gain from winning.

That is true for most games, yet people are still interested in the score.

that gets boring quick

streak of 6 with a mean error of 0.03

This seems like a lot of work for... a not very fun or playable game? to prove some kind of point?