I mostly directed it into making training corpuses for classical classification models instead. "oh I need 10k ground truth images to train on" *cracks knuckles* "better get tagging then!"

@gsuberland I'm very lucky I was able to turn my literal addiction to mapping power infrastructure in OpenStreetMap into a job...

@gsuberland I did a bit of data entry as a temp in my early twenties and I can confirm it can be very therapeutic, especially with a stock of podcasts to churn through while you do so (this was in the days of them being in an actual iPod). I was doing it for some medical research and the person supervising was astonished at how quickly I got through it all, just locked in and tapped in those bone measurements.

@jon it's the sort of thing where I'd end up with a custom frontend to maximise my flow and auto-check common errors, driving whatever they usually want data entered into. and then I'd *fly*. brain off copy paste loop machine.

@gsuberland I did consider it but I was on a very locked down Windows machine (rightfully so, I was working in a hospital on patient data).

there are widely-used open training datasets with fewer entries than the ones I have produced by myself by hand. I find it moderately alarming.

(the knuckle cracking isn't for dramatic effect, I just have fucked up joints and can't type if I don't crack them relatively frequently)