My use case is finishing the master project¹ of my M.Sc. Comp Sci degree that I’m working on! It’s about beating Adobe’s compression method for RAW photos using the SPIHT wavelet compression. It’s also a good example what happens (to me at least) when you only track the input, not the output for the entire thing is supposed to be 300 hours. And I’m above 400 already and that’s not counting the many hours I poured into it before I started beeminding: spiht – phi – beeminder
As you delightfully put it I noodle on the things that are already good enough (in my professors’ eyes) rather than moving on.
And that’s because I picked the topic for this project myself and am way too passionate about it and there is so many things still that I could play with and experiment but I gotta wrap this up.
And yes, I did actually manage to beat Adobe’s compression. Less is better:
Now with it having grown that much (the documentation got 60 pages long instead of what should have been < 20 pages) my professor suggested that we could publish it! So if you see me at an academics conference talking about wavelets later, now you know why 
It’s also my Beeminder origin story:
¹ Not to be confused with the master thesis. It’s supposed to be sort of like a very implementation focused mini bachelor thesis.