Learning things in the just-in-time fashion seems like a good idea (so you can apply what you've learned right after learning it). So I agree it makes sense to combine study of meta-skills like "how to learn" with study of some object level skill, like learning some branch of math. Maybe try a different notetaking technique for every chapter in your math book, for instance. Productivity metrics are hard. Some ideas I've had are number of commits (for programmers, assuming you can get yourself to make each commit about the same amount of improvement to your software), using predictionbook.com to make predictions about when you will accomplish goals and learn to have an accurate model, and just keep qualitative observations (write down how much you got done at the end of every day or something). I recommend keeping an "experiment queue" in a text file, and each evening plan some experiment you will perform on yourself the next day. You can populate your queue by reading stuff written by productivity/learning gurus, or just brainstorm experiments for yourself. Here are some initial experiment ideas: meditation, log your activities every hour, work in a café or library, autofocus. An interesting technique that I used to good effect was to alternate 50 min. of work with 10 min. of relaxation (generally taking a short walk), and listen to white noise during the work period. It was actually pretty easy to condition myself to work when I heard white noise. Actually, I would focus so hard with the white noise that I also had a lower gear (listening to Internet radio station groove salad while chewing a particular flavor of gum) that I used for activities that didn't require intense concentration. I was getting deeper and deeper into this sort of radical self-hacking until last august when I got stuck using voice recognition to communicate with my computer, became really depressed, etc. I have a lot of ideas related to it, but it feels dishonest to share them because I'm n