So anyway, August review time:

- A/B testing: the lineheight test (http://www.gwern.net/AB%20testing#line-height) found no difference, so I did a quick one where I tested an empty test to check the A/B testing tool I'm using (http://www.gwern.net/AB%20testing#null-test); I successfully failed to reject the null. The next test is whether underlining hyperlinks annoys people or not.
- http://www.gwern.net/Book%20reviews I wrote a Haskell program to parse my GoodReads ratings & reviews into flat Pandoc Markdown; it works somewhat well, it seems to be eating blockquotes & neutering hyperlinks, I'm not sure why. Was also an opportunity to clean up some reviews: inline some of them, spellcheck, expand references & links, which was a lot of work. But it's nice to have my reviews gathered somewhere with a readable interface. kiba thinks the work may pay for itself in affiliate revenue with Amazon, but I'm skeptical.
- http://www.gwern.net/docs/2002-radiance Added a hundred pages or so. Annotating some of it is quite difficult; Scholz's familiarity with Wagner's operas is a challenge, since I've only ever read his Ring Cycle.
- I've started two new self-experiments:

    - Magnesium citrate on mood/productivity: http://www.gwern.net/Nootropics#citrate     - Treadmill desk usage on spaced repetition scores: http://www.gwern.net/Treadmill#spaced-repetition
- http://www.gwern.net/Touhou#growth-over-time made a little more progress on Touhou music, with # of releases vs time: seems like we may've reached peak Touhou in 2009 or so.
- http://www.gwern.net/Silk%20Road#archives-of-sr-pages I've started hosting public copies of subsets of the black markets
- https://groups.google.com/forum/#!msg/mnemosyne-proj-users/8yZz9BbqKl4/BxAIjo_2WscJ I've been analyzing my Mnemosyne data and the giant public database for time of day effects. While my results aren't conclusive, my analysis of 48m flashcard reviews from the public database finds that the best time to study your flashcards seem to be noon. A little surprising, you'd think that late at night, before bedtime, would be the best time.

My Sand polls aren't done because I got a much lower rate of responses to the polls than I was hoping for and only got enough responses in the final poll the other day. :(
Shared publiclyView activity