Max's beeminder journal


Must do tasks

I am still working on the implementation of these two goals, I need to do some more thinking and testing this week before I know what I am going to do with them.

Pocket reading task

I worried about my new pocket goal before creating it, considering ways of giving higher points to older posts, wondering if I should split off a backlog to handle separately, etc. In the end I added a simple +1 per article read as a ‘better than nothing’ solution with the intention to revisit it later.

Instead of messing with the goal, I found a simpler way to nudge myself to read through the older and ‘harder’ articles: I found the setting to reverse the list order in the pocket app.

I now see the oldest articles first, and need to scroll allll the way through the list to see new ones. I am working through the backlog, causing the ‘harder’ ones to float to the top, and removing the incentive to add a bunch of quick, easy things I would previously have read immediately instead of adding them to pocket.

The only issue would be that I will become out of date, or no longer interested in an article once I get to it, I see this as positive. Few of the things I save to pocket are time sensitive (few of any of the things I read are time sensitive). Things that no longer interest me I can delete, saving myself reading time.


That might be one of the most ingenious solutions to the backlog problem that I’ve heard! I bet it would be helpful for a lot of kinds of backlogs… It almost makes me wish for a reverse-chronological-order view for when I want to clean out my email inbox.


I have killed and added various goals around must do tasks and todo-list items this week, but everything is still very much up in the air, so I will write about it once everything has crashed to the ground and I have inspected the pieces.


One straght-forward goal I added this week is to make one metaculus prediction each week. I used predictionbook quite a lot for a year or so before losing the habit. Metaculus recently added the ability to create private questions as well as adding predictions to the well written public predictions, so I decided to restart that habit.
I like making predictions, I feel it makes me think more clearly by making me frame questions in a precise way (one needs to state clearly in which cases the prediction will resolve as correct – the world tends to be messier than expected). In theory, making lots of predictions might make be better calibrated as well, and thereby make better decisions.


I made a small improvement to my use of Beeminder this week. I have added a weekly task to check over my goals, retro-ratchet anything that is ‘too green’, adjust slopes to be harder or easier where necessary, and – very important – book any beeminder holidays in advance (this is something I always forget until two days before a holiday otherwise).

This week I made my pocket and anki-learning goals slightly harder, retro-ratcheting them so they float back up into the blue/yellow part of my list, so I can’t forget about them.

As the number of goals increases, I find myself using the colors more and more, ignoring anything blue or better. This works great for goals that I need to touch less frequently, like the goal to write this journal entry, but for daily habits such as anki study, flossing, etc., letting the goal get ‘too green’ lets me stop doing it for a few days, which seems to weaken the habit somewhat.

I now understand the need for the auto-retro-ratchet feature!