Beeminding deep work

Lots of people say that it’s much more effective to work on something for fewer long periods than more short periods. The idea is that there are big switching costs. Most of my goals are programming, which requires remembering what I was trying to accomplish last, the structure of the program, details about stuff it interacts with etc. It takes a while to load this stuff into your head and it falls back out of your head pretty easily. This comic is representative.

Unfortunately, the way Beeminder is structured is such that I have a bunch of different tasks to do every day. I have 10, 5, 5 and 20 hours per week allocated to different projects which ends up meaning I work on project a for a couple hours in the morning, then project b and c for an hour each, then project d for four (I have weekends off turned on). It’d be much better if I did solid blocks - spending Monday and a quarter of Tuesday on goal a, 0.625 of Tuesday on goal b and so on - but I’m akratic enough that my goals are almost always red.

Anyone have solutions for this?


A thought occured: I could get fancy with the rate editor. That’d require manual effort every day or writing a script and making sure it’s run daily.

Two ideas:

  1. If you start the goal with enough buffer (or get yourself to a point that you have the right amount of buffer), then, for example, you could set it up so that project A (and only project A) hits the red on Monday. Then, because it’s the only one, you just work on that one and know you’ve got to build up a bunch of buffer so that you can do the same with the other goals. But I think you’re saying that you’re akratic enough that this won’t work for you. So, idea the second:

  2. A meta-goal, where each day it looks at your Beeminder goals and counts how many are in the red (or maybe orange). And if there’s more than one, it derails you. There are other threads discussing things like this (I haven’t tried it myself though).

Good luck!


I totally feel you on this.

Anyway, another meta-goal idea would be a do-less goal where you track how many of these goals you work on per day on average, with an optimal maximum of, say, 1.2. And then you could add a force the other way by making sure you work on each of those goals some amount per week. I’m not sure how easy the meta-goal would be to implement without scripting, but as a programmer, I suppose you’re not necessarily opposed to scripting :slight_smile:

1 Like

40 hours for your four projects is aggressive. You probably need to spend 50+ hours a week working in order to achieve that, even if you estimated the times perfectly. But you probably won’t estimate the times perfectly, so your project to which you’ve allocated 20 hours a week might take 22, and the one which you’ve allocated 10 hours a week might take 8. So unless you’re planning on working 60+ hours a week, you should probably dial down the goals.

Personally, I’ve found a lot of success setting these types of “chunky” goals at least 25% under my real goal, and giving myself a largish (right now I’m at about 10 days) buffer. I try to evaluate things every week or two (but definitely when my days til derailment on one goal gets significantly higher or lower than the others). If my time per week was too low or too high, I adjust the hours per week. On the other hand, if I spent a lot of time on one goal for what I consider to be a one-off thing, I’ll just retroratchet. Unfortunately the opposite of retroratchet (subject to the akrasia horizon) is a premium feature, so if I wasn’t able to spend a lot of time on one goal for what I consider to be a one-off thing, there’s no perfect solution, so I lower the goal temporarily and set a reminder to raise it back up in a week or so. If the reason my days till derailment on one goal is just that I’ve been lazy, then I don’t adjust anything and just work extra hours to make up for it. If all my goals move in the same direction, I haven’t touched anything. At one point all my goals had 12 or more days to derailment. Now they are at 8-12, other than my weight loss goal which bounces around wildly and is currently at 25 (the meaning of “days till derailment” is very different with a weight loss goal; if I gain a few pounds in one day that 25 will become a 6 overnight).

I think the key to this strategy is to be honest with yourself when one or more of your goals drift (which for a lot of “chunky” goals is fairly inevitable). Did you just make a bad estimate, or was there a one-off event, or were you just being lazy? You have to really give that some thought. Probably even think about it, come to a conclusion, then wait a day and think about it again, before modifying the road. I’m a GTDer, so I treat this as part of my weekly review, but no matter what system you use it is helpful to set time aside to do a meta type analysis.

Anyway, I’ve rambled, but I found to be a very good article on how to deal with “chunky” goals. I’ve basically adapted my strategies from that.

1 Like

By the way, maybe there is a “perfect solution” for the one-off adjustments to lower your goal. In theory it should work, but I haven’t tried it yet. I’ll give an example. Let’s say I have a goal of doing 10 hours a week on some project, but I only have 4 hours done when Hurricane Matthew comes along and forces me to take the rest of the week off. Let’s say I decide that there’s no need to make up those 6 hours of lost work.

So what I could do is immediately change the goal to 4 hours a week, and set a reminder to myself to change the goal back to 10 hours a week after one week. That’d act as a sort of anti-retroratchet, provided you have enough slack (7 days of slack would always work, less slack would work in some situations). If the event is something like Hurricane Matthew and you do derail, that’s when you email the folks at Beeminder and tell them the derailment was not legitimate. It’s when you don’t derail, but merely lose some of your slack, that some sort of trick like this is necessary. (I’m arguably facing this right now. My kids are off school due to Hurricane Matthew, and like I said above I have 8+ days of slack on all my goals, but I’m probably going to get less done today than I would have gotten done otherwise. Plus the kids are also off Monday for Columbus Day. OTOH, I think I’m just going to ignore it, as I’m pretty far from derailing, my beeminder goals are probably already set a little below my real goal and am probably going to be able to get some things done today.)

Anyway, just a thought. I haven’t actually tried it.


Here’s an idea for a meta goal:

Give yourself a point for every solid time block that you work on a project. It could be work on any project, but for at least the length of time that you think your deep work focus requires.

I do something like this to back up my tagtime goals: to reduce the odds of needing to rely on randomness on an emergency day, I create a supplementary goal that is based on wall-clock time. It strikes me that a similar solution might help you avoid the emergency-day-on-everything syndrome.


I like how you converted it into a do-more goal. If you start out with your regular goals well in the green and that meta-goal in the orange or red, it should work out so long as the goals are realistic and you’re focusing on the more urgent goals first. If you find yourself fulfilling the meta goal by focusing on the least urgent goals (e.g. working on a 7-day-to-derailment goal while there are three 2-day goals pending), maybe you have to make the meta-goal more complicated (or add a separate, focus on the right priorities, meta-goal), though.