If on the morning of a beemergency you realize that you will only be able to make it halfway to the absolute minium the “rational” thing to do is to not even do that because you will derail either way. This got me thinking about partial derails, when you get halfway to resolving your beemergency you only pay half your pledge. In a way this is an extrapolation of Beeminder’s philosophy of turning lumpy incentives to make lumpy progress into incremental incentives to make incremental progress. Going from pledge sized incentives and daily progress to cent sized incentives and fractions of daily progress.
Looking back at my own derailments there would have been instances where I wasn’t motivated to do everything needed to save 5 bucks because that felt overwhelming but would have done a few minutes more to save a few cents. Maybe this would even have lead to me not derailing after all by repeated application of this “just a few minutes more aren’t that bad”.
For goals with a post derail respite this would lead to the new incentive problem of do 99% of the bare min, pay 1% of the pledge, get all of the post derail respite. This can be solved by also applying the post derail respite proportionally. This might even be worth it on its own since it makes the breathing room I get after a derail more proportional to how much I struggled with the goal.
There is also the question of if this makes sense financially for Beeminder as this will on average decrease the revenue they make from a derailment. On the other hand knowing that you don’t have to pay the full pledge if you partially do the bare min might lead to more ambitious goal settings. This plus partial post derail respite would lead to more frequent derails, hopefully making it not a revenue negative feature.
One idea I’ve played around with that gives you some of the same effect is “tiered goals”. Works best for autodata goals (otherwise it’s extra data entry, or using IFTTT or something to do the same).
Let’s say I want to do 10k steps a day, measured by my fitbit/apple/garmin device. I link it to two goals: one is 6k/day, the other is 10k/day. Then if I only manage 7k steps, at least I only derail on one goal.
I’ve been thinking about this issue for a while as well, less about paying half the pledge, but more being able to partially ratchet
I have several goals like “Do not eat junk food before Noon” and “Do not eat junk food after Noon, before sleeping” (i.e. splitting the day into 2 Do More goals to achieve “doing less of something”)
Let’s say that I eat junk food in the morning at 10 AM, and this will cause me to derail.
Then there are a few possible options:
If deadline is set for sometime during the day (e.g. Noon) – I derail at Noon and now need to do 2x of that goal that day, which is impossible, so I either need to fake data or have 1 day buffer (which then frees me to eat more junk food for another 1-2 slots guilt-free)
Deadline is set at midnight – maybe I eat junk food at night too because I know I’m going to derail that night, so I either (a) derail 2 days in a row if buffer is 0, or (b) have 1 day buffer (which again frees me to eat more junk food for 1-2 slots)
And there’s no fractional ratcheting, so I couldn’t say ratchet half a day away to solve the problem.
I finally have a solution to this, Clive’s idea got me thinking about it.
I’m testing out 2 goals, one for each block of time in the day, which then both feed into a metaminder goal, which I set to the same rate as my original goal.
Now I can ratchet the subgoals independently, optionally add separate derailment fees, etc. while still maintaining the overall baseline rate in one-place via the meta goal.