Commitment Devices are Great but Don't Last Long

The Problem with Commitment Contracts like Beeminder and Forfeit

Everyone is composed of two selves, their long-term thinking self (cortex-driven) and their monkey self (limbic-driven). Much of human action is an internal fight between these two, also called akrasia: there is some action one part of you knows you should be doing but instead a different part of you chooses to do something else.

At the start of a new habit, motivation is high enough for long-term thinking you to quash monkey you's desires (e.g. do hard exercise for the long term benefit). After this motivation wanes however, monkey you starts winning those battles (e.g. skipping out on exercise because he doesn't feel like it). Commitment devices like Beeminder and Forfeit use this to their advantage because you're initially highly motivated to create the commitment contract that later constrains monkey you to act in the way long-term thinking you wants.

After a while however, monkey you wants to leave this contract/system because they can only be punished, or reach neutral at best, and will do so when able. As a result, long-term retention is low, and habits fade.

Also, because the use of a commitment contract is somewhat extreme, you have to be highly motivated to begin using the system. Lots of casual users who do want to start new habits but aren't as intensely motivated don't use the system. As a result, the activation rate is low, and only early adopters but not mainstream customers use the system.

To solve the retention issue, rather than using positive punishment (which is what the commitment devices are based on), what is needed is positive reinforcement. Monkey you is excited to use the system because of rewards from accomplishing habits. We know from habit science that rewards are the most critical component in establishing and maintaining a habit, and that positive reinforcement is the most effective long-term strategy. As long as the reward exceeds the friction of the task, and it is the best reward he can consume in his environment, monkey you will fall into an optimal default aligned with long-term thinking you's wishes.

To solve the activation issue (that you must be highly motivated to begin using the system, which turns away most people), perhaps a less daunting commitment contract, positive reinforcement, some combination of the two, or something else is necessary.

One idea to solve the activation issue is to turn the concept of hyperbolic discounting to your advantage! Monkey you chooses small immediate benefits over large long-term costs (hyperbolic discounting, which has valid evolutionary origins) while long-term thinking you will delay immediate gratification for those greater long term benefits. When choosing to engage in a commitment contract, the main reason you don't is because a bit of monkey you comes out and complains that he'll be constrained in the future to pain or loss of opportunity.

What if you told monkey you that he would only be constrained for 30 seconds? That maybe would be fine with him; after all, anyone can do anything for 30 seconds. But you would also tell him that the amount of time he would be constrained would slowly increase (exponentially, say 1%) over time, such that in 200ish days he might be constrained to about 5 minutes of activity. Since it is so far in the future (and intentionally nebulously defined), monkey you performs his usual hyperbolic discounting and agrees to it! In this way, by using temporal, exponential (slow in the beginning so that it's ignored, fast later so that it becomes useful) modifications to the system, we can trick monkey you into agreeing while also benefitting long-term thinking you.

Email me at max@atnself.com if you have some thoughts (the temporary solution until I add comments).