A Closer Look at Self-Commitments

What are Commitments — Part 3

Published in

CommitPool

12 min readMar 24, 2021

During this series of posts, we’ve shared how we at CommitPool think about commitments. We’ve outlined a Taxonomy of Commitments, which organizes commitments by their purpose and by the mechanism they use to constrain the committer’s actions. In Part 2, we examined how and when Incentive Constraints function in the context of commitments meant to persuade others. Now, we turn our focus towards commitments meant to keep the committer’s own actions in line with their preferences. We call these Self-Commitments.

Before we dive in, let’s recap our helpful characters:

Charlie is our committer. He seeks to constrain his own actions in order persuade Peggy to act as Charlie desires (persuasive commitment), or to ensure that he himself acts as he currently desires (self-commitment).
Peggy is our second party in a persuasive commitment. She is the person Charlie seeks to persuade by constraining his own actions.
Trey represents our third party. He is enlisted to enforce the incentive (e.g. penalty) on Charlie as part of an incentive constraint.

Self-Commitments: Deconstructing Charlie

In our previous posts, we treated our committer, Charlie, as a single entity. This was a useful simplification for examining persuasive commitments that already involve multiple characters. For self-commitments, though, where there are typically no other characters, it’s an over-simplification.

The reality is that there are two versions of Charlie. Charlie-now is the version who exists at the time Charlie decides to take an action at some point in the future, and may (or may not) make a self-commitment to that action. Charlie-later is the version who exists at the point in time when that action is supposed to be taken, who’s choices the commitment is designed to constrain. Restating our simple example of a self-commitment illustrates the distinction:

Charlie-now has a goal to exercise regularly, but he recognizes that when it comes time to actually do the exercise he often puts it off in favor of another activity. Making a self-commitment to exercise is how Charlie-now removes the discrepancy between how he wants to act and how Charlie-later acts.

This discrepancy between Charlie-now and Charlie-later exists for persuasive commitments as well as self-commitments, though the mechanism is different. In a persuasive commitment scenario, the discrepancy between Charlie-now’s selected actions and Charlie-later’s selected actions is primarily due to shifting circumstances. For example, circumstances may change because of Peggy’s action, or perhaps because of Charlie learning new information.

Recall that a persuasive commitment is a mechanism for Charlie to convince Peggy to take an action that she would only take on the condition that Charlie will respond a certain way. Charlie’s future response is the incentive for Peggy to do what Charlie wants. But once Peggy takes the action, then Charlie has already gotten what he wants and it is no longer necessary for him to persuade Peggy.

Without the need to persuade Peggy, there is no reason for Charlie to spend the effort to respond unless he literally cannot break his promise (physical constraint) or unless he will be penalized for breaking his promise (incentive constraint). The constraint in a persuasive commitment exists to ensure that Charlie-later acts as Charlie-now promises despite the new circumstances.

*Illustrative scenario: the Charlie Car Company needs Peggy Supplier to make some custom parts for their newest car model. The* ***Action*** *row describes what happens if Charlie makes a credible persuasive commitment to Peggy; the* ***Action (comparison)*** *row describes what could happen if he doesn’t.*

In the illustrative persuasive commitment scenario depicted above, Charlie’s preferences don’t change: he always prefers to acquire parts from Peggy so he can build his cars. But Peggy’s actions do change the circumstances that Charlie-later faces. The parts Charlie needs only work for Charlie Cars. If Peggy makes the custom parts for Charlie, as the only buyer Charlie-later will have considerable bargaining power and can demand a lower price. Knowing this, Peggy will be hesitant to make the custom parts for Charlie unless she has a commitment from Charlie-now to buy at an agreed-upon price. The shift in circumstances is what Charlie-now’s persuasive commitment is designed to address.

In a self-commitment, on the other hand, there is no Peggy action to shift the circumstances. Instead, the primary discrepancy between Charlie-now and Charlie-later is a shift in Charlie’s preferences over time. Specifically, Charlie-now prefers that he take an action at a particular future point in time, but once that time arrives, Charlie-later then prefers to take a different action. Even when the circumstances are exactly the same, Charlie-later often views other actions as more attractive than the action Charlie-now originally preferred.

We’ve all felt this preference-shift in the form of procrastination. In school, when a paper is first assigned, how many of us have made plans to work on it incrementally over several weeks only to choose to spend time with friends or video games or a book each time we meant to sit down to write? And how many of us have planned to go for a run tomorrow only to end up sitting on the couch instead?

*Illustrative scenario: Charlie makes a plan to start running regularly — beginning in a few days. The* ***Action*** *row describes what happens if Charlie makes a credible self-commitment to this plan; the* ***Action (comparison)*** *row describes how his shifting preferences might derail his plans without a self-commitment.*

In the illustrative self-commitment scenario depicted above, Charlie-now wants to get in better shape, so today he makes a plan to start running consistently beginning in 3 days rather than vegging on the couch. When the fateful day arrives, his circumstances haven’t changed (e.g., both his couch and running shoes are exactly where they were 3 days earlier). However, his preferences have changed! As the cardio labor looms but its health benefits remain far in the future, at this point Charlie-later would prefer couch-sitting. This preference-shift is what Charlie-now’s self-commitment is designed to counteract.

Physical vs. Incentive Constraints

A self-commitment is designed to counteract this shift in preferences. A physical constraint makes it impossible for Charlie-later to choose Charlie-now’s non-preferred action, no matter how attractive it may be in that later moment.

This can work well for preventing unwanted behaviors, such as not spending money or not eating unhealthy foods. Throwing away (or not buying in the first place) unhealthy snacks can be an effective way for Charlie-now to physically prevent Charlie-later from eating them. Or freezing his credit card in ice can be an effective way for Charlie-now to physically prevent Charlie-later from making an impulse purchase (by the time the ice has melted after taking it out of the freezer, Charlie-later’s better judgement may prevail).

But, for the most part, physical constraints don’t work very well for ensuring positive behaviors. Charlie-now can chain himself to a treadmill, but that doesn’t mean Charlie-later will actually do the workout. Self-commitments to positive behaviors typically require incentive constraints.

Penalties and Self-Commitments

To create an effective commitment, incentives must be applied fairly and be strong enough to influence Charlie’s actions. In the context of a persuasive commitment, Charlie often cannot rely on himself (as a first-party enforcer) to fairly enforce a penalty on himself. In certain circumstances he can rely on Peggy (as a second-party enforcer) to apply a penalty, and in others he can enlist various versions of Trey (as a third-party enforcer).

In establishing the incentive constraint for his persuasive commitment, Charlie can select from a few different types of penalties, ranging from financial to social & reputational to physical to psychological. See Part 2 for a rundown of these penalty type and enforcement dynamics for persuasive commitments.

What do these dynamics look like for self-commitments?

First-party enforcement rarely works

In a self-commitment, as in a persuasive commitment, first-party enforcement only works via psychological penalties. Charlie-now can’t trust Charlie-later to fine himself if he fails to meet the commitment. One thing Charlie-now can do is take advantage of Charlie-later’s expected loss of sense of self worth (or ego hit, or shame) on failing to keep a promise, by making exactly such a promise. However, this approach relies on the strength of that expected loss, which may not always be sufficiently high.

(Recall that if Charlie’s will-power were strong enough, or if he inherently enjoyed the activity on which his goals depend, then he wouldn’t need a self-commitment to help him stay on track.)

Second-party enforcement barely exists

Since Peggy does not exist in a self-commitment, there is no direct second party to serve as an enforcer. Unlike in a persuasive commitment, nobody else has an inherent reason to enforce a penalty on Charlie-later.

That said, while there are no direct Peggys for Charlie-later to potentially disappoint, Charlie-later may have future interactions with future Peggys in which his reputation could determine success. And to the extent that his future reputation includes Charlie-later’s actions with respect to the present commitment, the incentive to maintain a positive reputation would constrain Charlie-later’s present actions. This is in large part the mechanism people are tapping into when they publish (e.g., on social media) their personal goals.

Charlie’s friends and family are another type of indirect Peggy. While they probably don’t care about Charlie’s personal goals quite as much as Charlie himself, they do care enough that Charlie’s success or failure would impact them somewhat. But the other side of that care coin is a bias towards leniency and forgiveness, hardly effective ways to ensure that Charlie keeps his promise.

Third-party enforcement can work, but only if set up appropriately

Charlie can enlist a third party to enforce a penalty by granting Trey control over something valuable (e.g., money) Charlie owns, to be returned if Charlie-later succeeds in upholding the commitment or withheld as a penalty if he fails.

For this transaction to take place, Trey must meet some basic criteria:

Trey must have a reason to provide this penalty-enforcement service to Charlie
Trey must be trusted to impose the penalty fairly according to the rules established by Charlie-now. Note that this includes both the process of assessing whether Charlie-later has met the commitment and the process of penalizing Charlie-later based on that assessment. Trey is in control of both, and so Charlie-now must trust that he will be fair on both accounts.
Trey must be expected to not run away with Charlie’s money

In Part 2, we outlined a variety of actors who can play the role of Trey within a persuasive commitment. Let’s see how well they meet these criteria in the context of a self-commitment.

Individual person
A friend or family member of Charlie’s potentially cares about him enough to be willing to help, and is unlikely to run away with Charlie’s money. However, just like in their second-party role, their care for Charlie may bias them toward leniency and let Charlie get away with failure without a penalty (true “tough love” is a rare thing). This might not work well.

An enemy doesn’t have much of a reason to help Charlie. An enemy would probably be too strict and over-penalize him anyways, and is liable to run away with the money immediately. This is not going to work well either.

A random person might be neutral towards Charlie, but has no reason to help and, even if he did, also has incentive to run away with his money or over-penalize Charlie to keep some for himself. Same story.

Organization / Company
An organization of which Charlie is a member looks a lot like a friend/family, and an opponent organization looks a lot like an enemy. Similarly, a disconnected organization looks a lot like a random individual. However, its status as an organization opens up a couple angles to address some of the issues.

An organization (likely a company, in this case) can create a business out of providing penalty enforcement services. Paying for these services is an excellent way for Charlie to give the business a reason to help him. Additionally, that kind of business relies on repeat customers as well as brand reputation to attract new customers and earn greater revenue in the long term, and so would be less likely to simply run away with Charlie’s money.

But there’s a problem: what happens to the penalty itself? Since a service like this is in full control both of assessing whether Charlie should be penalized and of imposing the penalty, any scheme where the penalty money flows through the company — and especially when kept as direct revenue — creates a perverse incentive for the company to over-penalize Charlie.

Any incentive to over-penalize undermines the value proposition of the penalty enforcement service. When Charlie has reason to believe that his penalty may not be enforced fairly, its influence over his behavior is weakened and his commitment will be less effective. Companies facing this problem can counteract it with customer service policies that give Charlie final say over whether a penalty was applied fairly, but that reverts the third-party enforcement back to first-party enforcement, which we know is typically not effective.

Legal system
A legal system can’t enforce anything without a second party to trigger the enforcement (e.g., by bringing a lawsuit). Even if it were technically possible to take yourself to court, Charlie-now wouldn’t trust Charlie-later to do so — for the same reason that first-party enforcement doesn’t work.

Smart contracts
As with persuasive commitments, smart contracts work well for self-commitments because, once established, they are guaranteed to follow the rules for assessment and penalty enforcement encoded within them, and can do so without relying on individuals or organizations. A smart contract can accept payment from Charlie-now and it will never steal his money nor inappropriately penalize Charlie-later.

Tools: a different type of enforcer

Smart contract penalty enforcers function without intervention from the types of actors typically thought of as third parties. Should, then, they be classified as third party enforcers? This is a valid question!

In fact, we claim that smart contracts are best thought of as a different category altogether: a tool. Unlike a first-, second-, or third-party enforcer, a tool does not have agency. It has neither interests of its own nor the ability to intentionally diverge from the behavior intended by those who wield it.

Specifically, smart contracts are a first-party enforcement tool. Charlie can use a smart contract himself to establish a penalty, and then relinquish control over its enforcement to the smart contract! Using such a tool, Charlie can now hold himself accountable without relying on a second or third party to keep him honest. And he can do it with stronger and more reliable incentives than his own will-power or psychology can muster.

CommitPool: Come for the tool…

We’re building CommitPool because we think a first-party enforcement tool is the best way to help people (including ourselves!) stick to their personal goals. Third-party enforcers for self-commitments, in the form of web2 services, sound great — and sometimes can even work quite well — but ultimately fall short because they rely on faulty incentive structures.

A third-party enforcer gets paid if you fail to meet your self-commitment, and they have control over both the assessment of whether you fail and the decision to penalize you. The mere possibility that they can cheat you means that eventually they will be mighty tempted to cheat you. They can prevent themselves from cheating — and give you more reason to trust them — by allowing you to refute their assessment and/or enforcement decision, but all that does is give Charlie-later the means to avoid penalization, and we know how untrustworthy Charlie-later is.

This is why CommitPool uses decentralized web3 technologies to assess whether you fulfilled your commitment, and smart contracts to enforce the penalty conditional on that assessment. We are extremely sensitive to the fact that CommitPool is only as a credibly neutral tool as its most centralized component. Soon, we’ll have a post up exploring the trust assumptions within the CommitPool protocol and outlining our plans for progressive decentralization.