Yassine Meskhout's taxing vs punitive commitment contracts

This weekend I was at a party (hooray vaccines and the end of the pandemic!) and met a fascinating person who studied econ at GMU and has created some elaborate commitment contracts for himself. He had independently reinvented multiple aspects of Beeminder and figured out one thing in particular much faster than we did. Quoting him with permission:

The main issue I’ve struggled with is how to structure the contract and the accompanying penalties. The way I have categorized it could be considered “punitive” vs “taxing”, or “punitive” vs “strict liability”. With “punitive” regimes, you generally have some sort of “escape valve” in place [Beeminder’s “fine print” feature]. For example, if I’m supposed to work out every day but I come across a day where I literally have no time, or I’m feeling way too tired, or whatever, then the contract could allow an exception for those circumstances.

The problem with “punitive” contracts is that there is way too much discretion, and it quickly loses its force. Because you start to redefine what counts as tired or unavailable “enough” and start applying to more and more questionable situations.

In contrast, the “taxing” system doesn’t give a fuck about your excuses. You’re too tired or too sick or too busy to go work out? Doesn’t matter, pay up. The paying up is a clear demonstration that you’d prefer eating the penalty, and is a good and honest demonstration of what your capabilities are that day. Sometimes this system can be too harsh, like if you have the flu and it’s literally impossible to comply with your commitment contract.

Our current thinking is that if you want the “taxing” system, you opt in to “no-excuses mode”. Or you will, as soon as we replace “weaselproofing” with “no-excuses mode”. [UPDATE: now done!]

I’ve been groping my way towards Yassine’s taxing-vs-punitive idea for a while now and am pretty pleased with his much more succinct version. My version is collected in a bunch of sprawling notes for an eventual blog post called “Paying Is Not Punishment” [UPDATE: also now done!]. This has been pretty confusing for me because we rave about punishment all the time (mostly recently in our “Contra Positive Reinforcement” post) and many users genuinely prefer the punitive system. I personally prefer the taxing system and tend to be most impressed with users who think of Beeminder that way. But people are different and the same person may even need different approaches for different goals.

It’s probably worth clarifying, at least in your own head, when creating a goal, whether you want Beeminder to punish you for crossing the bright red line or just tax you for crossing it.

5 Likes

So true! And as someone who does better with the punitive model over the taxing model, there are still past goals of mine that absolutely did best under the taxing model than the punitive model. (The reverse is probably true for many who prefer the taxing version generally too, I suppose.)

Nodding head hard. May have just given myself an RSI.

It’s been more art than science for me to figure out which is which and I sometimes only discover it after my first derailment, but I’ve been trying to work out some basic markers about which do better under which and I’ll be curious what others think about it so as to steal their ideas collaborate to crystallize those thoughts. Either way, it’s worth thinking about it when you start your goal and committing to sticking with it until after at least your first derailment, when you have a little more info about the goal.

3 Likes

This is a near stream-of-consciousness reply, so caveat emptor.

  1. As another with economics training, I really like this punishment vs tax distinction. Somewhat embarrassed I hadn’t thought of myself, given my supposed training.

  2. For better and for worse, thinking of the Beeminder payment as a tax would give it less of a “I’m bad” sting than thinking it as punishment. Likely incentive effects.

  3. Thinking of Beeminder payments as a tax prompts me to think, “Well, how about I just tax myself.” Obviously, how I spend the “tax revenue” would impact incentives.

2 Likes

Agreed! I think for me, the distinction is how much in my power accomplishing a goal is – if I have a lot of control over a goal, then a tax system works great. If I frequently find myself physically unable to accomplish the goal for reasons entirely out of my control, taxing doesn’t work – it ends up feeling like I’m paying a tax for someone else’s screwups which leads to resentment.

For instance, my “do everything on my to-do list” goal works great as a tax. I’m not sure I’ve EVER called non-legit on that goal for non-technical reasons. I’m 100% in control of what goes on my to-do list (I write it out every night), and all I need to do is make ANY progress on every item to count, so even on the worst days it’s maybe 15 minutes tops. The feedback loop is short (“wow today I put WAY more on my plate than I could handle, I guess I’ll make sure tomorrow’s list is shorter”) and totally under my control. If I derail, it’s because I consciously decided that I’d rather pay $90 than spend the 15 minutes to touch everything on the list (and on rare occasions I have made that choice).

On the other hand, my eternally-problematic bedtime routine goal is absolutely one I would not have as a goal if I had to pay up every time I didn’t accomplish it. I have a toddler, and on nights when she’s not even asleep when my bedtime routine goal is due (9:45, sob) having to pay $20 on top of spending the evening with a non-sleeping child would absolutely lead to an immediate archive-with-extreme-prejudice. On the other hand, plenty of nights (…like tonight! but tonight’s a safe day :wink: ) I stay up late for no good reason doing pointless things before my basic daily chores are done, and paying a punative $20 feels more than fair. Does that lead to some waffling on whether said toddler going to sleep at 9:30 led to a legit or non-legit derail? Not gonna say it doesn’t, but the overall benefit of the goal is still there for me.

1 Like