About a year ago I decided to try using one of those apps where you tie your goals to some kind of financial penalty. The specific one I tried is Forfeit, which I liked the look of because it’s relatively simple, you set single tasks which you have to verify you have completed with a photo.

I’m generally pretty sceptical of productivity systems, tools for thought, mindset shifts, life hacks and so on. But this one I have found to be really shockingly effective, it has been about the biggest positive change to my life that I can remember. I feel like the category of things which benefit from careful planning and execution over time has completely opened up to me, whereas previously things like this would be largely down to the luck of being in the right mood for long enough.

It’s too soon to tell whether the effect will fade out eventually, but I have been doing this for ~10 months now^[1] so I think I’m past the stage of being excited by a new system and can in good conscience recommend this kind of commitment mechanism as a way of overcoming akrasia.

The rest of this post consists of some thoughts on what I think makes a good akrasia-overcoming approach in general, having now found one that works (see hindsight bias), and then advice on how to use this specific app effectively. This is aimed as a ~personal reflections post~ rather than a fact post.

Thoughts on what makes a good anti-akrasia approach

I don’t want to lean too much on first principles arguments for what should work and what shouldn’t, because I was myself surprised by how well setting medium sized financial penalties worked for me. I think it’s worth explaining some of my thinking though, because the advice in the next section probably won’t work as well for you if you think very differently.

1. Behaviour change (“habit formation”) depends on punishment and reward, in addition to repetition

A lot of advice about forming habits focuses on the repetition aspect, I think positive and negative feedback is much more important.

One way to see this is to think of all the various admin things that you put off or have to really remind yourself to do, like taking the bins out. Probably you have done these hundreds or thousands of times in your life, many more times than any advice would recommend for forming a habit. But they are boring or unpleasant every time so you have to layer other stuff (like reminders) on top to make yourself actually do them. Equally you can take heroin once or twice, and after that you won’t need any reminder to take it.

I tend to think a fairly naively applied version of the ideas from operant conditioning is correct when it comes to changing behaviour. When a certain behaviour has a good outcome, relative to what the outcome otherwise would have been, you will want to do it more. When it has a bad outcome you will want to do it less.

This is a fairly lawyerly way of saying it to include for example doing something quite aversive to avoid something very aversive; or doing something that feels bad but has some positive identity-affirming connotation for you (like working out). Often though it just boils down to whether you feel good or bad while doing it. The way repetition fits into this is that more examples of positive (negative) outcomes is more evidence that something is good (bad), and so repetition reinforces (or anti-reinforces) the behaviour more strongly but doesn’t change the sign.

A forwards-looking consequence of this framing is that by repeating an action that feels bad you are actually anti-reinforcing it, incurring a debt that will make it more and more aversive until you stop doing it. A backwards-looking consequence is that if the prospect of doing something feels very aversive it’s probably because it had a bad outcome on average the previous times you did it.

In some cases it’s easy to see upon reflection that this accounting your brain is doing is wrong. E.g. doing the washing up has small positive effects that persist for much longer than the amount of time it takes. Or it could be something like applying for jobs, where you expect to get only negative outcomes until you get one big payoff that fixes the expected value. The accounting can also be not-wrong, your brain can be tracking some real negative outcome that you are trying to talk yourself out of.

I think it’s good to see this as real and as something to be negotiated against (by trying harder to make the outcomes positive, or layering on additional punishments and rewards), rather than simply powering through with repetition.

2. The approach should be counter-cyclical

It should be easier to get back on right after you fall off than it is to stay on.

An example of something that doesn’t have this property is aiming for a streak of doing something every day. If you have a 90 day streak on Duolingo, that’s good for motivating you to keep it up. But if you fail one day then it’s much less motivating to do it the next day because you know it will take 90 days to get back to your previous PB.

This is especially bad because you don’t usually just fail things randomly^[2], but because you’re going through a period of stress. So you are already prone to fail several days in a row and then you also lose the positive nudge from maintaining the streak at the same time.

Streaks are good for getting you from 90% to 99% compliance with your intentions, but not from 0% to 80%^[3]. For most things each percentage point is about equally valuable, and therefore it’s more important to have a system with stronger incentives (positive or negative) at the lower end of compliance^[4]. For example the “just commit to doing 5 minutes of the thing you're avoiding” trick is in the right direction on this.

The analogy to counter-cyclical fiscal policy works quite well, with the idea that the underlying system is overly pro-cyclical to begin with. One tends to go through phases of things going well, this motivates you to commit to more, eventually this becomes overwhelming (possibly triggered by some short term setback) and you are forced to pull back a bit (maybe bail on some commitments or drop some side projects), this is then demotivating and causes you to pull back more than you should.

Because of this it can be worthwhile to follow a strategy that is exaggeratedly counter-cyclical, where basically most of the incentives are about getting you back on the wagon when you fall off rather than keeping you there.

3. Poor planning can be as big a problem as poor execution, and improving one improves the other

Akrasia is sometimes identified with procrastination. It is assumed you already know what the right thing to do is, and if only you could motivate yourself to do it everything would go brilliantly.

I don’t relate to this that strongly, I think it’s just as often that my plan of what to do is bad and so the procrastination is my brain reacting to the expectation of lacklustre results.

A specific genre of bad plan is a vague intention of doing something that is not that relevant to your life. “I should take cold showers every morning” or “I should finally get round to learning Mongolian throat singing” one thinks to oneself, to which one’s subconscious thinks “that is the dumbest idea I’ve ever heard, I need to make this the most aversive task imaginable so that maybe he’ll get round to answering those emails”. A plan to answer the emails would be more effective.

There is a vicious cycle here of not having a particularly good plan, then not executing said plan wholeheartedly and therefore not gathering much evidence on whether it was a good plan or not. Improving the plan or the execution would improve the situation, a better plan would be more motivating because you would expect a better outcome, and better execution would more effectively update you on whether the plan was good or not and what to change about it.

I like the definition of akrasia as “acting against one’s own best judgement”, where the best judgement may be better than the thing you have come up with so far, and so spending more time planning/scheming counts as far as overcoming akrasia is concerned.

Using Forfeit specifically

The way the app works is you set a specific task (a “forfeit”) with:

• A deadline
• An amount at stake if you don’t complete it
• A verification method, with the default being sending a photo or screenshot, which someone manually verifies

This is what it looks like to set a forfeit:

Chronologically, the way I got into using it was:

1. I came across it while I was browsing the Beeminder blog one day, and started using it for some small admin things
2. At the time I was trying to get this project over the line, and I started setting forfeits to help with this (£15 for “2 hours in vscode”, £30 for “>100 lines of code merged”, £100 for “publish the announcement post”). This was the first “killer app” that made me realise how effective this was as a motivating tool
3. Despite having seen the light, my usage petered out once that project was shipped and I eventually stopped using it for a couple of months
4. Then when I was off work over Christmas I thought to myself “wow everything in my life was going way better for a few months when I was using that app”. So I took some time to plan how to use it consistently for things that are at least somewhat important to me:
   1. Since I had found it useful for tracking screen time, I set up RescueTime to do that better (previously I was using the default Mac screen time)
   2. I set up a recurring forfeit for limiting time on Twitter (details below, this was the second killer app that turned out to really surprisingly effective compared to stuff I had tried before)
   3. I set it up as a backstop for Habitica (daily habit checklist), where I would have to pay a forfeit if I didn’t at least open Habitica once per day^[5]
5. This was enough to make it self-reinforcing, and since then it has been gradually eating the world of how I organise my life. Most things that I used to put in my calendar or a to do list I now put straight into the app

One lesson here is the fact that you can start with small one off things makes it a very easy onboarding experience compared to (to me, seemingly overcomplicated) alternatives like Beeminder.

Another is that it’s not immune to the problem of simply stopping using it despite it being effective that most productivity systems suffer from. But the ability to set recurring forfeits as a backstop for sticking to your own system makes this much easier to avoid, because unlike other systems you can’t just close your eyes or walk away without losing some money.

What I have found it most useful for

Here are some stats to prove that I do in fact use this a lot every day, the bulk of these are over the last 4 months:

I’ve looked back at the ones I have completed and thought about the categories that have turned out to make the biggest improvement to my life. Here they are in roughly decreasing order of value (absolute benefit, not benefit/cost):

1. Commitments to reply to people (or message them in the first place)
2. Work related forfeits that commit the bulk of my time each week
3. Small admin things (work or non-work related)
4. Non work related forfeits that commit big chunks of time (e.g. writing this post)
5. Reducing time on twitter/youtube
6. “Daily habits”

The top two being replying to people and work commitments fits well with the idea of relevance to my life being the key factor. “Your relationships with other people” and “whatever you do all day most days” account for the bulk of how well one’s life is going, and so it’s unsurprising that mundane commitments related to these have turned out to be the most valuable. It's a bit disappointing though, I would have hoped that creative endeavours scored higher.

I like SMART (Specific, Measurable, Actionable, Relevant, Time-bounded) as a good generic piece of advice for setting goals. Although I now think relevance and measurability are where most of the action is. “Does it matter if you do it?”, “Will you be able to tell whether you’ve done it or not?”.

Commitments to reply to people (or message them in the first place)

Committing to replying to people has been by far the best cost/benefit (due to low cost), and I would say narrowly beats out work related ones for absolute benefit (these take up a lot more time).

I used to feel guilty a lot of the time about putting off replying to people, and would become more averse to doing it the longer it went on, and then also feel bad about any new messages I received which added to the sense of obligation. This is obviously hurtful to other people too because they get the idea that you’re deliberately ignoring them if you do it enough.

I am almost completely cured of this now. Now whenever I get a message that I don’t respond to immediately I set a forfeit to reply within a day or two. I usually make this low stakes (£5), to be proven by a redacted^[6] screenshot.

If I fail then I recreate it with an increased amount, but I rarely fail these ones. The vicious cycle only starts once I’ve put it off long enough that I start to feel guilty so setting a small forfeit nips it in the bud.

Work related forfeits that commit the bulk of my time each week

For work, I mainly set two quite high stakes forfeits each week, an output one (£50) and an input one (£30):

1. We do weekly sprints (I’m a software engineer), so I set one to have all my highest priority tasks at least in review by the penultimate day
2. I set a goal for overall time spent on the “Software Development” category in RescueTime, which corresponds quite well to “time doing the main things I’m supposed to do for my job”^[7]

The output one I’ve hit 7/8 times since I started it in its current form. In the 8 weeks just before I started using Forfeit regularly I would have hit it… 2/8 times. The hours worked one is more like a backstop to make sure I’ve put in an honest attempt, I set this one at a fairly comfortable level and have always hit it.

Obviously going from rarely hitting my goals to almost always hitting them is much better for a million different reasons. Among them not being derailed on longer projects due to underestimation. And also of course, I feel better now knowing when I start a week that I’m very likely to do the things I’ve written down.

There is a question of how much of the change here is due to increasing output vs reducing underestimation. I don’t have good enough data to show this convincingly (I don’t have screen time data from before I started these commitments), but I think the answer is that it’s:

1. Mainly reducing underestimation
2. With some benefit from not veering off into doing things that are easier but less important
3. And a smaller benefit from increasing time-on-task (<20% increase)

Small admin things (work or non-work related)

As with replying to messages, I create a low stakes forfeit for every small admin thing that I don’t do immediately (taking the bins out, paying bills etc).

Individually these are pretty small, but in aggregate it has made a big difference to not have the feeling that there are a million little things I should really get round to. Hence the relatively high ranking.

Non work related forfeits that commit big chunks of time (e.g. writing this post)

I often have various schemes and side projects going on at any given time, and only rarely do these come to fruition. I have been using financial commitments to put more time into these. However I have barely a blog post to show for it, so I can't rank this too highly.

I hope this is just a matter of things in the real world taking time to get right, so in a year's time I would consider this to be a more valuable category.

As mentioned the one shipped thing where Forfeit made a noticeable difference was manifolio. It made it a lot easier and less stressful to commit the time to, but there was a hard deadline so I think I would have found a way to do it anyway.

Reducing time on twitter/youtube

I have the following setup for making sure I don’t spend too much time on twitter:

1. RescueTime installed on all my devices
2. Freedom (app blocker), with the “locked mode” turned on so it can’t be disabled mid session. The main purpose of this is to permanently block the twitter website on mobile, because RescueTime can’t track individual websites on android, only apps. I also find it very useful in general for blocking things temporarily
3. A daily £30 forfeit for spending less than 2h30m on twitter + youtube combined across all devices. Proven by screenshots from RescueTime
4. A separate weekly forfeit for confirming that RescueTime is active on all my devices (partly to prove I haven’t cheated by disabling it, partly because I have found RescueTime to break sometimes)

The main goal of this is to make sure I don't ever waste a whole day on twitter, hence the fairly high limit and high penalty. For this one I have evidence that it has made a significant difference. Here is a graph of my time on twitter + youtube over time (units removed because I’m too embarrassed):

Again this data only starts at the same time I started this commitment so the actual decrease is a lot bigger.

This one has freed up time to do other things I think are more valuable (like some of my ill fated side projects). I’m not ranking it higher because I actually do love twitter so I feel like I’m missing out a bit now.

“Daily habits”

Some things in this category:

• I have one for going to the gym 3x/week
• I have one for doing the washing up before I go to bed
• I had one for doing Anki reviews for a while
• I had one for journaling for a while
• I have the one for making sure I open Habitica, where there are a few more things I have to tick off

The prospect of losing money has made me better able to stick to things like this. Although, things you do every day are relatively easy to stick to anyway. Also I haven't found anything like this to be completely life changing (unlike the concept of replying to people in a timely manner). Honestly, the washing up one has actually made the biggest difference to my life, though I expect I’ll thank myself for going to the gym eventually.

In a way I feel I have at least got the “little daily habit” mind-virus out of my system. Previously, I was unable to stick to a not particularly stimulating habit for long enough to really know if it was worth it. Now I can do this quite easily. And, having tried journaling daily for a couple of months, it just didn’t do anything for me.

Having gained the ability to make arbitrary commitments fairly reliably, it seems to me now that the main reason every-day habits are valorised is because they are easy to stick to, and that in fact every day is way too frequent for most things.

I now try to be quite strict on cutting daily tasks that I don’t think are that useful, because even if they are individually small, having to do them every day means they contribute a lot to the total hassle of random stuff you have to do. You don’t want to end up having 10 tasks you need to clear before you can leave your house on a given day.

Tips & tricks

Organised in roughly decreasing order of importance (plus bonus points for non-obviousness and simplicity):

Learn to embrace, or at least tolerate, the work of verifying and operationalising

It can feel stupid for it to take longer to verify that you’ve done something than it takes to actually do it.

I see your feeling stupid and raise you “actually, it’s fine”. It might only take 10 seconds to reply to a message from your mum, but if creating a forfeit to do so increases the chance of replying promptly from 70% to 95% then it’s probably worth it, she misses you. In the olden days replying to your mum would take at least 30 minutes to write a letter and find a postbox, and even this would not be a good reason not to do it.

Generally speaking the category of things you will want to put into the app will be things where you feel averse to doing them even though they are pretty clearly worthwhile. For these it’s especially worth it to spend some time up front thinking about how you can break it down into verifiable commitments.

Use RescueTime, via screenshots (not the integration)

As referenced above I use RescueTime a lot for time-on-task and limiting-screen-time commitments. In addition to recommending you set this up at all, I recommend that you submit screenshots rather than using the integration.

I haven’t been able to get the integration to work properly, and in general my experience with plugins for niche apps is that they tend to break. Screenshots work perfectly well though.

Set the amount at stake high enough that you care directly about the money and not about the personal failure

Setting a stake of £1 for something that really needs to get done is not that much different from writing it down in a to do list. You should set the stake high enough that the cost of losing the money is in the same ballpark as the cost of the task not being done. If you earn a lot of money, or the task is very important, this might be really high!

The actual process I follow in my head for setting the amount on an individual task (that is not trivial) is like so:

• I have some standard levels that I usually use (£5, £15, £30, £50, £100)
• I pick one that seems reasonable
• Then, given that I know my failure rate is about 5%, I ask myself “would I pay 5% of this money right now just to make this commitment (e.g. if I had to pay a fee to a bookie)? Would I pay much more?”. I then might go up or down a level as applicable and repeat

For big ones I sometimes write this down outside my head. The reason I think this is better than basing it on the full amount is that, given that most of the time you don’t fail, it’s fine for losing the full amount to be much more painful than not completing the task, in fact that's the point. The ex ante expected cost is a better point of comparison.

I also keep track of my monthly expenditure in the app and try to adjust things up and down (by committing to more/less, or changing the amounts) based on this. This might not be that useful to you unless you are a power user (fail several per month), but for me it’s more useful than reasoning about the individual tasks.

I think of the actual expenditure per month as the level of “incentive pressure” I have put on myself, and having decided on a specific expenditure to aim for my goal when setting forfeits is to size them correctly on average to hit that amount the next month. So even if I thought ahead of time that I was committing to things that are really difficult, if I had only spent £20 at the end of a month that’s the ultimate evidence that they weren’t in fact that difficult, so I would be more ambitious next time.

Meta forfeits are overpowered

This is possibly the main thing that sets this apart from other productivity systems, you can build the incentives to use the system effectively into the system itself.

Break down large tasks into multiple sub-tasks

There are at least 3 reasons to do this:

1. You usually want to get partial credit for partial success, to avoid the situation where it becomes clear that hitting your big goal is impossible long before the deadline and hence giving up
2. The UX isn’t great for tracking a big deadline far in the future (it doesn’t stand out from the other tasks). For things like this I usually try to break them down into shorter-term checkpoints (in addition to the big deadline task), or at least have a self-verify “check in on [big task]” forfeit once a week
3. Loss aversion and common sense says it’s easier to be rational about several small losses than one big one

Non-cheatability should increase with the stakes

As the amount you will lose goes up it becomes more tempting to cheat, and this makes you feel conflicted even if you don’t end up cheating. The whole point of the app is to take away the willpower aspect, so making tasks hard to cheat is good. On the other hand it’s more of a hassle to write tasks that are really hard to cheat, so you should try and increase how hard it is to cheat as the stakes increase.

You can make it harder to cheat by:

• Using a more strict verification method: GPS > Photo > Self-verify (I don’t use the Timelapse option but that also seems very hard to cheat)
• Being more lawyerly about the wording, trying really hard to cover edge cases
• As above, breaking things down into multiple tasks, or at least multiple forms of photo evidence required for one task

Challenging task with less at stake > comfortably doable task with more at stake

I started off doing things more on the comfortably doable end of the spectrum, and have shifted over time to the more challenging end (while trying to keep the overall rate of expenditure similar, so lowering the amounts to account for failing more often).

There are a couple of reasons I have found this to be better:

• I tended to put things off until close to the deadline anyway, so thought it was better not to pad them out beyond what I thought was a reasonable time to take
• When I didn’t fail any for a while I got “streak-brain”, where the feel good factor of having a 100% success rate was motivating me rather than the cold financial calculus. I think this is bad because losing money at the end of that makes you want to quit (which, fortunately, you are unable to do immediately). Losing £5 every day or two is easier to see as just a budget line item.

Don’t be too precious about appealing/bending the rules

It can be quite hard to set a goal that perfectly captures what you want, and I have found it’s better to not get too bogged down in this and just use the appeal feature to change it if needed (I also just use the “Normal” leniency mode as opposed to the no-excuses mode). This involves explaining yourself to someone (it appears to be usually a co-founder, Josh, that reads them). This is naturally a bit embarrassing which puts you off really abusing it.

Also, I think it can be fine/better to appeal when you nearly hit a goal, again to avoid all-or-nothing giving up. E.g. for “spend X hours doing Y before 7pm” tasks it sometimes gets to 5pm and I realise it’s now not possible to technically hit the goal (because there aren’t enough hours left in the day). In this case it’s better to still try to hit X hours but go over the time limit and appeal, rather than giving up altogether.

See also: Eight Short Studies On Excuses

Recreate forfeits that you fail once

It doesn’t prompt you to do it automatically, so this is an easy way for things to slip through the cracks.

Ask them to raise your max stakeable amount

It is set quite low initially, I think £50 for the UK. I don’t set too many above this but it’s nice to have the option. You can ask in the chat (top right of the screen) to have it raised.

Verify retroactively where possible

Some things can only really be verified by getting your phone out and taking a picture while you are doing it. This is annoying for a number of reasons:

• If you’re with other people, it’s annoying for them and embarrassing for you
• You have to break out of what your doing and open the app more frequently, you can’t batch things together
• You can forget to take a picture. If you then allow yourself to appeal in such cases this is a slippery slope to cheating, because you’re back to relying on your own honesty rather than evidence

It’s much better for the evidence to be created automatically, which you can then submit at your own leisure. It’s not always possible to do this, but here are some tactics for doing so:

• As usual, use RescueTime or other trackers that can run in the background for anything on the computer
• Set GPS forfeits where possible, even if they don’t perfectly cover what you are aiming to do (e.g. “be at the gym” is good enough for me usually). You can also use your google maps timeline (which generally records everywhere you go unless you disable it) as reasonably good evidence if you need to appeal something where you forgot to take a photo
• Sometimes you can come up with a trick. For instance to make sure I take 5g of creatine a day I have a forfeit to show that the pill bottle is 35g lighter than the week before. It's worth putting in the time to come up with tricks, especially for ones that repeat

“Have done X the previous day” is generally better than “do X today” when possible

For things that are like “avoid doing X” (such as limiting screen time), this is much better, because you can’t be certain you have avoided it until the very end of the day.

Even for “do X” things I also find this to be better, it makes you more likely to remember to do it automatically (which is not necessary but doesn’t hurt), and means you can verify all your tasks from the previous day in one go.

Try to avoid deadlines right at the end of the day

At least sometimes you will put things off until right before the deadline, in which case you will be glad not to have set it at midnight. I tend to set them at 1pm by default.

“Roll up” forfeits from daily to weekly once you have established a habit

This saves a bit of time in verifying once you know you won’t forget. In general it’s good to try and constantly prune and aggregate the things you are committing to so as not to end up with a bunch of stuff that you were going to do anyway.

1. ^{^}

    10 months total, and 4 months as a power user doing >5 forfeits per day

2. ^{^}

    See Ludic fallacy

3. ^{^}

    I don’t want to bloat this part of the post with examples but another thing I think is particularly bad in this way is Anki. If you miss a couple of days the cards rack up and make it much more aversive to pick it back up again, and the longer you have been using it the worse it is when you miss a day because there is a deeper stack of cards queued up that become due on subsequent days. Also the algorithm doesn’t handle missing days at all well (because it sets the new interval based on the actual time between reviews), which means essentially your whole deck gets bricked if you leave it too long because everything gets scheduled way into the future. It’s almost like it’s deliberately designed to make you go through cycles of using it and then rage quitting.

4. ^{^}

    Another step in this argument is that the first few percentage points are easier to achieve than the last few, getting from 99.9% to 100% is very hard

5. ^{^}

    It has a crippling design flaw where there is no penalty for missing your daily habits if you don't open the app

6. ^{^}

    I just use meme generator to cover the content of the message

7. ^{^}

    This mainly counts time in vscode or on localhost, which excludes quite a lot of things, e.g. looking things up. But I have found that even over the course of one day this number correlates strongly with basically how much proper work I did, and 4 hours is roughly “an honest day’s work” (and corresponds to about 5/6 hours of coding time)