Proposal to parcel out goals stingily

Get it? Stingily, but also sting-ily. Cuz you get another goal when you get stung.

This feels like a nice step towards more pledge focus.

Let’s try some help-docs-driven development…

:wavy_dash: [begin wavy imagination lines] :wavy_dash:

How many goals can I create?

Initially 3! You can jump straight to unlimited goals with a premium plan or you can get an additional goal each time you derail a non-zero-pledged existing goal. Wait, what? Are we trying to bribe you to fail at your goals? Not at all, I mean, well, it’s like this. We’re kind of bribing you to derail but derailing is not failing. It’s like how if you never make mistakes then you’re not learning. Pushing yourself is the point and that will mean occasional derailments and that’s fine. Creating a ton of goals off the bat is a common recipe for actual failure, the kind where you get overwhelmed and quit altogether. (Another form of actual failure is setting your goals so unambitiously that you never derail but also never do more than you would’ve without Beeminder, which is part of our derailing-is-not-failing argument. See for details.)

Anyway, so that’s partly why we parcel out additional goals so stingily. If you still want another one after derailing then you’re probably ready for it. Also it’s how we make money. Again, you can pay for unlimited goals directly with a premium plan.

Finally, whatever your goal limit, it only applies to active goals. So the other way to create another goal is to archive one of your existing ones.

Finally-finally, we’re also happy to bump up your goal limit in exchange for feedback. Get in touch!

:wavy_dash: [end wavy imagination lines] :wavy_dash:

What do you think?

PS: I made a new Twitter thread about “derailing is not failing”:


What if someone calls non legit, “I only detailed because I wanted another goal”?


I think it’s potentially difficult to navigate mixing derails and premium like that, unless it’s done more generally: I feel like the most “exquisitely fair” thing to do, if you gave more goals on a derail, would be to automatically turn on Premium for an account that’s payed the cost of premium in derailments for a month already… but I’m not sure how to generalize that. I know I’ll always want infinite goals, so I’m always paying $8/mo (well, actually, a lot less, since I pay once every two years, but for the sake of argument…)… does that mean my first $8/mo of derails are “pre-paid”? Does that dilute the sting too much?

I’m not sure if that is the most exquisitely fair thing to do, but I think it feels _un_fair to tell a hypothetically highly derailing user “no, sorry, you can’t make an eighth goal—you’ve only paid us triple the cost of premium, which gives you infinite goals”…

But I’d like to think there’s some elegant way to put sting-charges and premium payments in the same bucket!


Maybe a middle ground of “three goals, hard limit, until the first time you derail, at which point you have unlimited goals”, and you now have $derail_amount of premium credit if you want other premium features. This obviously reduces total revenue by $average_first_derail_amount_from_people_who_would_get_premium_anyways but it might have enough positive knock-on effect to be worth it? I’m not sure.

(or make “has derailed ever” the only way to get unlimited goals and take it out of premium plans, which is maybe the most “pledge-focused” approach? otoh, I only have infinibee for the unlimited goals, and I imagine that’s common, so that might not be a good revenue-idea right now)


Ha! That would take some cheek! :joy: But simple patch: only legit derailments yield free goals. Standard reply (not that I expect it to come up!): “Ok, do you want the derailment and the free goal or do you want neither?”

Ah, yeah, we’ve talked about that before and there was a lot of pushback – sting dilution is dangerous, as you say. This proposal also suffers from sting dilution but maybe there’s a line we’re not crossing by just unlocking additional goals? I’m not sure.

Figuring out how to do this has led to so much hand-wringing in the past that it’s a big part of what pushed us toward pledge focus where we just want to steadily put more emphasis on pledges and maybe gradually phase out premium, perhaps with some principled exceptions.

Not crazy! I guess this proposal of getting another goal each time you derail is a baby step in that direction.

Also not crazy!


I don’t have a horse in this race, so feel free to ignore, especially since I’m “predicting” the future. :smiley:

This feels like the kind of thing that is definitely clever and matches Beeminder’s philosophy but that you’d later come to think was not a good idea in practice and would want to back out of. For example, you’d find some users confused by it, and some users who’d abuse it (as already suggested), and it’s extra complexity (in code, quals, documentation, user support Q&A, and the legit check process).

I think the user confusion is the main part that’s giving me concern. The newest users would be seeing docs about this feature when they went looking for information about limits on goal numbers, and for some (many?) of those users everything will be at least a little bit confusing. When they’re reading about how this extra goal thing works, they have to also be understanding concepts like premium plans, pledges, “non-zero-pledged goals”, derailing, “derailing is not failing”, goal life cycles… All those concepts are important, but a new user would have to hold them all in their head simultaneously when all they want to do is understand how many goals they can have. I feel like this would be an “OMG WHAT NO, I give up” moment for some.


How would you compare this proposal to allowing one new goal per (say) week?

I think this version appeals to me because derailing several goals at once seems like a worse failure mode for new users than never derailing?


I do like some things about this proposal, and I definitely agree that something or another along these lines is needed. But I do have a bit of a concern about this particular setup. Specifically, I think that giving extrinsic to the goal in question when derailing can lead to awkward (even perverse) incentives in some situations.

For example: imagine I’m about to derail today on one of my goals—for concreteness, let’s invent a goal: let’s say a goal with a $10 pledge to do the dishes. I’m sitting on the couch, not really feeling like I want to get up and do the dishes, but Beeminder makes the choice concrete: $10 or wash dishes, my choice which, no hard feelings either way (paying is not punishment, after all.) That’s really good: a stark binary decision where I decide what I currently value more.

But imagine that I also get a new Beeminder goal if I derail. Maybe I’ve been musing about starting to learn the guitar, but I’m all full up on Beeminder goals right now, so I haven’t actually made the commitment and started on it. Now, sitting on the couch and wondering if I have the energy to stand up and go wash dishes, things are a bit more complex: I can either: a) pay $10, not wash the dishes, and make a guitar goal tomorrow; or b) wash the dishes and either kick the can down the road a bit on studying the guitar, or try studying the guitar without a Beeminder goal to keep myself on track.

This isn’t quite the cut and dry decision the original was. It’s entirely possible that I’d prefer get up and wash the dishes and not pay $10, but the guitar situation pushes me over the edge and I decide to leave the dishes in the sink.

I’ve just described a situation with an actual perverse incentive, where Beeminder would push me to not do the thing I wanted to do—to choose the worse of two options. As things stand with Beeminder now, when confronted with such a situation I can shrug my shoulders and say that whether or not I do the dishes reflects my revealed preference about whether I value not doing the dishes at over $10 or not. With this proposal, who knows? You could modify the preceding statement to be “… at over ($10 + the value of an additional goal) or not”, but that is a lot messier. Dollar values are clean and clear: I know exactly what $10 is. A major part of the value of Beeminder is the ability to put numbers on things, to figure out how much I actually value doing the dishes or whatever, as a scalar number. As soon as that becomes a vector instead of a scalar ((-\$10, +1 goal) instead of -\$10) it loses all that.


This is a very good point.


I agree with Nicky; that’s a really good point. We’ve been good about avoiding creating perverse incentives or mucking up the clean, bright lines but this does fog things up at least as much as some of the other things we’ve avoided. I hadn’t thought about it from the angle of the “Do I act now and do my goal’s task or not?” perspective at all.


Repeating from Beemail: now I’m wondering about parceling them out on some other metric like time or number of data points added. Or how many goals you have at $5 or more, so even being willing to be charged (rather than starting at $0) works… I actually kind of like that idea: no perverse incentive, because you can just start a goal at $5 if you want more goals sooner without having to derail, but it nudges people into real commitments.

For precedent on communications stuff, Postcrossing let people send postcards in a similar parceling-them-out way which we can potentially look to: How many postcards can I send?


I got some brilliant feedback about this from beemailing about it. First, here’s what I said in the beemail:

Most of you like [the proposal to parcel out more goals each time you derail] but @zzq brought up an astute concern. Namely, if you get more free goals when you derail then that’s a bit of sting dilution. Instead of a nice clear dollar value, you get your penalty bundled with a nebulous perk. That might be an unfortunate compromise of our behavioral-economic principles.

Or maybe it’s only an issue for newbees and they’ll have all the goals they want soon enough so it’s an elegant way to gently inculcate people. Don’t go too crazy with too many goals too quickly, derailing is not failing, etc. [1]

@alys, interestingly, said that the proposal is philosophically kosher but she predicts newbee confusion problems. I think that those can be overcome with an “achievement unlocked” framing (plus Nicky’s brilliant help docs exposition). I’d probably want to convey it as a kind of paternalism, which I think we need more of in general. [2] “You only get 3 goals for now but steady ho,” says the avuncular voice, “you’ll get more after you’ve more fully seen what you’ve gotten yourself into with this Beeminder thing”.

It’s all a bit speculative, obviously, and I’m eager to hear your opinions. STRAW POLL: Would this matter to you personally, either because you don’t have the Infinibee plan and would like more goals, or because you don’t want the sting dilution? If not, what’s your prediction about whether this would help or hinder newbees getting up to speed?

[1] Related reading: and

[2] I know some of you will bristle at that. I’m confident there’s a way to do it right, that won’t encumber the beeminding of you veterans. But do holler if you think I’m wrong or wrong to frame it that way. It could make for a good blog post, perhaps a generalization of (which I’ve found is still controversial in some circles but is really standing the test of time, I feel!).

And here’s some recap of the replies so far:

  1. Finding a balance between paternalism and mechanisticness is tricky. (Also, my own commentary now: paternalism may collide with anti-magic. My inclination is to take anti-magic as the harder constraint but aim for both. Eg, if there’s only one way to get more goals, that’s higher paternalism and lower magic/complexity. But it’s like we say in the anti-magic post: it’s just a heuristic / guiding principle. Pragmatism may still carry the day.)
  2. Can this be just about onboarding? Maybe at some point you hit infinite goals and don’t have to think about this again.
  3. More support for the alternative proposal of unlocking more goals via another metric: number of datapoints, amount pledged, consistency of adding datapoints, amount of edge-skating, tenure (like 1 goal unlocked per week of beeminding), …
  4. Derailing-is-not-failing is a pretty advanced, counterintuitive concept. Maybe newbees aren’t ready for it?
  5. Achievement-unlocked framing is great.

Where I’m at so far:

I think I agree that tying new goals to derailments is risky. But risk is good. There’s a chance it could work extremely well, pushing people to adopt the derailing-is-not-failing mindset. Conveying that derailing is not in fact failing feels crucial to me and I don’t think most newbees are hearing it. But I too am nervous about sting dilution. Can we find the best of both worlds? Or at least a good compromise? Like if you hit unlimited goals soon enough then the sting dilution would only be temporary.

Another thought that may not be realistic: make it a black box. We experiment with the best metrics for unlocking new goals and just tell users to beemind their guts out and The Algorithm will bestow more goals on them according to Its Infinite Wisdom. Ok, it sounds pretty dumb when I put it that way but we’re still brainstorming here so what the heck.


Folks do best when they skate the edge and take their lumps.

But this isn’t about diluting the monetary sting, and indeed it’s hard to compare access to creating one more goal with $$$ , so also hard to argue that dilution occurs. Even more so when the initial derailments are at $0. It’s nothing like comparing with premium credit which can more easily be expressed in monetary terms.

For a newbee there is literally nothing monetary to dilute other than a dollar sign on the graph and the inexorable Damoclean pledge value increasing to non zero.

Psychological dilution, perhaps — a desirable thing if we believe that edge skating is sane and value creating behaviour. We seek to normalise derailment as rational in pursuit of a worthwhile objective.


But what is really the goal here? To give users more goals (so they derail more?) - at least that’s what I understand from starting out with a link to the pledge focus blog post.

Your proposal sounds fancy and enveloped in all kinds of reasonings, but it would be a lot simpler to simply allow users to buy additional goals for one-off payments. Of course, that takes away all the complexity and fancy reasoning, but it gets you to the goal of users having more goals. Personally, I would jump at the chance of buying more goals than the few ones I have now. I am never going to pay the pretty big monthly price (big compared to other things I can get for similar amounts, not in terms of cups of coffee at Starbucks) but would definitely use Beeminder more (and probably derail more often) if I could create more goals.


Good question! I’m glad you mentioned this wasn’t clear to you, because it is worth making that really clear. I originally proposed something like this (I think it may have been forgotten it was my suggestion, but it was :sweat_smile:) because if we are pledge-focused (which we are), then features people are looking for (like being able to add more than three goals) should ideally not be hidden behind a premium payment.

At the moment, the way to get more free goals is to give us feedback (which is pretty much unlimited, though there is some degree of diminishing returns in that we don’t like giving out 3-4 new goals a pop no matter how much feedback you give us), or to subscribe to premium.

The good side of more goals being gated this way is that typically a new user cannot go ahead and make 10 goals, derail them all within a week, and then get upset because they feel that they didn’t understand what they were signing up for. The three goals you get by default are enough to play with and get used to the system, typically without becoming overwhelming.

So, my original proposal was intended to be a way to let free users have more goals without the problem of them being able to create tons of goals at first. So allowing users to buy additional goals for one-off payments would not answer that brief at all, and that’s why it’s not an option we’re re-considering (having done exactly that for a period before I joined the team; I don’t know why it was stopped).


Oh yeah, it totally was!

My latest thinking (thanks especially to @clivemeister who I chatted more about this with) is that the black box idea can actually work. We could say something like this:

You get additional free goals based on criteria including how often you’ve pushed yourself enough to have had at least one derailment (that’s a good thing!), how many datapoints you’ve added, how long you’ve been beeminding, etc. We experiment with these criteria to try to find the right balance between beeminding All The Things and minimal risk of burnout. You can always jump straight to unlimited goals with a premium plan if we’re being too conservative in parceling them out for you.


Parceling out goals sting-ily is now the law of the land! Here’s the announcement on the blog:

And here’s what you currently see in the UI if you hit that limit trying to create a new goal:

You're using all of your 3 active goals!

You can earn more goals by hitting Beeminder-y milestones. In particular, derailing and getting stung :moneybag: which, as we’ve argued extensively on the blog, means you’re using Beeminder as intended — to push yourself!

Or you can make room for more goals in your dashboard by archiving some existing goals.

(Unlimited goals are a premium perk if you prefer to just pay for them directly :carrot:)

Or have you not had enough time to experiment with Beeminder?

Let us know your special circumstances, or give us some feedback about how we can improve Beeminder, and we’ll give you another goal.


(Here in the forum that’s not a real button but check out the hover text on it.)

As we discuss in the blog post, we still want to expand this to bribe you to do other valuable things on Beeminder besides just derailing. Eager to hear more of your thoughts!


i love the mouseover text, @dreev! i also didn’t realise/forgot that there was a limit to free-goals-thorugh-feedback! glad you pass them out stingily now ;) (saying this as someone who’ll probably eventually get a subscription, but hasn’t yet.)

to briefly dig up an old point made here, because it caught my eye reading through the history of this thread:

unfortunately, that’s not always clean and clear for me, because currency conversions.
the difference of what $10 is in € has always been within a 40-cents-range for me so far, ever since i’ve been here / started derailing. sure, that’s not a whole lot. but it’s not nothing. and it’s not out of this world to imagine that one week, $10 might cost me 9€, and another month, $10 might cost me 11€. (maybe a similarly big difference has already happened with other currencies; i know too little about these things.)

i guess my point is that i am less susceptible to sting dilution arguments like this one, because the exact extent of the sting is already a bit unpredictable for me, in a very, very minor way (so far)?

i’m very excited about free goals being passed out stingily now! funnily enough, “unlimited goals” has never been my one and only reason for considering a paid plan. (i need to whittle down my gmail-e-mails eventually, so i need more than just “do more” goals. also, auto-ratcheting sounds like it would be useful to me the more goals i have! [i imagine i can only manually ratchet So Many goals each morning until it gets tiring.])

i wrote a typically long reply to the last weekly beemail, “parceling out goals sting-ily is now the law of the land,” but i guess most points could be interesting to discuss here in public as well:

  • quote from the beemail: “If so, that could be our excuse to introduce these other milestones and only give out free goals for a subset of derails, like the first one at each new pledge level.”—two thoughts on this:
    • i’d much prefer you to get really fancy with it, like, creating a whole complex algorithm that gives more goals for derailments only if the derailments are within a certain window. only derailed on this goal once in a whole year? no free goal, because the difficulty is clearly too easy! derailing on this goal multiple times a week / derailing right after the safety buffer of your last derailment ran out? no free goal here either, because this goal is clearly not working well for you (yet) / the commitment is too difficult!
    • personally, i find the idea of only getting one free goal for each new pledge level very limiting. not a lot of people (i imagine) set their pledge cap at $7,290, or even $2,430. but even if, that would mean only 8 possible additional free goals, max.
      i would pinstead prefer something like limiting the additional free goals to one per goal. (so derailing once on goal A gives you +1 free goal; derailing once on goal A and twice on goal B gives you +2 free goals, one per goal you derailed at least once on.)
  • general thought, that’s closely tied with the above point as well: i think the total possible number of additional free goals should be (very) high, if not infinite in theory! (how about +1 free goal for the 5 millionth data point you enter? +1 free goal for the 50 millionth data point, etc. etc.) i strongly believe that the average bee should have all their goal-needs met and then some, only by using beeminder in its intended way! (through derailing, adding data points, etc.)
  • looking at the other things mentioned in the beemail that could, potentially, in the future lead to additional free goals, i would suggest that most of them should also work retroactively—especially the ones you couldn’t otherwise achieve (again). “you get a free goal for setting up your first goal” can only happen once. people for whom this has already happened shouldn’t fall short of that additional free goal, imo! “your first $5 paid, you get another goal” could instead be rephrased to “your first/next $5 paid, you get another goal,” etc.
  • if you do go with the “+1 free goal for Xth data point entered,” i would suggest making this an unlimited (in theory) way to get free goals: +1 free goal for the first added data point; +1 free goal for the 50th; +1 free goal for the 100th; +1 free goal for the 500th; +1 free goal for the 5 000th; +1 free goal for the 50 000th data point added, etc. etc.—personally, i just don’t think that there should be a limit on this! i think adding more and more (and more!) data points should be very much encouraged, always! would i ever reach 5 million beeminder data points over the course of my life? who knows! but i would probably feel more motivated knowing that reaching 5 million data points would get me an additional free goal, rather than knowing every data point beyond the 500th (or whichever number) does nothing to increase my free goals. if a limit is set here, it shouldn’t be arbitrary. only counting the very first data point i would have no/less issue with, but passing out free goals for a larger number of data points and then suddenly stopping would irk me.
  • additional ideas i thought of, that could maybe lead to free goals:
    • +1 free goal for “the first/next time you add data on a beemergency day to get yourself out of the red”—should be very much encouraged!
    • +1 free goal for “the first/next time you add data on a goal that’s already outside the akrasia horizon,” or, alternatively (additionally?), “… data … that brings the goal outside the akrasia horizon”—should also be very much encouraged, i think!
    • +1 free goal for “the first/next time you’ve archived a goal”—should probably also be encouraged. don’t be afraid to bin something that’s not working / to set goals with an end date/final achievement in mind!
    • +1 free goal for “the first/next time you’ve consistently beeminded a goal for 1 year,” maybe. not sure what “consistently beemind” would mean exactly, probably having at least a certain number of data points per week, on average.

looking forward to thoughts if anyone has some to share!