Beemium feature request: User-specified road widths

I want to be able to use a Beeminder contract for goals where:

  • The interim measurements are substantially more spiky than the current
    road-width algorithm can handle.
  • The initial time to build up a buffer zone is unreasonably long
  • Staying on the road is part of the fun

For example, consider a weight-loss contract, with these properties:

  1. The rate is half a pound per week
  2. The contract initiates with the current weight on the middle line
  3. Accommodates that upon occasion, but less than often than one day in
    ten, the contract-bound party anticipates
    1. drinking two pints of lager and two pints of water for a weight
      gain of four pounds. When salty food is part of the meal, the weight gain
      will persist for more than 24 hours, and then plummet.
    2. eating meals with substantially more fiber than usual yet well
      within the daily calorie allotment, increasing scale weight by three or
      more pounds while they work through the system.

Given the above, I would be willing to put money at risk with a
five-pound-per-side road width. Any narrower would, in my opinion, be a
contract I would be certain to lose.

If I want to reduce my weight by five pounds in ten weeks, I want to be
able to put in the goal weight and the goal date when I initiate the
contract. Starting flat and staying flat for ten weeks - the entire
duration of the goal - just to create a five pound buffer, seems
unreasonable.

Hey P.U. (or presumably – per recent conversation on twitter.com/bmndr – Thomas),

My first response is that we have a road-width override implemented but not exposed (for reasons I’ll explain below) so we’re game for trying this as an experiment. We’d expose a parameter called abslnw (absolute lane width) for custom goals, which means it would be available with the Bee Lite plan.

But first, let me try to change your mind! :slight_smile:

I’ll start with the First Rule of Beeminder (so named by Jill Renaud, I believe):
Things that make staying on the road easier make reaching your final goal harder. There’s no free lunch. Any leniency today will get paid for down the – wait for it – road.

That applies to myriad things we’ve tried in the past like grace periods and 3-strikes rules. I believe it also applies to wide roads. We make this case in blog.beeminder.com/roadwidth (start reading at “we’ve learned this the hard way” and note that your proposal means staying below a fixed bright line, namely, the top edge of your fixed-width road).

Of course none of that addresses your point about pints of lager and salty food. For that we need to describe how to game Beeminder’s current algorithm…

The “can’t lose tomorrow” guarantee is that if you’re in the right lane (blue) today and you’re weighing daily then the road will auto-accommodate any fluctuation tomorrow and you’ll still be on the road. The emergency backup strategy is that if you’re off the road (red) today you still have till the end of that day (till midnight) to get back on the road. If you end the day still in the red then you derail (though there’s a loophole even here, in that the graph doesn’t automatically refresh till 3am – but you’re playing with fire if you try to exploit that loophole :slight_smile: [1]).

Also there’s the fact that you start with a week of flat spot, which I know seems unsatisfying, but just think of it as a way to have more wiggle room – the reason you want a wider road. In other words, start losing weight immediately despite that flat week and if you do so perfectly, you’ll stay parallel to the road with 7 days’ safety buffer. Try to maintain that buffer and you’ll always have plenty of room for temporary fluctuations, just like with a wider road.

In conclusion, we think that, if you’re willing to play Beeminder’s game [2] the current algorithm will allow for your lagers and salt case, if you’re careful. Your proposal with a wider road is more lenient but, again, confer the First Rule of Beeminder.

Here’s how it might play out in the current Beeminder:

If you’re in the wrong lane you don’t have the “can’t lose tomorrow” guarantee so no lagers and salty food. If you’re in the right lane and you weighed in yesterday then you do have the guarantee, so go for it. (Ideally you’d wait till you’re in the green, but let’s be reasonable.) You’ll weigh in 4 pounds heavier the next day but you’ll still be on the road because it will auto-widen. Now you’re skating the edge but as long as that 4 pounds comes back off at the road’s rate, which it certainly should, you’ll be ok. If it was really a temporary gain then you’ll be back in the right lane in a couple days and are free to repeat the whole game.

Eager to hear if that was persuasive!
Danny

[1] Why I say “playing with fire”: if you were in the red and you weigh in after midnight, that will cause your graph to refresh for the new day. Which ought to derail you since you never made it onto the road the previous day. But if the new datapoint has you back on the road then Beeminder lets bygones be bygones – slightly violating the no catch-up principle! – and says you’re ok. So if you’re in the red you can get away with weighing in by 3am instead of midnight but only if that new datapoint has you immediately back on the road for the new day (not just yesterday’s emergency day). (And if this footnote doesn’t make sense, don’t worry! Just treat midnight as the real deadline and you’ll be fine…) (Also, we reserve the right to close that 3am loophole any time! Though I guess we’d at least say so here first.)

[2] Despite calling it “gaming” we think it jibes pretty well with losing weight in a healthy way – the best you can do for akratics like us.

My first response is that we have a road-width override implemented
but not exposed (for reasons I’ll explain below) so we’re game for
trying this as an experiment.

I have participated in a user-specified road-width experiment in the past.
It was successful from my point of view, but a failure from the Beeminder
point of view: I achieved my goal, but Beeminder earned no revenue. When I
saw the notice for the Beeminder Premium plans, I dared to hope that *other
poweruser settings *included user-specified road widths. “Hooray!” I
cheered, “I will be able to craft a contract to which I am willing to
commit, and pay Beeminder for the service!!” But alas, no.

Things that make staying on the road easier make reaching your final

goal harder. There’s no free lunch. Any leniency today will get paid
for down the – wait for it – road.

I agree with this, and I do not look to a contract-enforcement agent for *
leniency*. I look to contract-enforcement agent for merciless enforcement
of agreements.

Here is my reasoning:

  • If Ulysses breaks free of the ropes because the rope-tiers show
    leniency, his doom is deserved.

Do you agree with that?

  • Ulysses should specify bindings which he believes will be sufficient
    to restrain him. If Ulysses is confidant that the proposed restraints will
    not bind him, he should not agree to the contract.

Still with me?

  • Ulysses wants to hear the sirens, but not become their prey. If
    Ulysses was simply concerned with not becoming prey, he could stuff his own
    ears and lock himself below decks; but Ulysses is a clever man, and he
    wants, by means of his wits, to experience the calls and live to tell the
    tale. Ulysses needs to think carefully about what bindings will allow him
    to hear the sirens but not escape, and then contract with merciless agents
    to enforce exactly those bindings.

That applies to myriad things we’ve tried in the past like grace

periods and 3-strikes rules. I believe it also applies to wide roads.

Grace periods and 3-strikes rules are examples of leniency, surely; but
what makes wide a fair adjective to apply to a road? I attempted a
Beeminder contract for weight maintenance, and after two days the algorithm
was displaying a road width of 3 ounces. Drinking a glass of water with a
vitamin pill was enough to put me into Lose In One Day state, with bold red
alerts in the UI and klaxons in my email. Every road has a width, but a
road is only wide if the width exceeds the expected variation.

Also there’s the fact that you start with a week of flat spot, which I

know seems unsatisfying, but just think of it as a way to have more
wiggle room – the reason you want a wider road.

First, a flat start does not address Beeminder goals for weight
maintenance. For weight maintenance, the entire term of the contract will
be flat. Second, for a weight reduction goal where the plan is 8 ounces per
week, it requires a flat start of six weeks to achieve the minimum buffer I
would consider prudent for a contract.

Try to maintain that buffer and you’ll always have plenty of
room for temporary fluctuations, just like with a wider road.

My weight can vary three pounds or more in a single day, so I would not
describe an eight-ounce buffer as plenty of room.

I have tried a number of strategies with Beeminder, including simple
reliance on the auto-widening road; setting the goal at five pounds higher
than my real goal; forcing the road to widen by adding ten pounds to my
weight-recording every ten days; and specifying the road width at contract
initiation. Only the last one worked. Maybe I am especially sensitive to
goal credibility. “If I steadily progress at -1.125 ounces per day for 20
days, the road will get narrower; if I then catch a headcold, take an
over-the-counter remedy for symptomatic relief and force fluids, 48 hours
later I can expect to forfeit, even though I am running a calorie deficit
during those 48 hours.” The magnitudes are grossly disproportionate. Two
pints of Hale and Hearty chicken soup and a two pints of water is 320
calories and weighs two pounds. Two pounds of fat is 14,400 calories. The
ratio is 45 to 1. The algorithm will misfire.

I want two things. I want to succeed in achieving my goals, and I want to
compensate Beeminder justly for acting as contracting counterparty and data
recorder. I have other goals besides weight control which do not have any
exposure to preposterous road width, and I will happily use Beeminder for
those goals.

Edit:
Two pints of Hale and Hearty chicken soup and a two pints of water is 320
calories and weigh four pounds. Four pounds of fat is 14,400 calories.
The ratio is 45 to 1. The algorithm will misfire.

PU has some good points about weight fluctuations and I have to agree that my experiments with beeminding weight loss have not been that successful for some of the same reasons. My weight can also fluctuate 3-4 lbs depending on a number of factors, both healthy and unhealthy.

I agree with Daniel that I don’t think widening the road is exactly the right answer, but there does need to be a larger tolerance applied somehow.

I have been using the Red Blue Green indicators a lot since they were added. I use them as a meta-goal to motivate me. I try to keep checkmarks in all of them. If I have a setback and start slipping on my goals I start eating into that safety buffer. When I get in the red I have to fight back tooth and nail to get back on track into the green and blue.

Right now those indicators represent 1,2 and 3 day requirements. What if they were adjustable? PU could set his blue indicator at - 5 lbs. As long as that is checked he knows that he has wiggle room for a Beer and Salty food night. But when he weighs in the next morning 4 lbs heavier than normal he is going to be in the red.

I know that this isn’t really different than the current way safety buffer works. What it is, is a more visual way or representing your pre-determined safety buffer which I think is what PU is asking for.

Jake

I’m still mulling all this but I wanted to make sure people aren’t missing the point that no amount of fluctuation is too big to for the current algorithm to handle (no matter how disconcertingly thin the road appears) as long as the fluctuation happens from one day to the next.

[UPDATE 2020: Everything in this thread is super wrong. See Death to Auto-Widening Yellow Brick Roads, Part 2 for why!]

Let me clarify what my issue is rather than try to solve it.

  • Trying to lose weight using Beeminder
  • Very quickly I was very close to the dark line
  • Now I have a “heavy” day for whatever reason and weigh in over the line. PU described two realistic scenarios
  • Now I get all the warning signals, red, Alerts, etc.
  • If I don’t get my weight below the minimum in 24 hours I default
  • Here is the problem:
    It might take me more than 24 hours to get back on track. If it was beer and salt, maybe 3 days of normal eating. If I was full of fluids, 1-2 days. Since I don’t have time for that I am forced to resort to instant weight loss techniques: diuretics, sweating, avoiding heavy foods, dehydration, etc or fail my Beeminder.

I think the healthy option is to get back on the horse and follow whatever plan you were on before the heavy day. If you have more good days than bad days you should trend down over time.

I know that you are going to say, build up more safety buffer. That is really hard and if I was really good at that then I probably wouldn’t be a Beeminder user.

I think there are 2 choices. Set a shallower road to begin with and/or set it to become shallower in a week so you only have be aggressive for a week before it becomes easier to stay on track. Or actually be a little aggressive with weight loss efforts to counter the beer/salt day rather than just go back to normal mode the day after.

If you’re wanting to force a certain weight loss rate and you want to be able to have a high risk intake day, then be prepared to counter it with a really low risk intake day (s) plus extra excercise. It’s that or a shallower road to allow for less effort overall.

You could try doing 3 long and/or intense workouts the same or following day as the beer/salt and see if that knocks out the problem.

When the setting-up-days-off-in-advance feature happens (currently most popular on uservoice), you could preset your road for a built in beer/salt day once a week or something, making 2 automatic flat days follow. Maybe you’re not that scheduled with when you go out for beer and salty food, but you could try to plan it ahead to help your weight loss goal.

Short of diuretics and feeling dehydrated, other natural methods of rapid weight loss for one or 2 days are ok, such as eating less than normal and doing extra workout time (while still drinking enough to not feel dehydrated) are safe for a day or 2 and can make the difference to get you back on track.

Melanie

Melanie Reeves Wicklow, HFS
Beeminder’s Resident Fitness Expert

I think setting the road to be shallower has the same problem to which Jake
was referring. Namely, that akratics will still end up nestled up against
that new, more shallow top of the road and then the fluctuation will still
push it over the edge. But I don’t think a custom road width will help with
help that since we’d just nuzzle up to that new width too. Of course, I’m
jumping in mid-stream here and so don’t have the background info about
what’s already been discussed. (Hi, btw, I’m new to posting to the group!)

I think the trick is something like training the akratic mind to see being
in the yellow zone as the disastrous “not to be done” rather than seeing
being in the red zone as the enemy to be avoided and the yellow as “on
track”. This is probably more true with the weight loss goal than with any
other and perhaps derailing in these cases is the consequence of not
learning that lesson. After enough derailments, or with a high enough
limit, the blue line will start to seem, more and more, like the only safe
line. And perhaps that’s as it should be. Learning to live with a buffer is
a great thing for us akratics to pick up, I think. (Maybe with the weight
goal, being in the orange zone should trigger the “emergency day” email,
rather than it being triggered by moving above the yellow road.
)

Of course, this all rests on an assumption that short-term weight gain is
mostly the result of one-off actions that are relatively easily balanced
out the next day. There are other times, though, when water retention and
other weight-affecting things that have nothing to do with calories/fat is
the result of a medical condition (such as being non-male) and can last for
several days and then drop below the original point without actually
missing a beat in terms of true background loss. In the case of some
medical conditions or of being a woman, you can easily have three to seven
days of “gain” that aren’t really capital-g “Gains” at all and so you’re
still going to derail, despite the fact that goal-wise, you’re actually on
track. (I’ve included an image attachment of such a type of case, but don’t
know if that works on this board.)

In this case, perhaps a little fine print that allows flexibility in
reporting is the answer. If someone knows (in advance) that there is a
non-%-of-body-fat-related reason that their weight fluctuates, maybe
they/we just write into our fine print that when one of those times occurs,
we’re allowed to postpone entering new data for up to a maximum of X days
(however long the situation usually takes to resolve itself) and then,
after those X days, you have to enter the data and pay the piper, so to
speak. (Of course, you lose the QS data in those cases, but it’s a
trade-off, for now.) I haven’t done that yet, despite knowing that one
derailment that I paid for was a legit derailment of this sort. But since I
was recently away from home and unable to weigh (and so unable to track)…
leaving a gaping hole in my data, I feel less upset about not having a
perfectly reported graph. That ship has sailed.

Or, perhaps the weight goal could be set up differently. But I suspect the
coding required would be complex and not worth the investment given the
other projects on the go. Something like: you get pick X number of days in
a custom setting and, when the road widens due to being over, the upper
level both widens and flattens for those X days, but then returns, steeply
to where it would have been. The middle of the road retains the same slope,
but the upper limit flattens and then drops sharply to meet where it would
have been had it, too, kept the same slope. There’s the opportunity to
abuse and then end up screwed and derailed… but if you know you’re
inclined to do that, set X to 1 or 2…

(Sorry so long… I appear to rarely be as concise as I intend!)

I have not been clear.

Let me try expressing my point another way.

  • I do not want a commitment contract which permits high risk intake
    days. Beeminder does not want to offer such contracts, and I would not be a
    customer for such a contract if Beeminder offered it.
  • I want a commitment contract where I agree to maintain either a steady
    calorie deficit (during weight loss) or a stead calorie balance (during
    weight maintenance).
  • A high risk intake day would be a day where calorie intake was high.
    A day where water intake is high is utterly without risk to the
    calorie-based goal.
  • Weight is a convenient to measure.
  • Weight change can be used as a proxy for calorie deficit or calorie
    balance, but it is a crude proxy.
    • Weight can increase during a period of calorie deficit
    • Weight can decrease during a period of calorie surplus
    • Weight can remain constant during a prolonged period of calorie
      deficit

There is evidence from personal experience:

In 2010, I agreed to a single-point commitment contract which required me
to weigh no more than a specified weight on a specified day ten weeks away,
a reduction of 7.5% of my body weight. At risk was a multi-thousand dollar
donation to a notably corrupt and devious political candidate. I tracked
calorie deficit scrupulously, but nonetheless had many worrisome moments
due to weight not following calories. I succeeded in my goal and the
candidate lost, so it was an unmitigated triumph.

Last autumn, my purpose was to reduce my weight 2.5%, so I set a goal of 5%
For weight loss, I pick a target weight below my desired weight, because
during a calorie deficit my glycogen stores are depleted, but will quickly
return. Beeminder setup a five-pound-per-side road contract for me. During
the time of the contract, there were multiple times when my weight
increased 3.5 pounds or more over the course of four days, each day higher
than the previous, and subsequently decreased. I succeeded in reaching my
goal.

As I mentioned earlier, I have also tried the standard Beeminder contracts,
with the auto-widening road. I failed, and for me they became
dismotivational. The alerts would fire on days where I had met my calorie
goals. My weight would return to the desired level within a day of
forfeiture, and I would shake my head in dismay at my own foolishness in
agreeing to the terms in the first place.

Based on the principles in the bullet points and on my own experience, I
would be reluctant to enter into a calorie-deficit or calorie-balance
commitment contract which uses body weight as a proxy unless I could
specify the road width.

On Saturday, April 13, 2013 1:30:17 PM UTC-4, Melanie Reeves Wicklow wrote:

If you’re wanting to force a certain weight loss rate and you want to be
able to have a high risk intake day, then be prepared to counter it with a
really low risk intake day (s) plus extra excercise. It’s that or a
shallower road to allow for less effort overall.

I think setting the road to be shallower has the same problem to which
Jake was referring. Namely, that akratics will still end up nestled up
against that new, more shallow top of the road and then the fluctuation
will still push it over the edge. But I don’t think a custom road width
will help with help that since we’d just nuzzle up to that new width too.

Hi Mary,

I am akratic, and I when I tried a glycogen-aware-road-width contact, I did
not end up nuzzled against the top of the road. Being in the wrong lane
felt bad and being in the right lane felt good.

I don’t end up nuzzled against the top of the road for any of my other
Beeminder goals, and it is the same akratic mind.

The reason I use Beeminder is so that my far-thinking mind can set the
goal, Beeminder can immanentize the goal, and my near-thinking mind can
make the desired daily choices.

The only thing I did not like about the glycogen-aware-road-width contract
was worrying how Beeminder could get paid, if what it offered was sensible
contracts to people who wanted to achieve their goals. That’s why I was
hoping that the Beeminder Premium would include road-width-setting as a
user-customization.

Gotcha.

(And perhaps my feeling that “akratics” end up nuzzled up against the top of the road is due to the fact that I haven’t been using Beeminder long enough to have stopped that behaviour yet.)

I’m with Essentiae, I feel like the things that I really need Beeminder for are things where I’ll end up skating the edge. If the warm feeling of being in the right lane were enough motivation then I could use any number of other QS tools [blog.beeminder.com/trackhack] or just an excel spreadsheet.

So I do think that a True Akratic, if given a 5-pound-wide road, will end up skating the edge of it and end up 5 pounds above their actual target. So I still prefer the solution of, if you need a guaranteed road width of 5 pounds, starting with a flat road and waiting till you’re 5 pounds below it before dialing your road to reach your target weight at the target time. You can even reduce the wait and initial derailment risk by making the road slope up for a bit first.

Picky Ulysses’s example of a road that was 3 ounces wide I think is a red herring. The road width is fluid while you’re below the centerline and becomes fixed when you cross it. [1] The key is to work hard to not cross the centerline, and if you do, work even harder to get back below it. We have really good reason to believe that if you play Beeminder’s game then even if you have wild fluctuations, you don’t need to wait more than that initial week with a flat road before hard-committing.

(Btw, to clear up one bit of confusion: Beeminder does warn you when you’re in the wrong lane. It says you “could lose tomorrow”, ie, you’ve lost the “can’t lose tomorrow” guarantee. That’s the warning Picky Ulysses didn’t like. But again, instead of widening the road, first get enough below it, and then I think the existing warning scheme will be just right.)

I’m thinking maybe Picky Ulysses is a fairly special case because I think the problem of “I’m akratic so I’ll end up skating the edge and also salty food [2] makes my weight jump up temporarily” is fundamental. Contrary to Picky Ulysses’s experience, a wider road doesn’t solve it, it just postpones it. Confer the First Rule of Beeminder. With a wide road it’s just a matter of time before you’re skating the top edge anyway. A single bright line that you’re up against and that any fluctuation can push you past. Beeminder’s auto-widening is just a (perhaps too elaborate) trick for having progressive bright lines. The centerline is the first bright line. The consequence of crossing it is forfeiting the “can’t lose tomorrow” guarantee and, more specifically, triggering the final bright line, the top edge of the road, which is fluid until you cross the centerline. The top edge of the road is the bright line that you have to be under every day by midnight or you derail. I don’t buy the complaint that that line is too close to the centerline (that the road isn’t wide enough). The top edge will be as far from the centerline as it needs to be based on your actual fluctuations. The only way that can fail is if you drift over the centerline and then have beer and salt for dinner. So don’t do that. If you’re in the wrong lane you’ve got to be super careful. That’s just how Beeminder works.

Said again in terms of the actual strategy you’d employ:

The initial flat week plus the fact that when you start a weight loss goal Beeminder shifts the road up to start at the max of your initial datapoints means that you can be sure to be a few pounds below the centerline at the end of week one. You’ll also be able to see from your trend line if you’re matching the road steepness. If not, either buckle down or dial the road easier. If so, you’re going to be following the road but mostly in the green. Perfect. One day you’ll have beer and salt for dinner and the next day your weight will shoot up by, let’s exaggerate and say 10 pounds. No problem, as long as you’re weighing daily. The road will auto-widen and at worst you’ll now be at the top edge (which is now a fixed bright line because you crossed the centerline). But no problem, if your weight made such an extreme upward jump it’s safe to say it won’t make another upward jump tomorrow. You’ll get back to the right lane in a day or two.

[1] I wonder if it would be less confusing/misleading if, whenever you were below the centerline, the road simply had zero thickness. Only when you cross the centerline does the width of the road manifest, and when it does so it’s fixed. Until you get back below the centerline when the road shrinks to nothing again.

[2] Or menstruating, to use another common example.

“(Btw, to clear up one bit of confusion: Beeminder does warn you when you’re in the wrong lane. It says you “could lose tomorrow”, ie, you’ve lost the “can’t lose tomorrow” guarantee. […])”

That’s true. I actually have a filter that shows me only the emergency ones in my email… I should undo that so I’ll see the “could lose” ones as well (and start thinking of the “could lose tomorrow” line as the “crisis” line, at least where that one goal is concerned, given potential fluctuations.

“The top edge will be as far from the centerline as it needs to be based on your actual fluctuations. The only way that can fail is if you drift over the centerline and then have beer and salt for dinner.”

Or if you’re the cozy-up to the top of the road type and have something in particular that causes you to to go up for a day or two in a row, through no actual fault (really, not just the BS kind). For the cases of which I’m thinking, though, I think that knowing that one has potentially wild fluctuations at times, using the “Don’t gain, maybe lose” (now possible manually with retroratchet or automatically with one of the premium levels) is a good middle of the road compromise since it’s something only a handful of people will really have to think about, I think.

For something more predictable, like footnote #2, getting the timing right for a +x for one week (where x is the likely fluctuation) and then a -(2z + x) the first week and -z for two more, where z is the actual desired weekly loss, would do the trick… but remembering the timing is, like the weekend thing, a bit of a pain. (In fact, I’m working this weekend without the earlier intention to do so because I can just never plan that right (and really dislike emailing for fixes or legit derails since I tend to want to save those for the times when I screw up my graphs!) but will derail otherwise!

On Saturday, April 13, 2013 3:56:37 PM UTC-4, Daniel Reeves wrote:

I’m with Essentiae, I feel like the things that I really need
Beeminder for are things where I’ll end up skating the edge. If the
warm feeling of being in the right lane were enough motivation then I
could use any number of other QS tools [Trackers vs Lifehackers | Beeminder Blog]
or just an excel spreadsheet.

Based on this standard, it follows that I do not really need Beeminder.
Being in the right and wrong lanes is enough motivation for me in all the
Beeminder contracts I have tried so far. The right/wrong lane for me is a
big part of Beeminder’s value.

So I do think that a True Akratic, if given a 5-pound-wide road, will
end up skating the edge of it and end up 5 pounds above their actual
target.

Based on the results of the glycogen-aware-road-width experiment, where I
did not skate the edge, it follows from this axiom that I am not a True
Akratic.

So I still prefer the solution of, if you need a guaranteed
road width of 5 pounds, starting with a flat road and waiting till
you’re 5 pounds below it before dialing your road to reach your target
weight at the target time. You can even reduce the wait and initial
derailment risk by making the road slope up for a bit first.

This is the bogus goal solution. If I want to get from a current weight
which fluctuates around 195 to a new weight which fluctuates around 190,
where the fluctuations are spiky, it seems I should either:

  • set as a goal “maintain weight,” and then commence my
    half-pound-per-week calorie deficit. After ten weeks, when I have lost the
    weight, I should change the goal to my then-current weight
  • set as a goal “gain five pounds,” then commence my half-pound-per-week
    calorie deficit, then after the road has sloped up to a five-pound buffer
    change the goal to five pounds above my then-current weight

Have I misunderstood?

Assuming I have not misunderstood: I do not want to commit to a bogus goal.
If I ask a friend or family member to monitor my contract progress, I do
not want to have to have a discussion starting with “Wait - I thought you
said you were trying to lose weight; this chart shows you trying to gain weight.
Which is it?”

To summarize:

The stated purpose of this Google Group is for akratics to discuss what
techniques they have used in the past in order to achieve their goals. One
technique which I have used to achieve a weight-loss goal is is a
rudimentary glycogen-aware commitment contract from Beeminder, which
required special setup. I have also used commitment contracts from
Beeminder for other non-weight-loss goals that did not require special
setup. However: In none of the cases did I find myself skating on the edge;
rather, the right-lane / wrong-lane and below-the-road feedback was
adequate. Since the right-lane / wrong-lane signals were adequate, it
follows that I am not a true akratic, and do not really need Beeminder.

I apologize for the misunderstanding.

I don’t really think anyone was meaning that you probably have no use for Beeminder, but rather that, for at least some of us, the skill to care at all about the centre of the road is something that is completely lacking. Perhaps the word “True” is best replaced with “Still kind of hopeless”. For others, the “less completely hopeless” akratics like you, there’s a slightly smaller motivational kick in the pants required (i.e. the fear of getting too close to a will-fail-soon position) that wouldn’t be there without a system and that provides enough juice to get you going faster and better than having nothing. The “if I could do that, life would be amazing and I wouldn’t need any help ever” thing that we in the “still hopeless” category might say is probably more a fantasy than anything else because, in reality, as willpower develops, although a smaller stick might be required (like that of a 2-day threat instead of a TODAY threat), a stick still helps until we become non-aktratic productivity and self-discipline ninjas. Maybe what we’re saying when we say “if I could do that…” is more like “but… but… I can’t (even) do that…”

BREAKING NEWS: custom road width is now live in advanced settings on custom goals! (Ie, available for Bee Lite subscribers.)

[UPDATE 2020: And then we killed it]

I do see what you mean about wanting the centerline to go from your current weight to your goal weight. I don’t think you lose too much by following a yellow brick road where “following it” means staying parallel to it a few pounds below it. But your point stands. That’s not ideal.

Still thinking about this stuff and hugely appreciate the conversation! Eager to hear if others repeat Picky Ulysses’s experiment with custom road widths.

Thanks again, everyone!

PS: Bethany proposes that picking a custom road width means that you have to always be on the road – no good side of the road. Could be an interesting deterrent to abusing the feature…

I’m going to add a little bit to this discussion as I’ve been using beeminder/kibotzer since April 2008 (holy cow 5 years now!) for my weight and it has been moderately successful.

First we did some crazy 3 strikes rule, which really did not work for me because it made it too easy for me to slack and I ultimately did not reach my goal. This lead to the idea of being on your road every day (which I think is much better) and if you are off of your road then you lose. I wound up doing a lot of baby sitting for Danny/Bethany that summer. (Which was okay because Faire is pretty adorable).

Kibotzer had for awhile that the first weigh in of the day was the official weigh in, but I think that lead to me doing crazy things like waiting until 1 or 2pm to weigh in without eating or drinking water… which was probably bad for my metabolism :slight_smile: The current rule of “lowest” weight of the day is probably fairer than the first weight of the day because then you have the day to do something productive towards weight loss. From April 2008-Aug 2010 I lost about 10 lbs that I’ve basically kept off - I think I haven’t been above 165 since I reached the point a few years ago - which means a sustained 5% loss over 5 years (yay beeminder).

From about Aug 2010-Jun 2011 I lost 18 lbs on the variable road width algorithm. I think that I succeeded in this goal but I developed some really perverse (and I think unhealthy) eating habits to keep my road wide. If I was on the “right” side of the road I’d completely 100% pig out with reckless abandon because I knew that my road would autowiden for me. Then I’d eat pretty normally for a few days until I got back to the right side of the road – and repeat. I think that this made me lose weight in a very strange manner and also sort of killed the weight loss mentality for me. In July 2011 I was down 33lbs from my beeminder starting weight.

I got married in August 2011, and gained a lot of “honeymoon” weight which I’m now trying to take off. I stopped weighing myself for awhile and ate a lot of crappy food while my husband was assigned to a case in London (while I was in NYC with the dog!) Beeminder rule #1 for weight loss – just keep weighing, daily. Even if you’re worried that you’ll be above the road – whatever – weigh, pay the fee and keep beeminding, you’ll wind up at a lower weight at the cost of a bit of money to Beeminder. I stopped weighing and just hoped that when my safe days ran out I’d be able to make it back on the road. I coughed up something like $500 to danny/bethany in June 2012 for that one – they still “owned” that contract as a relic from the kibotzer days.

Mostly I didn’t bother with tracking weight for awhile until Dec 2012 when my primary care physician yelled at me for being overweight and told me to lose some weight. Based on my pigging out behavior while on my previous contract (to widen the road) I chose that a fixed road width would be better for me. I’ve elected to have a +/- 3lb road width (fixed). It has helped me a lot with the crazy binge pattern that I’d developed. Now when I’m on the right side of the road I don’t worry about trying to widen it. I just work on keeping up my mojo. It seems to be working so far. I have the self imposed rule that if I’m home I have to weigh every day. If I’m on the wrong side of my road for 7 days I’ll force myself to reset/recenter.

TL;DR: I’ve lost about 20lbs with beeminder over the past 5 years (at one pt I was up to a 33lb loss). I fully support custom/user specified road widths. Weighing daily is key.

I don’t think you lose too much by following a yellow brick road
where “following it” means staying Parallel to it a few
pounds below it.

Here is what you lose: when you don’t adjust your calorie budget downward to account for the fact that the weight reduction accomplished has decreased your resting metabolic burn, you start drifting up into the Yellow Brick Road. When that happens, Beeminder sends “Great job! You are on the right side with an X-Day safety zone!” when what it should be sending is warnings.

That’s huge.

With a glycogen-, water-, fertility-cycle-, lipid-, and/or muscle-aware road width, there’s a much better likelihood that warnings and praises based on body weight will track the disciplined adherence to a calorie-deficit regimine.

Picky,

Are you also tracking things like calories, etc? I’m just guessing that you might be, 'cause it sounds like you’re pretty on top of your fitness data. I’ve been getting into the habit of tracking calories, satfat, & exercise and have just recently decided that having a major weight goal too is kind of overkill. I’m keeping my weight chart, but with a very small decline (-0.25), instead of my real goal, and a setting to have it just trim the excess safety days when I have over 10. That way, it works kind of like the “don’t gain, maybe lose” concept, but also keeps me having to keep my other fitness-related goals tuned well enough so that they’re at least somewhat effective. It allows enough flexibility for fluctuations over a short time (but headed upwards longer than a day), yet captures my losses and changes the road to match them. It doesn’t force me to keep my pace, but the behaviour-related goals do that, really.

Would that be anything that would work for you too (esp. if you have behaviour-related goals).

(Btw, guys, I love the idea of having the road trim our excess safety days. I haven’t seen it in action a lot yet, but that’s my new favourite feature.)