Beeminder Forum

Protocol question: When someone tells you to nevermind


#1

I was supposed to do something for someone and, when I followed up later to get details so I could follow through, they asked me to put a pin in it. Does that count as a completion, or should that commitment be deleted?


#2

Good question. First, deleting anything is super fraught and all the beta users need to understand exactly how this works (and then we need to figure out the UI to make this work for normal people who will just use this thing and not read things). I’ll add that to the FAQ – http://dreev.commits.to/add_deleting_to_stew_FAQ

But probably the most straightforward answer is to void the promise, making it count neither for nor against you, but leaving it accessible at the original URL for posterity.

These cases are tricky and one pro tip might be to phrase your commitments in terms of things wholly in your control.


#3

Another option is to treat the deadline as conditional. Like “I’ll do XYZ within a week of getting prereqs ABC” and then just keep snoozing the deadline.


#4

It is something that was wholly in my control, in some sense, but it can be a waste of time to do the thing if not needed anymore and an “Oh, yeah, don’t do that now” decision is made. (And, actually, in this case it would have been counterproductive because the situation had unexpectedly changed and blah blah.)

To capture all of the “I will” statements one makes and to be careful to make sure that you can do what you say you’ll do is one thing, but I’m wary of the idea of trying to pack all possible unexpected changes into statements about what you’re agreeing to do. I mean, there are lots of everyday examples of teams changing directions and being like, “Oh, right, this is a better idea. Okay, instead of you doing this, you should do that.” It strikes me as hard to plan that into the “I will” statement in a way that’s not… super weird. [Edit: …in terms of linguistic conventions, I mean]


#5

PS - Void works, though! I went back and voided it instead of leaving it as complete.


#6

Yeah, this is philosophically super interesting. Can we reverse-engineer the right answer to this from the scoring function? That function maps every I-will statement to a number describing how true it was. So the possible answers from that perspective are:

  1. Void it because it turned out not to be an I-will statement, the way we conceive of such.
  2. Count it complete because you stayed on top of it; or that it’s vacuously true that you did everything you ought to have done.
  3. Let it count against you because the literal I-will statement was a falsehood.

For #3 it would serve as an incentive to be extra careful about what you say you’ll do! And if it’s understood that that’s the norm then it just means that lower overall scores are still very good.


#7

My first reaction was to go with (1) but now that I’ve thought about it a bit more I’m leaning towards (2). Part of the reason is in thinking about what happens in a scenario where you committed to do something by a certain date, and it is now one week late, and then you find out that the action you committed to is no longer necessary. Your reliability score should still suffer, because you in fact did not do the thing by when you said you would; the fact that it later turned out to be unnecessary doesn’t nullify that. So going with (1), i.e. voiding, wouldn’t really be the right thing.

I think an argument can also be made that if you make a commitment but then it turns out to be unnecessary before the due date, that should actually count as you 100% fulfilling your commitment. In other words I’m imagining that every commitment implicitly has an extra clause like “…unless the entity I committed to no longer wants me to do it” — which, I would argue, is how commitments actually work in the real world.

However, I could also see the argument that although being late on a commitment should always lower your reliability (regardless of whether it is later deemed unnecessary), it shouldn’t increase your reliability when a commitment is nullified prior to the due date when you haven’t actually done it yet. (If you promise to do a thing, but the person you promised later tells you to forget about it, do others see you as more reliable than before? Probably not… though maybe a little bit, actually, since human psychology is weird?)

So anyway, perhaps another possible solution is to simply tweak the semantics of “void” a bit: it does decrease your reliability score if you void a commitment after the deadline (by the same amount as if you had completed it then — this means you’d have to record the time that a commitment was voided), but it doesn’t change your score to void something before it is due. And what if you want to void something and have it not count at all, even after the deadline? I would argue: you shouldn’t/can’t. (In practice of course, one could snooze the deadline into the future and then void it…)

Oh, and also, I’m pretty sure that (3) is terrible.


#8

Good points. This happened to me on the weekend however I almost forgot to go and void it until Monday (I did remember just in time), however, could we backdate to the point at which I found out it was void?


#9

Wait, actually, this is weird. What if your reliability score is currently 80% and you then void a commitment one week after it is due. Should that increase your overall reliability?