I guess I picked a biased forum to float this question! The first response probably exemplifies what a random jury of one’s peers would say. Alas, here we are!
Isaac’s “inspired halfassery” is cracking me up. But that sums it up nicely. Figure out a way to make it happen. It might end up being a little lame, but not as lame as if you weren’t bound by this yellow brick road at all!
Here’s what Bethany and I have been groping our way towards:
Be as lenient as possible but it has to be based on a generally applicable principle, not special treatment. For example, anything that could remotely be considered Beeminder’s fault should never cause you to derail. This includes missing features that the ideal Beeminder would have, like specific warnings first thing in the morning on emergency days, even if you normally have reminders come at night, or don’t use bot reminders at all.
In fact, we made an accommodation for Michael Merchant last month on that principle, when he went off a road because he didn’t realize it was an emergency day, when the bot told him so late that night:
What do you think of this principle: modify things retroactively to
make it as if Beeminder worked the way it ideally would. So we imagine
that if Beeminder had sent an extra “OMG Emergency Day!” reminder in
the morning, you would have dialed the road down, and then worked your
butt off to get back on the right side of the road. By that principle
we added a flat spot that puts you in the green today, but leaves you
on the hook for the 30/week rate until the 21st. Does this seem fair?
(In retrospect, even that was kind of pushing it! :))
So I think in the absence of a general principle to apply in this new year’s eve in vegas case, we have our verdict: No special flat spot inserted within the akrasia horizon!
And, as others are suggesting, lest it seem harsh, these commitment contracts are powerfully motivating and thus hugely valuable. The cost (either the cost of doing work while drunk in vegas, or just paying the $30, whichever’s less) is small in comparison.
In fact, we’d like to make that a principle in itself, but we don’t know how to operationalize it:
Whenever an $X pledge is paid, the person shall have derived at least $X worth of pants-kicking value out of Beeminder up until the point of derailment. We do think that that is turning out to almost always be true.
Having said all that, here’s an attempt at the other side of this argument:
The true point of a commitment contract is to eliminate akrasia. So anything your long-term self would’ve agreed to, there’s no need to be hard-nosed about. It’s sort of in the spirit of the SOS clause: emergencies or, more generally, truly unexpected circumstances, should be exempted. So if this had been a surprise last-minute trip, it would’ve been in that spirit. The fact that it wasn’t a surprise doesn’t change the equation much (other than making us feel like Merchant got what he deserved for forgetting to adjust his road). He obviously meant to and his long-term self would’ve approved of the exemption, so what’s the harm in pretending he did it?
Man, I’m almost convincing myself! But it’s a dangerously slippery slope – the value of bright lines is huge and “my long-term self totally would’ve been cool with this” is a very dim, fuzzy line!
If anyone has changed their mind or hasn’t chimed in yet, speak now or forever hold your peace! If the current consensus stands then Mike has to start busting his ass to get as much safety buffer as possible before he leaves. 
Yes, I’m really
I’m wondering if you
Thanks again to Bethany and Danny for a great service. I’m