Do-less goals must be anti-cranial-silicosis, but how?

Akrasia
1

(This started as a daily beemail last month and @mary pointed out that it really wants to live in the forum!)

Do you want to philosophize with us about how the Platonic version of do-less goals should prevent you from sticking your head in the sand and not reporting your transgressions?

Of course you do, you people eat that shchnitzel up.

Let’s do it as a STRAW POLL. Taking it as a hard constraint that do-less goals must derail if you ignore them, which approach is best:

[PPR] Pessimistic Presumptive Reports. Nice and simple, if I don’t report, Beeminder assumes the worst. Specifically, that I did twice my daily limit.

[AUCH] Auto-charge. Don’t mess with my data, just give me a countdown totally independent of where my data is with respect to the road. If I don’t report in that many days, charge me as if I derailed and keep the graph going, with another countdown. (Deadman switch still applies so eventually this will stop if you’re not interacting with Beeminder at all.)

[METAM] Meta-minding. The underlying problem is that I need to commit to entering data even when the graph itself isn’t making me do so. We can solve that problem with more Beeminder! Namely, make it easy/automatic to have a meta goal that forces me to enter data on the main goal at whatever frequency makes sense for me.

I think those are exhaustive. We have to derail you, so we can either do it directly (AUCH), do it with data (PPR), or do it with a separate goal (METAM). Anything I might be missing there?

Here results from the daily beemail folks:

[METAM] Meta-minding: 10 votes
[PPR] Pessimistic Presumptive Reports: 5 votes
[AUCH] Auto-charge: 4 votes

But I’ve concluded that AUCH is no good. Too much machinery and UI complexity and special cases and confusion. Unfortunately METAM, which may be the most principled and Beeminder-y and QS-y solution, has a long-ish list of prereqs before it will be sufficiently newbee-friendly, which means we’re probably stuck with PPRs for a while. (I mean, some of us genuinely like them. Also we’ve been improving them markedly lately, as you can see in the changelog – roughly http://beeminder.com/changelog#3147 through http://beeminder.com/changelog#3167 ) In any case, there are a lot of exciting things related to meta graphs in the future.

2

Some thoughts from @adamwolf on METAM, which he points out is nice because it’s using Beeminder to force yourself to enter data exactly how you want to be forced. “It’s the best tool we know of for imposing restrictions on yourself!”

But it probably needs at least 3 things (this is still me quoting Adam):

  1. Smart defaults.
  2. Dramatic visual integration on the goal page and dashboard, showing how the goals back each other.
  3. Optimizing for the newbee case. The newbee case is “I just want to stop dinking around on youtube so much, why do I have 2 different goals now?!?!” This maaaay mean that in the (likely two most common) default cases of “I want to have to enter data everyday (daily)” or “I want to have to enter data by midnight the next day (yesterdaily)” we have a shortcut where we don’t even show the second goal on the dashboard, and just tightly integrate a checkbox-or-not 1-past-week visual display on the first goal.

Back to me: This sounds right but kind of daunting! Now I’m wondering if we could just embrace the two-goals thing rather than trying to avoid it with Magic. That’s kind of what doc.bmndr.com/metamind is suggesting. Also do-less is already a premium feature and could potentially be bumped to Bee Plus and be power users only.

2 Likes
3

PPRs are really annoying and fundamentally don’t make sense without configuration. Why double the daily limit? And polluting the data stream isn’t the right way to go about it either.

If you’re actually going to invest effort here, I’d say just make this a special case of the more general feature request “Power multiple goals from one data stream” so I can do the same thing with other goals. For example, two Do More goals from the Fitbit integration (one with aggday sum to count my steps and one with aggday nonzero to count my syncs). Then the Do Less UI just needs to make it easy to Beemind both 1) the main goal as Do Less with whatever aggday and 2) the meta goal as Do More with aggday binary.

3 Likes
4

How would this help with manual entry goals? Or would it?

1 Like
5

I thought

was reasonably clear. The point is you would have a single manual entry data stream and you’re Beeminding both 1) you’re not going over your quota and 2) you’re entering data points.

6

I’m also firmly in the metaminding category. A Beeminder goal is the best way I know of to draw a line in the sand and make at least as much progress as I want to over time. Many oldhats with a billion goals already metamind a lot of goals–let’s make it easy and obvious to beginners too!

3 Likes
7

Paraphrasing some of the pro-PPR feedback I got:

@minstrelofc:

PPRs make do-less goals as easy to reason about as do-more goals. If you don’t enter data you move toward derailment at a predictable pace. (Note that there is a temptation to let the PPR be the implicit ‘max report amount’, and not enter data just to avoid reporting a even higher number.)

@lanthala:

I like how clean PPRs are: assume the worst, so you are almost always better off giving the real data. I think meta-minding is interesting but complicated (please don’t give me more goals to have to add breaks to), and the auto charge countdown offends my soul for some reason – I guess because the idea of being charged for something other than a derail feels extremely un-beeminder.

And another person said they like PPRs because they mean you have to think about the goal daily, and a downside-qua-upside is that if you have a lot of safety buffer then you can forget about the goal and let it fill up with wrong data but that’s kind of an autoratchet-lite for free.

2 Likes
8

Yeah, scheduling breaks on a lot of goals is Painful with a capital P. But I kinda feel like that’s an issue all on its own… :stuck_out_tongue:

2 Likes
9

That seems like a clear argument for meta-minding: ensure that I enter data every so often, regardless of safety buffer.

1 Like
10

Right, and the utility isn’t limited to Do Less goals. It can be useful for Do More goals to ensure that your integration didn’t go braindead and you have to burn up your entire safety buffer to notice.

2 Likes
11

If the goals can be coupled like that I’m in the METAM camp as well. Otherwise I don’t think METAM is hands-off enough to be honest.

1 Like