Beeminder's 2022 Priorities

Bugabee
1

I took a poll in a beemail in January about how we should prioritize some projects in 2022 which I’m repeating and formatting as a new poll here:

  • [AUTOD] Steadily adding integrations (Focusmate has been So Huge!); in particular the meta integration
  • [UNCLE] The uncle button (as in crying uncle, like “yes yes, stop zeno’ing, I’m intentionally taking the derailment on this”)
  • [SCHED] Revamping/optimizing the pledge schedule
  • [BOARD] Just gradually making Beeminder better and clearer and onboarding better and emailing people more when they fall off the wagon, etc etc
  • [GRAPH] Dynagraphs, aka merging graph.beeminder.com into Beeminder proper
  • [STAIR] Bright Red Staircase? [hopeful-face-emoji]

0 voters

Feel free to cast votes again if you were part of the original poll. It’s approval voting this time – see side rant below – so vote for any that you approve of as being a top priority!

Better yet, here’s a prediction market:

Excerpts from the original replies to the beemail
  1. Ooooo bright red staircase, making the red line graphically consistent with time/derailment
  2. Upgrading the mobile apps (multiple people added this as a write-in)
  3. Just snoozing zenos for the rest of the day could be simpler version of UNCLE
  4. Various autodata integration suggestions which we’ve added to the list on consider.it
  5. Better goal stats (max datapoint, averages over different windows, streaks, etc)
  6. “Integrations are everything”
  7. Onboarding OMG it took me forever to really understand Beeminder
  8. “UNCLE should be easy, and be a nice quality-of-life improvement”
  9. UNCLE is a double-edged sword; not having it has kept me from derailing

And a million other opinions, often fully contradicting each other, like “yes, imperative that you do X before anything else” and “absolutely avoid X at all costs”.

Astute observers will note that we’re already 16% of the way through 2022. And we have in fact made progress. We eliminated the biggest chunk of onboarding confusion we knew of with UVI#4017, though that’s clearly only the tip of the iceberg. Also we slotted in No-Excuses Mode in case you missed that. Mostly we’re still boringly wrapping up some infrastructure upgrades (Ruby on Rails and Stripe Checkout, which also involves dropping support for PayPal).

Back to the poll results, we got 30 responses and the average rankings were like so:

GRAPH   2.82
UNCLE   3.18
AUTOD   3.32
STAIR   3.60
BOARD   3.77
SCHED   4.32
Side rant on voting mechanisms

This is a nice illustration of why I dislike Ranked Choice Voting and prefer Approval Voting. It’s pretty hard to get a real sense of the aggregate opinion from those average rankings. Other than that the top couple options are ranked significantly higher than the last couple. I tried looking at a smoothed histogram of the rankings but that doesn’t help much, other than to highlight how unliked SCHED is:

image
image1344×804 100 KB

I even tried instant-runoff / iterative elimination on the rankings. That yielded this ordering:

  1. GRAPH
  2. AUTOD
  3. STAIR
  4. UNCLE
  5. SCHED
  6. BOARD

In conclusion, Approval Voting Wins. Yes, it’s dumb how I’d already concluded that and then did Ranked Choice anyway. It was an interesting experiment I guess!

I think in this case I was tempted by how Ranked Choice felt like it would demand less strategizing. And I imagine everyone did just truthfully give their rankings. Whereas with Approval Voting you need to understand the strategic choice you’re making. If you approve of everything, you’re wasting your vote. If there’s something you want to vote against, you can vote for everything but that, or if there’s one stand-out favorite you want to put all your weight behind, vote for just that. Or anything in between.

Slightly harder for the voters but I think the results would’ve been a lot clearer. (Also I don’t think Ranked Choice is actually better than Approval Voting even in terms of robustness to strategizing, if the stakes were high enough.)

Anyway, the conclusion is that dynagraphs (GRAPH) and the uncle button (UNCLE) are most important (the latter especially because it’s actually quite easy) and autodata integrations (AUTOD) clearly need to keep happening periodically whatever else does.

Probably y’all are wrong to have rated onboarding / goal creation stuff (BOARD) so low but that makes sense because I forgot to clarify whether you should answer from your own perspective or Beeminder’s. (And usually I do mean to poll your personal opinions!)

3 Likes
2

A slightly edited version of my email response to this beemail:

In my opinion, none of these come close to the urgently needed priorities, in this order:

  1. DEBUG - thoroughly debug Beeminder / aka project PlayNice / make everything work together well

  2. DOCS - thoroughly document all features in one clear, easily accessible, comprehensive, unified, indexed reference guide and tutorial

  3. INTERFACE - fix the confusing and unintuitive web interface where settings are divided into 5 settings panels (one of which is called “settings” lol), apparently randomly, making it difficult to find

  4. DROID - give the Android app all functionality of the website, i.e., support mobile-only workflow

I think all of these, which don’t involve adding new features, are more important than anything you listed - there is a failure mode where you add new features too fast to properly assimilate, document, and debug them, and I think it would be best to not add new features until the current ones have been polished a bit more.

3 Likes
3

Except for the last one (which totally shouldn’t be a priority, IMO), this is basically BOARD. And I do agree that from Beeminder’s perspective that one should probably have the highest priority.

2 Likes
4

Ah, I see they slightly changed the definition of BOARD - in the original email it was “Revamp goal creation, onboarding, email people more when they fall off the wagon, etc.”

That’s a little different from my suggestion, which is specifically not to add new features, and is not about onboarding or goal creation, but is just to fix bugs and documentation. So I think it should be a separate category.

The new BOARD added “Just gradually making Beeminder better and clearer” which is really vague, since the whole point of any of them is to make Beeminder better.

From my perspective as a user, fixing bugs, documenting, and making the interface better should be the top priority.

I am still not going to vote for BOARD as I don’t think it reflects what is actually needed most, so if available I’d choose “none of the above.”

1 Like
5

The poll is a good idea, but let’s not forget that we’re looking at a biased sample of hardcore users. In other words, anyone who was frustrated by “on boarding” is not longer here and won’t respond to the poll.

My vote is to make it easier for new users to join and use this incredible tool. It would take just a couple of language tweaks (and maybe an example or two), and we can support many more in getting closer to their goals.

I agree with @ianminds and @zedmango

2 Likes
6

Just came here to say it’s good to see you alive zedmango :slight_smile:

Edit: okay, I’ll try to be constructive. To be honest, I have to agree with zedmango. None of these suggestions seem helpful (except BOARD which is kind of meta). What about you add zzq’s goal-collapse feature? That would be a UVI that I would notice (or not because I already use the add-on).