Derailment statistics: YAY or NAY?

almost three months ago now i wrote an e-mail to support because i’d been confused about one of my autodata goals (which i’ve since archived)—i’d vagualy remembered a derailment on it, but looking at the graph, i couldn’t find it anymore, and was confused. turns out some autodata goals overwrite datapoints when there’s been new progress synced since, for the day of the derailment! (which does absolutely make sense, when you think about it.)

(this story has not necessarily very much to do with the following post/question, but it might be helpful for context, maybe. or maybe not. in any case, i wouldn’t have known how to start this post without this bit of context XD)

do you think that there should be statistics about derailments in the goal settings?

(either optionally toggle-bly if one wants to see/hide them, or always, permanently there, maybe hidden in a more obscure place)

personally, i think that it would be super valuable to have statistics about derailments; like:

  • number of past derailments
  • dates of past derailments
  • maybe also “how much data would have been missing for a derailment to not have been a derailment, on average” (+ total minimum/maximum)
  • other ideas?

“number of past derailments” can of course sometimes be inferred from the current pledge amount, but that works Not At All when the pledge cap is low!
“dates of past derailments” (and thus also the total number) could of course be found out by scrolling through the graph, but, let’s be honest, even with my not-even-half-a-year-long-ones that’s absolutely too much effort for me already, for most of them. (besides, i would have to make a note of it somewhere else, which is, of course, tedious :p)

semi-relatedly, i would also love it if the starting pledge for a goal could be shown somewhere! (i’m the kind of person that forgets these things within minutes!)

of course, circling back to the beginning, all beautiful statistics would be completely lost/useless on autodata goals where a derailment data point will be overwritten the next day … *sad mara noisesvisuals*

@olimay, who answered my e-mail at the time, brought up, among other things, some valid arguments against having derailment statistics:

It’s not something we currently tabulate. One concern is that tabulating it and making it available (even it’s in a relatively obscure place) might end up being really demoralizing for some users, so there’s some caution against implementing it.

which absolutely make sense as well, unfortunately.^^

what are your thoughts on possibly having derailment statistics at our disposal? YAY or NAY (and why)?

5 Likes

Whenever I want to look back over my derailments, I do so through the payment history on the payments page. That doesn’t include zero-pledge derailments, of course, but those aren’t “real” in some sense anyway.

I’m saying this both in case you didn’t know about it, but also to register my skepticism at the risks involved in showing statistics about derailments even in some obscure place—such a place already exists, and it seems that anyone who would be demoralized by it should be similarly affected by the existing page.

(I find it useful to, at the start of each year, go over all my derailments in the previous year, and consider how to adjust my aggressiveness for the upcoming year. That’s my primary use for derailment statistics, and it’s well-served by the payments history page. I don’t have a strong opinion about if more detailed statistics about derailments should be shown somewhere.)

2 Likes

I do have a feeling that having derailment statistics would be useful, but only in comparaison to other goals.

E.g.

brushmyteeth, $0.30/day
stop_working_at_midnight $2/day

I think that I’m overspending on useless goals and not being agressive enough on some others, and seeing how much each goal costs daily might be a good indicator

4 Likes

This is very good idea. Bumping the thread.

2 Likes

very neat, thank you for mentioning this! (i searched through my bank account once to get similar stats, it has pretty good filtering options—so looking at all my “beeminder” charges, or specific “goalname” charges is pretty easy. [turns out my bank app has problems with numbers-only goalnames, though, so looking up “1923” returns no results.])
regarding your skepticism at the risks involved in showing derailment statitstics even in some obscure place: right after i started this thread it came to me that i forgot to mention Derailing It Is Nailing It | Beeminder Blog, which falls into the same vein of argument, for me.—i do agree that “derailing it is nailing it” probably won’t be the first, most intuitive advice/concept a newbee will embrace. however, it doesn’t seem philosophically very compatible with the argument “No Derailment Statistics Anywhere Because People Could Be Scared Off By Them”.

love this idea, that does indeed sound like a useful thing to track as well!

new idea! (additionally to those already mentioned)

having the derailment triangles of non-legit-derailments be another colour (green?) than red.
useful for people with bad memory (me!) and for people with long-running graphs (future me!). especially useful for zero-post-derail-respite goals, or for those autoratcheted to 0 buffer.

2 Likes