If you’re just tuning in, and if you care about this for pragmatic rather than philosophical reasons, you’ll want to start with (or stick with) our announcement that we have fully killed off auto-widening yellow brick roads.
This is the part where we philosophize about why this is a good idea. Equivalently: why our blog post from 2011, where we twisted ourselves into a pretzel trying to justify the complexity of auto-widening, was wrongheaded.
It’s pretty fascinating and embarrassing to read that old post. The first glaring red flag is right in the title: “The Magical Widening Yellow Brick Road”. You almost never want software to do anything seemingly magical.
Beeminding weight loss
Our rationalization of auto-widening starts with weight loss. Beeminder and weight loss go way, way (weigh) back. The first lines of code that would eventually become Beeminder (12 years ago!) were to draw a yellow brick road to your goal weight on a graph. But we noticed that one of the biggest problems with losing weight is that your progress is rarely steady from day to day, so that it doesn’t feel like a measurement that is wholly under your control. There are causes behind these fluctuations that careful attention can tease out. For one person it’s eating salty foods, or drinking a beer; for another it is having a uterus. But they commonly make weight loss bewildering and frustrating.
“It was circular logic”
Beeminder used to try to mitigate the pain of that for you, with the afore-mentioned elaborate magic to adjust the width of the road. But that doesn’t make it any easier to lose weight. Making goals easier makes them harder! Tampering with the road width just makes the failure condition more murky. Ultimately, we decided that it’s much more valuable to have a single, predictable, immovable bright line. And users seem to agree.
“Stay below this line and you won’t derail tomorrow”
Without getting too far into the depths of the Rube Goldbergian auto-widening concept, we claimed to give you a guarantee that if you kept your weight below a particular bright line — the centerline of your yellow brick road — then we’d guarantee, by auto-widening your road, that you wouldn’t derail tomorrow. This was circular logic! If you can actually stay below that bright centerline then you can just as well stay below the even brighter you-actually-derail-if-you-cross-this-line line. If you can’t actually stay below the centerline, which, of course you can’t, then the whole point is right out the window and you’re skating the real line anyway.
Also, realistically, no one was motivated by that guarantee anyway, not even us, since it was so hard to understand.
The one thing that auto-widening did for you was let you postpone choosing the right amount of initial safety buffer. But it’s much cleaner to just make you guess your maximum daily weight fluctuation at the start, as we now do.