« Beeminder home

Beeminder Blog

Cookie Monster cookie monstering a cookie

Our esteemed cofounder, Bethany Soule, gave a talk at the 2013 Global Quantified Self conference in San Francisco. We just got the video of it and wanted to share it with you, along with a transcript and the slides. See also our previous Quantified Self interview from 2011 and our previous Quantified Self talk from 2012.

1: Track Your Time.

Howdy I’m Bee. A couple years ago I founded Beeminder with Danny Reeves. Beeminder is Quantified Self + Commitment Contracts. That means you pledge money to stay on track toward your goals and we collect your data, graph it, and if you fall behind we take your money.

“Was I actually spending 18 hours of the day doing a depth-first search of the internet?”

This is a story of akrasia, time tracking, beehavior change, and the birth of Beeminder. Right. So. I’m akratic. Meaning, I do stuff against my own better judgement all the time. Definitely including that [points to cookie monster]. In particular it seems like I waste massive amounts of time when left to my own devices.

So I started to track my time. I wanted to know was I actually spending 18 hours of the day doing a depth-first search of the internet?

If you’re interested in tracking time too, I highly recommend RescueTime. It’s awesome. It just runs in the background on your computer, et voila. You are tracked.

My problem with the totally passive solution is it can only track things that happen at the computer and it can’t easily collect info like what project this code is part of, or if I am in fact daydreaming.

The solution we came up with was stochastic self-sampling. We call it TagTime. It runs in the background on my computer and at unpredictable intervals throughout the day it pops up a window and asks “What are you doing right now?” I add text tags which get written to a log file along with the timestamp.

It’s random — every 45 minutes on average — so over the course of a single day there’s high variance, but over time I get a very accurate picture of how I spend my time. I’ve been doing this since 2007. I’ve got over 72 thousand entries in my log. So that’s pretty accurate.

2: Graph Your Time

In 2007 when we started this I was in grad school. I wanted to make sure I was treating it like a full time job and spending 40 hours a week on it. So we started graphing the data. And we went a step further. This yellow line, the “yellow brick road”, shows how much I should be doing — the idealized path. The data points are how much I’ve actually done.

If tracking changed my behavior, it still did not make me more efficient. I didn’t keep on track — in fact, this looks a lot like what you’d expect from a student. Not much time spent in the beginning of the semester, then at the end I worked like crazy. I was exhibiting the same behaviors as before, only now I had proof.

3: Daily Progress; Risk Money

Fast-forward a couple years, Danny and I were using TagTime to track the time we spent on building Beeminder. We had read a Paul Graham essay about startups called “How Not to Die” in which he suggests that if you could fully commit yourself to not giving up, your chances of success would be 90%. So, naturally, we started to bet on our yellow brick roads! We committed to staying on track and never falling below 40 hours a week or we’d pay actual money.

At first it was to friends who we’d make literal bets with. Now we’re both up $270 and that money goes to our users. We also took the plunge and left our day jobs to work on Beeminder all the time.

4: Track the Right Metric

We were fully committed. We had rich data, and nice graphs. We had money on the line. But we got up to 2000 person-hours or so and we still weren’t publicly launched, our beta users couldn’t create new goals without going through an admin interface, and it was seriously tempting to throw in the towel and get a real job.

Tracking time is important — forcing ourselves to put in a minimum amount of time has sometimes been a necessary kick in the butt and sometimes is a welcome excuse to escape and get back to work — but it was not enough.

So we decided to commit ourselves to visible forward progress. In February of 2011 we made up a metric — User Visible Improvements, or UVIs — and we committed to averaging one per day. We tweet them to @beemuvi, so users can keep tabs on us.

If we fail — if we ever fall below that Beeminder road — we’ll pay a user $1000. Our users pay us when they go off track, and we pay our users when we go off track. We’ve been at it for over two years now, and we’ll hit 1000 UVIs on Nov 16. So the results?

5: Awesomeness!

“I totally still procrastinate. But I have a tool that lets me parameterize how much.”

Beeminder pays my bills now, and makes thousands of people more productive and awesomer every day! Minding that metric (UVIs) means the focus is on making our users happier every single day (on average!) Which makes them shockingly happy to pay us when they go off track.

Beeminder does more than just track time. It integrates with RescueTime, and Runkeeper, and Fitbit, Withings, Duolingo, Gmail. [UPDATE: Full list of Beeminder integrations at the bottom of beeminder.com/services.]

I track a bunch of other stuff too, both manual and automatic, like pushups, getting to bed on time, working on my QS talk, and my weight.

Beeminder’s combination of Quantified Self + Stakes has done wonders to improve the happiness of this overambitious lazy programmer. I totally still procrastinate. But I have a tool that lets me parameterize how much.

(Note: Above is the version of the slides that were supposed to be used for the talk but there was a technical glitch and an old version was projected/recorded.)