A Diary of the Surprising Launch of Guesstimate
Eight weeks ago, I didn’t know if people would ‘get’ Guesstimate. Guesstimate is a spreadsheet that runs Monte Carlo simulations, which can be really useful but aren’t known to most people. Was this an education problem? A tools problem? It was tough to tell, but I had lots of people telling me that almost no one would understand it.
I kept coding. I knew I wanted it, and wanted to know for sure if a larger community would as well.
On December 30th the app was ready for launch. I wrote a short blog post and posted it on Facebook. It wasn’t much of a marketing plan, but I figured I could re-launch once there was some more content.
Several of my friends shared it, and then their friends shared it. People started to sign up and make models. One estimated the time to Tear Down a Christmas Tree, then another How Long it Takes to Get Ready for Preschool. People I didn’t know were figuring it out and making wonderful insights.
Soon it was tweeted about by Ramez Naam, author of the Nexus trilogy. It started growing rapidly. It got late in the US and I prepared for bed, but then people started joining from England, Norway, then Russia. I eventually gave up on reading each model, and headed to bed at 3am.
I woke up early the next day. Bret Victor promoted it and got hundreds of retweets. Guesstimate was soon submitted to Hacker News, Product Hunt, and Reddit. Despite the lack of an informative home page, it gradually rose to reach #1 on Medium, #1 on Hacker News and #2 on Product Hunt Tech. I frantically scrolled through dozens of tabs to respond to an outpour of comments and questions.
People found the Github page, which started getting popular as well. Several issues were created and I received a lot of advice on how to add distributions. On the down side users had difficulty getting it set up on their own machines, and I just didn’t have the time then to fix the Webpack setup for them.
BoingBoing published an article, then did Flowing Data. I was contacted by journalists. Soon it was featured in Fast Company and the Online Journalism Blog.
During all of this I could barely sleep. I had wrist pain from typing so hard to urgent inquiries, and a throbbing headache from the excitement and sleep deprivation. I was delighted, excited, and very, very tired.
Eventually things stabilized. I got some rest, then got back to work on scaling. In total, Guesstimate had over 4400 sign ups and 3600 models, many of of which were quite sophisticated. It’s still growing, though not as quickly as during that first rush.
It’s hard to say if the response could have been predicted. I would have previously estimated this success as less than 3% likely.
But I think this demonstrates that despite the early critics, there are a lot of people out there who understand and enjoy great probabilistic tools. The notion that Monte Carlo simulations are ‘too complicated’ for most people seems wrong, especially when they are really easy, intuitive, and useful.
I’m more determined than ever to continue onwards.
