Problem summary:
PyCon, the North American conference for the Python programming language, an all volunteer conference was facing strong growth and was on the verge of outgrowing its existing site selection and planning system.
Role:
Chicago bid chairperson and PyCon local coordinator
Collaborators:
Members of the Python Software Foundation
Key insights:
1. For conferences, scale means increasing marginal costs, not decreasing.
2. Feelings of community ownership extend far beyond software in
the open-source world.
3. Volunteers are critical for the community character, but relying on volunteer effort makes continuity difficult and makes specialized skills required for large event planning difficult to maintain.
the open-source world.
3. Volunteers are critical for the community character, but relying on volunteer effort makes continuity difficult and makes specialized skills required for large event planning difficult to maintain.
Solution:
As a part of the Chicago bid to host PyCon, we developed a process to redefine and centralize the planning and site selection process. This enabled PyCon to grow from 450 attendees in 2006 to more than 1700 in 2012 without burning out volunteers in host cities long before the event started.
One essential component was engaging a professional meeting planning company to provide the required logistical and negotiating expertise from year to year – and we sold this to the volunteer-driven Python community continue volunteer control while outsourcing tasks volunteers were least suited to do.
Much of the conference experience was treated like a prototype and thus open to revision, adaptation and co-creation. One fun example was created the night before the conference began – a quick talk feedback system for participants exiting talks.
We'd had a difficult time getting attendees to fill out feedback forms for most talks (only the extreme ends would get reviews, with negatives getting much more attention in our experience). So, we asked ourselves, how might we enable rapid, low-cost feedback?
We'd had a difficult time getting attendees to fill out feedback forms for most talks (only the extreme ends would get reviews, with negatives getting much more attention in our experience). So, we asked ourselves, how might we enable rapid, low-cost feedback?
Our answer was to have attendees toss a chip into one of three buckets – green, yellow or red to rate the talk from "good", to "ok" or "bad", respectively. This simple approach was inspired by the (new at the time) idea of Net Promoter Score. Combined with a headcount in the room and we had a very powerful metric done quickly, on the cheap, giving us a real, usable feedback to a level that we'd never had before in the history of the conference.