37 Signals is a web application development company based in Chicago, consisting of 7 guys who produce some very successful products. The day-long workshop called “Getting Real” that we attended was all about how they do that — what methods they have that help them achieve that success with the products they build.
[link deprecated: Here are all my workshop notes], exported directly from OmniOutliner (which creates some seriously crappy html code, BTW.) If there’s anything there that doesn’t make sense — and my notes are really rough, so there will be — feel free to ask questions here.
So here’s the dirt on how they do that, boiled down: They start with a set of core philosophies that they all believe in, about the environment/culture they want to work in, and about what’s important about the products they want to produce. Then they set up some methods of working that reflect those philosophies.
Those methods involve how they interact with one another, how they make decisions, how they design, code, and test their products, how they launch them to the public, and how they handle customer support on their products.
By consistently following those methods, and constantly referring back to those core philosophies (to the point where they call these ideals “mantras”) during their decision-making processes, they are able to produce products that are consistent, functional and pleasing to their users.
When you’re looking at their applications, or listening to them talk about what they do, it becomes really apparent that they’ve identified and filtered out what goes wrong with most software development, especially development that happens at large corporations.
One the keys is that they don’t try to create large, complicated applications all at once. They boil their apps down to some core goals, get them functioning quickly (thus requiring little documentation, meetings, and endless wrangling in the process) and iterate additional features quickly, but only when those features benefit all users and make sense. All this occurs while they refer back to their core mantras for guidance in decision-making.
So are their methods scalable? Can they be applied inside a large corporation? They sugggested several times that skunkworks projects and small groups inside big corporations can achieve that, but I think it’s also possible with a larger team, as long as the person at the top has enough of a vision to set the core philosophies and get people to focus on them and work within them on a daily basis.
Pictures I took in Chicago — not many of the workshop, but lots of the city.