3 Maxims for Great Developer Platform Design
NEW YORK — Russell Miles opened his keynote at this week’s Devmio International JavaScript Conference with a guitar and a pictorial tribute to the historical figures who created platforms. He soon had the audience clapping and playing along.
Miles, a self-described listener, author, reader and developer, spoke on “How Developer Platforms Fail (and How Yours Won’t).” He has been building platforms for 25 years, including work on the Spring framework.
“I’ve been through hell as I’ve been trying to build those platforms and establish them in different organizations,” he said.
He pointed to some unusual inspirations, including American satirical author Kurt Vonnegut, as he outlined three maxims about platforms. Vonnegut claimed to apply science to stories and said that every story starts off with: things are pretty good and then things are not good. Then, by the end of the story, things will be pretty good again, Miles explained.
“I’m going to tell you a story like this except that it won’t be very Vonnegut,” he said. “I’m going to start with pain. I’m going to start with the why you shouldn’t build a platform or why you may not want to.”
Maxim 1: You Don’t Really Want a Platform
Platforms aren’t built overnight and they’re not a one-and-done proposition, he cautioned. It requires people dedicated to it, who will curate the platform.
“When I get into an organization, the very first question I’ll ask is who’s going to build it, who’s going to own it and how many years commitment have you got on this,” Miles told the audience. “If any of those answers are, that’s not what we’re thinking, then I’d say you don’t want a platform.”
The catch is, every company already has a platform, he said.
“If you’re shipping code, you’ve got one already,” he said. “A platform experience begins the moment a developer is trying to do anything in your organization.”
Building a platform requires understanding who your internal customers are. It also requires building roadmaps. It needs to be viewed as a product the company is going to invest in, he said.
“There’s a whole collection of effort and life that goes into curating a platform,” he said. “If you want the impact of a platform, if you want to go through Kurt’s story and end up with perhaps outrageous good fortune, then you need to frame it as a product you’re going to invest in.”
When he talks to companies about how they can establish a platform, he talks to them about how the platform feels to developers.
“I don’t start with where are your repositories or where [are] your build pipelines,” he said. “I start with, what does day one feel like to a developer here — what do they have to do?”
Then he finds a developer and sits and listens to their frustrations.
“I’m not there to judge them. I’m as clueless as they are,” he said. “That’s where a platform begins.”
Maxim 2: Platforms Are Not Product Process Alone
It’s the impact that sets a platform apart, he said. Again, he pointed to an historical figure, Christophe Plantin, a French painter. Plantin moved to Antwerp, Belgium, where he built the world’s largest printing house in the second half of the 16th century.
“What he invented was mass market publishing, printing. He created markets where you could distribute books,” Miles said. “Now this is when we’re talking between 2-5% of Europe could even read, and most of them wouldn’t read anything except theological disputations. So he was creating a mass market platform when he had no mass market to apply to, but he believed there would be one.”
Not every invention is a platform, of course. The printing press was not in itself a platform. Plantin’s work is a platform because he took all the innovations up to that point, curated them, then created and packaged the results in a way that is consumable by a large number of people.
“His intention was to make money. Sound familiar? No one really creates a platform without thinking there’s an audience out there for it,” he said. “Plantin created for the thousands of potentially literate that were going to be out in Europe, for millions eventually. That was his belief. But the platform had an unexpected impact, a surprising impact. Great platforms should surprise you with how people use them and that’s exactly what, of course, happened with Plantin’s work.”
Along the process, Plantin managed to reach two particular 14-year-old boys.
“Both of these young boys were going to benefit from the platform that Plantin was creating,” he said. “Those two boys were Galileo and Shakespeare, both 14 years old, both on the cusp of lives that would change everybody’s perspectives, one in science, one in English, and they wouldn’t have been able to do any of their work without what Plantin created. This is the essence of a platform.”
Great platforms should surprise you with how people use them, as Plantin’s work did. Too often, engineers are stuck on the idea of platform as a product, but the product isn’t what matters, he contended.
“This is the essence of what a platform should look like,” he said. “We’re trying to create inflection points.”
Maxim 3: Curate Platform Habitability
Miles took this maxim from “Patterns of Software” by Richard P. Gabriel. To explain, he compared creating a platform to creating a kitchen in a restaurant.
“Before you curate a platform, what you’ve usually got is everyone going out, sourcing their own ingredients, cooking their own dinners, all bespoke, and no one is getting any benefits of this, because people are just using the facilities,” he said.
Creating a platform is like building them a kitchen, he added.
“There are the people that are trying to eat meals, and we’re going to give them the menus, we’re going to give them the product interface, the APIs, we’re going to give them everything that we can to keep them buying food and eating it,” he said. “Then, the platform engineers are going to be behind the scenes making sure that we deliver the best menus and the best food they’ve ever had.”
It’s important to remember: It’s a kitchen, not a cathedral, he added.
“I usually encourage organizations to create an internal platform completely in open source, where anybody could consider contributing to it for this reason, because I don’t want people not to be able to see how it works,” he said. “I just want them to trust it enough to rely on the features of it. Sometimes, for engineers, that trust means I need to see behind the wizard’s curtain, so don’t prevent them.”
It’s important to listen, too, to what developers need. The only way to succeed is to realize you’re not the expert in what they’re trying to do, he warned.
“Build it and they won’t come,” Miles warned. “The only way they come is if they can see themselves in it, they can see their work benefiting from it.”
