An Exit Interview with Mapzen’s Dan Phiffer

On January 2, the mapping platform Mapzen announced that they were shutting down. CEO Randy Meech has declined to explain why, but we reached out to former Mapzen senior software engineer Dan Phiffer to reflect on his time with the organization in the following exit interview.

In case anyone is reading this who isn’t familiar with Mapzen, could you briefly describe what Mapzen was?

Mapzen was a company making open mapping more viable for developers out in the world. Our tools cover the full range of map-making: cartography, graphics, search, routing, data, transit, mobile apps. And we published all of it as open source code and, in the case of Who’s On First, the project I work on, as open licensed data.

The project began in 2013 with a small group of tech people, led by Randy Meech, our CEO. We were a part of the Samsung Next accelerator, but operated under a slightly different part of the org chart. We had a lot of freedom to work out in the open on the problems that were most interesting for making maps.

And now that Mapzen is shutting down—what happens to all of that work?

The vast majority of the code and data will remain online in our public GitHub repos. Each team operated as a kind of independent city-state in the larger Mapzen kingdom, and each team has had a different approach post-shutdown. The team that created the Valhalla routing engine is joining Mapbox to work on similar routing problems. Our transit schedule data project Transitland will continue to operate. Who’s On First will also continue on outside of the Mapzen framework.

In each case, the code and data were always designed to outlive the company. Specifically what that will look like in practice is still shaking out. Aside from the code that runs the metered API system, each service was always designed to run without there being a Mapzen company behind it.

Why design code and data to outlive a company? Is that a kind of strategic pessimism?

I suppose a lot of us have just been through startups before. We share some company backgrounds: I worked at (as did Mike Cunningham), which was acquired by Patch (where Randy was CEO before), which then got subsumed by AOL. Aaron Cope, who I work with on Who’s On First, had started a very similar places database project while he was at Yahoo, called Where On Earth. The similarity in the acronyms WOE and WOF always seemed like a tacit admission we had picked up where that project left off.

Basically we were always working on open geo primarily, and a lot of that work feels more durable than any particular company. But yes, the fact that we had openly discussed our work outliving the company had felt somewhat macabre before the shutdown. In retrospect it seems like a good strategic choice.

How hopeful are you that the tools you’ve helped create will continue to be used? Civic tech projects (especially those made by volunteers) have sometimes come under criticism for too easily falling into disuse if there isn’t an individual or a group taking ownership. Could that happen here, especially with the projects that aren’t going to continue outside the Mapzen umbrella?

Yeah it always strikes me as suspect when a project announces they’re “giving the code to the community.” Time will tell how much of what we built will be actively maintained. I can speak for Who’s On First and say that we are continuing work on it, to the degree that we can. In the cases of tiles and search, I know there has been interest in them doing consulting work for other companies. It’s good work, so I feel pretty good about the prospects that things won’t wither away now that we’re going our separate ways.

I have my own civic tech side projects that I hope will continue to work using the Mapzen stack. A lot of it is just how I’m accustomed to working, but stuff like Metro Extracts doesn’t have a great equivalent anywhere else.

If there are Mapzen components that have proven essential to your organization (that is, to you the reader), you should consider ways you can give back to it. We’ve had the benefit of working under the comfort of a big corporate sponsor, we may not have been as open to outside contributions as we should have been. I’m hoping that each team can figure out the best way to make that happen, post-shutdown.

Now, as an organization working under the auspices of Samsung, how important was profitability to Mapzen? What was the plan for long-term sustainability, had that been achieved? And do you have any advice for organizations in similar positions?

We were starting to figure those things out, but they’d never been a big focus for us early on. I can’t decide if that was what doomed us, or if that’s what made working at Mapzen interesting to begin with. Maybe both. The ways we were making money were through a metered API and through individual one-off deals we’d been making with organizations (World Bank, Tesla, Portland’s Tri-Met). We’d also been reselling access to our services through CARTO.

So yes, it was important that we figured that out, and I honestly have no idea if there are great lessons to take away on the business side. I don’t feel qualified to speculate, mostly I was happy that someone else was thinking about those things so I didn’t have to. I’d hope that our shutdown doesn’t dissuade others from attempting their own version of what we tried to do.

Is the future of civic tech (if you can comment on something that broad) for-profit or nonprofit?

I have no business making sweeping statements about the future of civic tech, but I think for-profit companies have a role to play if they can do it in a way that’s not extractive and exploitative. Nonprofits have their own challenges and biases, so I guess I would like to see more of all of it. I’m personally interested in platform cooperatives and public benefit corps, but I’m also willing to let big monolithic corporations have their chance to do the right thing.

How has your experience at Mapzen changed how you will run future projects, including the spin-off of Who’s On First?

I learned a ton working on Who’s On First, in part because of all the scaffolding and mentorship from folks like Aaron Cope and Nathaniel Kelso. Some of it was a lot of boring engineering stuff, about how to make sure whatever code you’re writing is going to work in two years or so. But also about the importance of writing about what you’ve done. I’m proud of the blog posts (also archived to Medium) I was coerced into writing. The final posts I wrote were about how to set up our data editorial tools on your own hardware, and it forced me to take my work home with me as a working server.

Writing is really hard, but I think the company blog has been an amazing way for us to explain why we care about open geo, and what kinds of problems we could solve with the tools we were building. Posts like Escape From Mercator still boggle my mind.

We’ve been discussing how Who’s On First should probably become more accommodating to outside contributions since most of the work I was doing on our web-based editor was still internal use only. We want to attract a larger group of editors to help us maintain the gazetteer, so that will likely inform my next steps with the project.

Anything else you’d like to add?

Lots of great people have passed through Mapzen who I’m really glad I got to know. There was a broad sense of sadness about the shutdown, but also a lot of gratitude that we had a chance to spend the time we did working on open geo problems. I should mention too that there were a whole range of great people, not just in engineering roles. And not just people who are around for the shutdown; we had Ingrid Burrington, Patricio Gonzalez-Vivo, Lou Huang, Michal Migurski, Dave Riordan, Alyssa Wright all passed through Mapzen, just a really solid group of people I’m happy to have spent time working with.

I encourage everyone to look through our Twitter list and offer us all fabulous new jobs! But seriously, I’ll be really interested to see what everyone goes out and does next. We all joined joined a new “postzen” Slack team and are already posting cat pictures and such.