I’m Leaving Microsoft Research for a Startup

Wow. It’s startling to even say … but after thirteen years, I’m leaving Microsoft Research and joining a funky little startup in San Francisco.

Where I Was

I’ve been carrying out research with Microsoft Research (MSR) since I finished graduate school, in 2004. MSR sits outside the direct product areas at Microsoft; its mandate is to both expand the state of the art of the field, and ensure Microsoft has the latest technologies.

The parody in this chart (from Manu Cornet / Bonkers World) feels a little dated today — but in the Microsoft chart, MSR would probably be the arms dealer.

MSR is a fascinating entity: part university department, populated with PhDs; part prototype shop.

Microsoft Research is truly an amazing place. In my thirteen years here, I’ve gotten to work with some of the smartest, most dedicated people in the world. It’s really a unique opportunity to be able to wander down the hall and talk to people who were involved in inventing whatever technology I’m curious about. I’ve seen ideas in early forms that, a few years later, would bubble out to the outside world. For people who are contemplating your next steps, and Microsoft Research is on their radar, I’d strongly encourage anyone to push forward with them.

Admittedly, I like Microsoft more then most rational humans do, or probably should. I think most of the hardware and software that the company makes is, by and large, pretty solid. Not only do I love my Surface Book — I even miss the Windows Phone. The microchip in my brain is firing just fine. I have a deep and abiding respect for the work that it takes to make Microsoft go: I’ve gotten to work with product teams that make the software we each depend on, every day — and I’ve gotten to join them in wrestling with some really interesting design questions.

Some of the projects I’ve worked on: network analysis of email; animated voronoi treemaps of corporate hierarchies; touch driven controls for machine learning; a novel architecture for Excel; textual analytics on a tabletop

My own research has gotten to explore a lot of exciting areas, largely centered on data visualization. I started out working on the Netscan project, and thought about how you could discover different social roles in an online community. In the last few years, I’ve written an O’Reilly book, visualized Big Data, and lots of other fun things.

There are challenges to MSR, too. The greatest strength of Research is that it operates at a remove from product; that is also its greatest challenge. The process of transferring technology with product teams can feel like an intricate dance of finding good partners and good technologies and fitting into their design calendar at just the right frequency. Getting noticed among so many smart people and exciting projects can be difficult, too — when a colleague and I started collecting the “top few” projects that people were most proud of, we got hundreds of nominations.

Why a Startup?

After thirteen years of research, it’s time for me to try something else. My next step is to try a context of working directly with a product, and in a place where I can actively affect its direction. I’ve decided to join Honeycomb, a startup in San Francisco. I’ve chosen Honeycomb because I believe it can succeed, I believe in its mission, and I think I have a real opportunity to make a huge impact, not just on the company, but on the world.

One of the most critical areas right now where visualization is being used is for system monitoring. It’s really hard to understand the complex behavior of a distributed system — of databases, and compute roles, and user interfaces. When something starts to go wrong, operators want to see what’s happening, and how it is different from before. That is a complex, time-sensitive task — and requires a sophisticated visual analysis to drill through layers of data to diagnose the situation.

The status quo, however, separates logging, monitoring, and tracing under different tools. This makes it even harder to understand precisely what is going on, or where a problem came from. (Yes, I know. When I heard that, I was baffled too.)

Honeycomb brings together all these layers under one roof. They’re bringing me in to strengthen the user experience, and to coordinate a vision of themselves as a visualization-forward company.

After a few weeks of much-needed vacation, I’ll be starting with Honeycomb in mid-June. I’ll keep living in Seattle; there will be a mix of telework and commuting. I’ll probably be at fewer academic conferences for a while, but I’ll try to keep a toe in the game.

You can reach me by email as danyel at gmail.com; my website is http://danyelfisher.info.

Please keep in touch.