Larry Birnbaum – Farley Fellow Q&A
Larry Birnbaum, professor of electrical engineering and computer science
Larry Birnbaum is co-founder and chief scientific advisor of Narrative Science, a company that extracts insights from pools of data, turning it into actionable stories and insights. His research focus has been in natural language processing, intelligent information systems and human-computer interaction.
How did you catch the entrepreneurial “bug”?
A lot of my entrepreneurial life — like a lot of my research life — is tied to my collaboration with Kris Hammond (McCormick professor and Narrative Science co-founder). We'd known each other in graduate school and had worked together there. When I first got to Chicago, our first entrepreneurial collaboration as grown-ups was thinking about recommendation technology and how to move it forward.
From my point of view, especially if you're in technology, there comes a time when the standard academic model of “I’m going to write this next paper, and I'm going to have a great student” begins to seem a little bit inadequate, especially with all the revolutions that are happening around us. You see technologies that were being talked about 10 and 20 years ago finally getting out into the world, and a couple things come to mind: why am I not part of that? And why can't that happen faster?
For example, 2011 was a miracle year for artificial intelligence. That was the year that Watson became the world champion on Jeopardy, and it was revealed that Google had cars driving around America, and finally Siri — millions of people are carrying on useful conversations with this technology every day. That all happened in one year, but if you ask people in the business, I think we all expected it to happen 15 to 20 years earlier. I wanted to be part of the process of making it all happen faster.
What advice would you give to young colleagues interested in business?
Feel free to fail. In this business [software], it isn't that expensive to fail. It's not like some other fields where you flush $10 million down the drain. In this case, you might have wasted part of a year and half a million dollars or something like that. The question is, how can we make the culture more — not just failure-forgiving, but failure-celebrating?
View Larry Birnbaum's faculty profileWhen you fail, it's actually kind of liberating, because once you fail and you see that it's not that painful, you go, “Yeah, I can do that again.” The first failure is the hardest. Plus, you really do learn so much. It wasn't a waste of time. I think that's the biggest worry, even more than the money: the waste of time.
Some people worry that if they take money from angel investors and fail, they will never receive money again.
I don't think that's true. They will. First of all, you can't overpromise. Of course, when you start a company, you have to live with the fact that there are constraints. The minute you take other people's money and you have investors and a board and you're responsible for other people's money, you are not a free agent the way you're a free agent in the academic world. Let's put it this way. It is not a research lab.
Have you found that timing is an important factor in success?
I don't think you should be worried about timing. The thing you're building may not be ready for a year or two, and you would have to predict where you think people will be in that time. A lot of it is luck.
Were you lucky with the timing for Narrative Science?
The timing was wonderful. Also, the idea is immediately attractive to people, but most important is the rise of big data. All this data is being gathered, all this data is being analyzed, and we need to convey it to people in a way that’s meaningful to them. It has potentially very high value, very broad applicability.
That also causes problems, by the way. One of the problems with Narrative Science is that this technology has such broad applicability, and we're a startup. If we were Google or IBM, we could say we're going to put 20 engineering teams on this, and they're going to try all possible market applications. But we have to pick a couple of spots to focus on. It’s kind of painful sometimes. It takes some of the romance out of it.
What about starting a company without investors?
Kris and I did a couple of small startups where we didn't hire a CEO — we did it ourselves, the whole thing was virtualized. None of those ever worked. They weren't as painful as the bigger failures, though, and I think I did learn something from them.
If you're afraid of a big financial failure, that might be a good way of getting your feet wet. Start it when you have a client, when someone comes to the lab and says, “I would actually buy this.” Then the amount of money would be $50,000 or $100,000 — whatever they'd be willing to pay. I can actually build a prototype for a customer for that amount of money if I do it on the cheap. We used the Small Business Opportunity Center at the Law School to do all the legal work. They incorporated the company for us, and they were cheap, cheap, cheap, cheap, cheap. It was like a hundred bucks.
You have to start with a customer in that case, because you still need some money. And don't use your graduate students for that — that'll get you in trouble. In the end, if this customer became a repeat customer and it looks like there is really a business there, that's the moment when you would scale the company and get professional management.