Why Uber Shouldn’t Have Poached CMU Robotics
Earlier this year, Uber went on a shopping spree at Carnegie Mellon’s robotics center. They hired away roughly 40 lab members from CMU, presumably to build the core of a cutting-edge driverless cars division. This was a bad idea that must have looked a whole lot like a good one.
Tactically, Uber’s move was a coup. In one fell swoop, they bought themselves a brand new R&D division, positioning the company to be at the forefront of the coming driverless revolution, as opposed to just letting it happen to them. The next 5-10 years will be good to Uber. From that perspective, I’m sure management was cracking open the champagne.
Strategically, this is a disaster for Uber. As the NYT story indicates (corroborated by my own conversation with a recent CMU faculty), Carnegie Mellon is not happy. Normally, this would be a good thing. If Uber stole a bunch of robotics guys from Google, sure Google would be pissed, but for Uber it would be an unmitigated victory. But this is the problem – not only is Carnegie Mellon not a rival, they were a potential partner.
There are very few things we know about best-practices in R&D, but one of them is building good relationships with universities. These might be formal, like attending/presenting at conferences, but they are frequently informal, like random chats over coffee. Ideas get exchanged, scientists get more productive, and companies invent cool stuff. Everybody wins.
Uber has not only sacrificed this kind of relationship with Carnegie Mellon, but it has also jeopardized its good standing with the academy as a whole. Academia is a small world, and specific fields like robotics are even smaller. Everybody knows everybody, and you can bet that most academics sympathize with CMU here. Academic circles already mistrust industry, given the latter’s typical focus on short-term profits over long-term, universally beneficial scientific advancement. Uber just showed why that mistrust is so well-placed.
Uber may have scored a coup, but history tells us that coups don’t usually turn out well in the long-run.
What Uber Should’ve Done
Criticism is easy, so let’s offer an alternative strategy. Was there a way for Uber to build a cutting-edge, robotics R&D division without sacrificing potentially valuable relationships with CMU and academia more broadly?
One potential route: Uber might have instead pursued a semi-formal recruiting partnership with CMU, thus providing it with a steady stream of roboticists. I can see three benefits over Uber’s poaching method, along with three downsides.
- This strategy is replicable across other universities with strong robotics programs (e.g., MIT), which both diversifies the intellectual crop (e.g., suppose the CMU approach is a scientific dead-end that needs to be weeded out over time), and boosts the speed at which you can build the division.
- This makes hiring easier in the future, since the relationship with CMU (and academia) is not at all damaged. So if Uber intends to hire more than this initial batch of 40 roboticists, it will find it easier to fill out its ranks with this strategy than with poaching.
- This strategy also builds up the relationship between Uber and CMU in a positive way, which should make that idea-exchange channel even stronger than it would have been absent any strategy.
- This is a decidedly slower method, so if speed is absolutely paramount to Uber in building this division, the added benefits would be all but moot points.
- This requires considerably more HR time and money, which Uber may not wish to spare.
- Getting a 40-person team, all at once, also buys their existing relationships. To the extent that these 40 individuals are already highly complementary assets, Uber is sacrificing some productivity in bringing in a stream of roboticists who have to become complementary (as, e.g., they get used to working with one another)
Ways in Which I Could be Wrong
Of course there are many, but let me try to identify a few:
- Uber values speed in building this R&D division more than the relationships it would have built with CMU/academia
- Uber’s relationships with CMU/academia are actually not that damaged, or at least the repair will not take that long
- Those relationships with academia are not actually an important channel of idea-exchange, innovation, and ultimately value-creation for a company (distinct from Uber’s valuing of that channel; it may just objectively be not-that-important for any company).
- CMU is, in some way, a rival to Uber. If not directly, then at least as a potential supplier of roboticists to Uber’s competitors. To the extent that this is true, and to the extent that CMU is now not more inclined to give preference to Uber’s rivals in access to and hiring away their roboticists, Uber has struck a preemptive blow against its competition.