Carnegie Mellon Reels After Uber Lures Away Researchers
on.wsj.com
on.wsj.com
However, salary for tenured professors is guaranteed regardless.
So, if you are a tenured professor, you can choose to do esoteric research that will not get a government grant, but you won't have any grad students.
It actually works out because the less practical, more theory oriented profs do not really need grad students. Whereas the more practical, application oriented profs need grad students to do all the work and they end up "setting up shop" like little companies---and get grants to pay for the students (mainly tuition, also a small stipend).
Not that I'm defending it, but it has a kind of internally consistent logic.
EDIT: Being innovative doesn't mean "at all costs".
Two examples:
When I see news about Tesla, I have a positive reaction. They constantly release new technologies, open sourced their patents, and so on.
When I see a news piece about Uber, I think, "What'd they fuck up this time?".
News pieces about how Uber was attempting to encrypt/destroy data remotely in their Canadian office while a law enforcement raid was in progress......does not help their case.
https://news.ycombinator.com/item?id=9606773
However, other comments suggest that the lab wasn't true academia anyway, but an applications-only wing that is non-profit in name only.
My impression is there's a difference between luring away researchers with good salaries, strictly according to their talents, and effectively acquihiring an entire team (but without the acqui bit). They're paying not just for the skill and expertise of the individuals, but for the institutional memory and project state of the research group.
Acquihires happen when the team is well motivated to remain but another company has deeper pockets. It's either cheaper to poach staff or buy the company and it can be a tough call which it is.
It also feels like academics are a bit inconsistent here. Universities are run like any other business but act as though they have a special position as dispensers of education.
Most companies would go through a full acquire and at least try to soft-land anyone they don't have a use for. Uber has made clear time and again that it has zero interest in individuals that it cannot use.
Also, probably the most important contributor in CMU SCS becoming what it is now was an act of "poaching" (of the Blums, from Cal).
While it upsets him that he invests both time and grant money toward students who don't finish the program, he understands the lure of a 7-figure salary for people who are living on a low 5-figure income.
Maybe, just maybe, instead of using grad students and post-docs as slave labor so he can slap his name on their hard work, he could instead advocate for better pay and working conditions for them using his privileged position of PI?
As far as slapping his name on their hard work, it's a symbiotic relationship. He works with his grad students individually on their research: guiding them away from dead-end areas, provoking them to dig into certain areas while providing references as starting points, helping them with the fine points of writing up their research, and using his contacts and reputation to secure academic positions when they graduate. In turn, yes, his name goes first on some papers. If he did nothing to help them, I'd say they have a legitimate gripe. In his case, he's available to them all the time.
"So basically, the principles of economics upset him, but he understands them."
The "principles of economics" are not the issue. His program is sought after, so he had to turn someone down in order to fund the person who decided to leave. Yes, it upsets him.
1) I have (at least) weekly meetings with advisors who can charge >$250/hr as a consultant. They are brilliant, and our conversations are enlightening, and they actually care about the outcomes of their students. This might be out of the ordinary, but I've never felt exploited by my advisors. In CS, advisors are very aware that their students are giving up significant salaries to do a PhD.
2) I get paid enough to live, not incredibly well, but I make a median salary in the area, and at this point in my life I'm more than willing to share a 2br 2ba apt.
3) I'm encouraged to do internships which do pay extremely well, and on top of my living wage at school, mean I live really well.
4) I want a job that requires a PhD. Sure I could drop out, take my masters in AI, brand myself as a data-scientist and get a job doing some ML programming at a hedge fund making 300k base with 1MM bonus, but working at a research lab, or at a tier 2 institute as a faculty member seems like a much better life in the long run.
The Lawrenceville neighborhood, where both NREC & this new Uber office are located, has recently become one of Pittsburgh's trendiest, and has seen skyrocketing real estate values— which means houses that used to cost $30,000 or less are now valued at over $100,000.
Is that what they were doing? Because I initially thought "what a waste... a lot of public research will now be locked up in one business's competitive advantage" - but if they were just building stuff for the military that would get locked up for decades anyway, this is better.
But that research is generally privately-funded and privately-owned, too. NREC is not---in general---a hotbed of publicly-published common-use research.
Only if CM replaces them. If they decide that the robotics lab is unrecoverable then the jobs will simply transfer from CM to Uber which has no direct benefit to Pittsburgh at all.
It seems like Stanford has a better way of managing these type of deals.
I think half is a bit of an overstatement in the typical case, but this same criticism can be levied at pretty much any research- or cause-oriented work, like charities and foundations.
Usually this cut is intended in part to fund the process that actually procures this funding and these grants in the first place, so under that assumption, it more than pays for itself.
You can't run an organization with literally zero overhead; it's not possible.
Nope, it's pretty typical. Most grants have a negotiated ~50% overhead in "indirects". I was surprised, though, to find a Nature report[1] that found the actual overhead was typically ~20% below the negotiated rate. Which is pretty wonderful.
1. http://www.nature.com/news/indirect-costs-keeping-the-lights...
A "doubling of salary" sounds nice, but it's still important to know the base. "Loyalty" to an institution is almost certainly an old-fashioned concept and probably a sucker's bet, but my sense of sympathy for CMU will still be somewhat dependent on the "pre-uber" numbers.
For instance, if you lose a post-doc earning 55k and uber now pays them 110k, well, it's time to stop complaining. This is a person who managed to get high grades in an very tough undergrad major with high attrition rates, gain admission to a top PhD program (almost certainly requiring high standardized test scores), and made it thorough a doctoral program with a 35-50% attrition rate (elite med schools, by contrast, typically have attrition rates below 0.5%).
Now, if these salaries were going from $250K with tenure to $500k, I'm a little more sympathetic to CMU (though even then, I don't begrudge these researchers their new and good salaries).
http://techcrunch.com/2014/09/02/google-partners-with-ucsb-t...
The Uber thing sounds like it is more of a complete transfer from CMU to Uber.
So yes, this is very similar.
That said, I think Uber will fuck themselves hard, and that Uber's robotics efforts will be viewed as a massive mistake within the next ten years. This is a situation where they should buy or borrow, but they're trying to build out of greed.
Autonomous driving technology is not going to be a differentiator. It's going to be a commodity. Everybody will have it. Investing heavily in a commodity product is a waste of money.
But it's exactly the sort of mistake that one could expect an egotistical asshole like Travis to make.
Uber won't be dead in ten years; they'll be the myspace of transportation. I guarantee it.
Sure, they are within their rights to get these employees, but it's worth considering whether they should think about the consequences for CMU. There's a lot of organization memory that makes a lab special, especially this particular lab.
If I start a medical startup, I'm more than within my rights to hire some doctors. But would I go and hire half the staff of my local teaching hospital? I think I'd think twice.
If your point is that society misses out on the positive externality of the knowledge they publish (which they mostly won't do at Uber), then the university or a nonprofit should work out an agreement with Uber to buy that info from them, e.g., by covering a portion of the scientists' salaries.
Of course. This is the economically naive explanation of why people are paid whatever they're paid (marginal product). It works fine for things like soccer players, where it's quite measurable whether some guy is contributing, how many shirts he sells, how much he'd add to a given team, etc. In this situation, it's not that clear. Uber have money, but that doesn't mean those people are suddenly worth more. They may make something useful, they may not. Having more chips doesn't make you right, it just wins you the hand.
>If your point is that society misses out on the positive externality of the knowledge they publish (which they mostly won't do at Uber), then the university or a nonprofit should work out an agreement with Uber to buy that info from them, e.g., by covering a portion of the scientist's salaries.
Right, and did they do that? It sounds like they just said "meh, let's not bother and just use this pile of cash to get what we want". Maybe it's more complex than that, but it doesn't sound like they care too much.
The question is: do the non-profits actually care about societal benefit, or are they really just managed by people interested in building prestige? CMU robotics wouldn't get nearly as much prestige from paying Uber scientists to publish papers.
A quick look at Glass Door shows that C.M. "post doctoral fellows" are being paid $55K/year. I don't know what the robotics guys were getting paid but it's not hard to believe that they would jump at the chance to get paid what a software engineer gets paid.
It doesn't seem fair to expect C.M. robotics engineers to work for half their potential pay.
It isn't like going into the priesthood where there is an expectation of sacrifice. The presidents of these institutions are certainly not setting an example of sacrifice (nor am I saying they should). [The president of C.M. gets over $500K a year.]
And then ask for an even bigger budget the next year.
In contrast, it is not a moral indictment of Uber.
For every $10 million you let remain in the private economy (instead of shifting to academia), you will get the same or more innovation.
IBM, Intel and Google have contributed far more to computer science than academia ever did, and the Internet would have happened either way.
I am (temporarily) in academia, and I can tell you, it truly is a Soviet-style allocation system.
Of course if you throw enough money at something with a ton of smart people you will get breakthroughs, but in general, it is corrupt, crappy, and only marginally honest.
I doubt that. The only way that something like the Internet could have been created is if a 'third party' of industry created it, which in this case is academia and DARPA funds. Before the rise of the Internet, there were private paid networks that were restricted and never talked to competing networks. The Internet destroyed those networks and now we get our Internet access (mostly) raw.
Also, do you realize that most of the basics and rise of computer science comes out of academia? If a Google or Microsoft could have patiented basic computer tech, they would have, stunting the growth of the field for 50 years or so.
That's not a justified conclusion.
If government academics had created airplanes or the internal combusition engine, people would say that was the only way they could have come about.
> Also, do you realize that most of the basics and rise of computer science comes out of academia?
A lot of the basics of computer science really did come from IBM.
> If a Google or Microsoft could have patiented basic computer tech, they would have, stunting the growth of the field for 50 years or so.
Patent laws are a government creation, so they're the same basic kind of thing academia is. They are not a fundamental fact about the way human societies and economies work. So we can't take that for granted and then use it as a premise to argue against something else.
(I acutally do support certain kinds of patents, though, and the effect of patents is usually the exact opposite of what you say here.)
After all those engineers and scientists don't belong to CMU or the NREC.
Let's say I want to open a new burger joint and I know there's a really good team in a kitchen down the road. Am I smart or a dick for hiring away that team at double their previous wages?
If this were Uber hiring engineers from Google or Facebook and paying them a huge premium, it would be considered the market at work and I expect there would be little debate on whether Uber was acting ethically or not (even when considering the tactic of opening an office nearby).
Right now it feels Google, Uber and Lyft will compete for the same space. Uber looks to be winning, but can it really win the race against Google.
Although to its own merit one thing comes to mind was that -- at least from PR it looks like it -- Uber is moving faster. Google which is always hiring(?) lost the chance to grab that CMU talent.
But overall, this will be good for the ecosystem both in terms of raw research project (more people working on a hard problem, though they aren't working together so we're not getting theoretically maximum benefit) and economics (more competition in the space as these projects do flourish into commercial projects).
In the end, this sort of thing should get the product to market faster than Google working on it alone, which will be saving lives sooner. Strict positive.
Honestly, the only thing I'd be concerned about is working for Uber itself. They're a company that has made it clear on multiple occasions that the moment someone is no longer valuable to them, they'll do everything in their power to zero out their responsibilities to that individual. Even if the triple-up-front signing bonus is a good deal now, I hope the researchers they hired aren't thinking Uber is going to be a career for them...
http://newsroom.uber.com/2015/02/uber-and-cmu-announce-strat...
My (uninformed on the particulars of this situation) guess is that if the strategic partnership has 'devolved' into a poaching situation, CMU shares its own bit of the blame. There's a lot that went on behind closed doors that we will likely never know.
disclosure: CMU alum as well.
I'm looking around the NREC web site. At the top, it is "research professors," with similar abusive quasi-academic positions. Maybe this will give some incentive to move from short-term grant-funded or teaching-funded positions to workable long-term faculty positions.
The pattern of decades of post docs is not workable. People want families. There is a biological clock (women, but also men if they marry similar age).
The NERC response is a little bit arrogant. Jeff's quote on Uber coming to them -- well, Uber hired short-term university employees into long-term positions. They did with those positions exactly what universities claim those positions are for (training ground for stable work). Jeff should be applauding them.
It'd be wise for them to consider this, if anything, a side-move (but a side-move with a 3x salary bonus, so definitely take that money and bank it!).
Neither is stable, but of the two, almost any industry, even including Uber, is much, much more stable.
Say what you want about Uber, but universities are clearly 1000x morally worse.