Hacker News Reader

Comment by rmk | Hacker News Reader

For people who are familiar with this, I have a question. How does research in an industrial lab vs. a university work, insofar as independence is concerned? In a university, aside from well-defined ethics guidelines for conducting research (e.g., involving animal or human subjects), staff are generally free to publish the findings of their research at any venue they see fit. Many are tenured professors, so they also have the added guarantee of tenure, which makes them unfireable save for clear violations of policies such as sexual harassment.

How do these things work at industrial labs, such as those of Google and Microsoft?

View on HN

> In a university, aside from well-defined ethics guidelines for conducting research (e.g., involving animal or human subjects), staff are generally free to publish the findings of their research at any venue they see fit.

I'm not sure what university were you at where that was true - academia is horribly political (in many ways more than corporations in some ways less) and until you get tenure your job is to put up and shut up. Your career is dependant on good recommendation letters and goodwill of quality advisors.

Attacking your organization, supervisor or lab will end your career faster than Google cancels products.

Even after you gain tenure, you're still dependant on external funding to keep your research going and employ postdocs and other researchers. That funding disappears really quickly if you decide to start an assault against your organization or people paying you. Like these Googlers did.

>Your career is dependant on good recommendation letters and goodwill of quality advisors

life pro tip: for any paper with more than 2 authors, the last author is the most senior and did the least amount of work (if any). i leave it to you to conjecture about quid pro quo.

The last author convention is not used in all fields.

All of what you said is true, but the meaning of tenure is that you are paid a salary by the university and get to keep your job. That's a very generous protection and it's offered to encourage free discussion and exchange of ideas, which is what the University is all about. There are professors who choose to ditch research and make teaching their calling, even at research universities.

Universities are not very well-run when compared to your average fortune 500 corporation, and there is a lot of skullduggery that goes on inside of academic departments, particularly those that do not receive a whole lot of research funding, such as the humanities. The path to tenure is brutal and is full of pitfalls (in fact many positions are explicitly not tenure-track; there are many, many adjunct positions where folks are paid a pittance and forced to sign very unfair contracts that must be renewed periodically). Even with all that being said, academics do enjoy a degree of freedom that those in the corporate world can only dream of.

MSR has a very free research culture. People there are technically fireable by normal means, but Microsoft chooses not to terminate employment for "outside-of-work" engagements, or for the subject of their research. Some very critical papers of Microsoft's AI approach were published under the MSR header. Not all of the findings were put into place, and Aether has the unenviable job of balancing ethical behavior with potential profits. But their ethicists are leaving because they're shouting themselves hoarse into the void, not because they're being asked to change their results.

You are talking about the Google ethicists who are leaving, right?

Also, can you clarify what you mean by 'shouting themselves into the void'? Are they being denied venues to publish their research, thereby rendering their shouting inaudible, or did you mean something else?

I believe they meant that they were allowed to publish critical research, just that it was not acted on.

In other words MSFT allowed them to publish freely, it just (allegedly) ignored the implications of the research.

I have a lot more respect for MS and MSR than I used to. I grew up, and I learned more about both. Nowhere is perfect but, in a different context of the US attempts to define extraterritorial control on data held by European hosted services, MS was the company which stood up to the department of state. Google and the others folded.

I am not sure I necessarily buy your comparison. Do you have any records where the employees have done something similar, e.g., giving ultimatums to manager, sending inflammatory emails to a large group of employees, leaking internal documents to outsiders, in MSR, but didn't get fired?

Anyone interested in working in a research position should care a great deal that their management hierarchy is separate from their research hierarchy, and that the organization supports researchers working across teams. That will reduce the impact of personality conflicts and assholes.

I know that some places feel that the people managers have to be the technical experts but I would strongly advise against working in such a place.

Apparently the way it works at Google is they fire you when you don't rubber stamp everything.

I hope you are using hyperbole here. Google employs hundreds of researchers who are free to publish their work as far as I can tell (plenty of seminal systems, AI, economics -- auction mechanisms and such, and other papers are testament to this fact). There appear to be some exceptions to this that are causing a furore.

Reply on news.ycombinator.com