Banazîr the Jedi Hobbit (banazir) wrote,
Banazîr the Jedi Hobbit
banazir

  • Mood:
  • Music:

Microsoft versus Google and Kai-Fu Lee: Repercussions for IT Jobs

Disclaimers: The opinions expressed in this entry are those of the author, William H. Hsu, only. The factual statements cited in this article are verifiable through the mainstream IT media. Relevant articles, both factual and editorial, are linked where feasible.

Some of you know that KFL, the former Microsoft exec who left for Google and whom MS immediately sued to keep him from starting Google Labs Asia, is my uncle (my mom's younger brother). I've been sitting on my hands somewhat regarding the controversy surrounding the case, owing to the probable perception of partiality that would color any comments I made. Recently, however, I've come across some misconceptions that I thought I would correct here. I also found some relevant articles:

  • CNet (Ed Frauenheim, 26 Aug 2005): `Buried in Microsoft's lawsuit against its former executive Kai-Fu Lee and Google is a legal doctrine that could make tech professionals shiver... in court filings, the software giant has also mentioned the theory of "inevitable disclosure," which holds that in some circumstances people can't avoid sharing or relying on trade secrets from their former employer when moving to a competitor.

    Thanks to this increasingly popular legal argument, techies and other employees could be in for a surprise when they try to switch companies. In states that accept the inevitable disclosure concept, employers can sue defectors even if they've signed only a confidentiality agreement--or even if they haven't signed an employment agreement at all, said Robin Meadow, an attorney with the firm Greines Martin Stein & Richland.'

  • Slashdot (28 Aug 2005): In a discussion thread on Frauenheim's article, readers compare the ramifications of the suit to a new species of feudalism and preemptive lawsuits in Jennifer Government (the inspiration for NationStates). Arguments range from "never sign a noncompete agreement" (patient: "it hurts when I do this"; doctor: "then don't do that") to "the employer should pay if it expects blanket coverage from a noncompete". A couple of cogent points were made several times: that most IT professionals have most or all the liabilities of a noncompete (not being able to work for a competitor on a broadly-defined or undefined set of R&D topics) and few or none of the benefits (the new company being willing to pay you for a year without your being able to work on what you were hired to do). Thus the potential downside if the court compels KFL not to start is that IT professionals may find it harder and harder to move to a new company.



I have a few points to add to these.

First, as some readers pointed out, the precedent that the case would set, enabling companies to chain down employees using non-disclosure agreements (NDAs) alone, amounts to a kind of indentured servitude, which is illegal when it comes down to it.

Of course he didn't have to sign it (not unless he wanted the job back in Redmond when he came back from MSR Asia). Of course he read it. Of course in 2000 the groups he had responsibility for were somewhat different than the ones he had in 2005. Think of it as effectively thus: a company hires an exec, sends him to his ancestral country to start a lab, brings him back, and has him sign a non-compete agreement then. Five years later, the company (purportedly) divulges info on beating competitor X to its senior management, including the now-promoted exec. Exec now has privy information. Some of the hundreds of people in this exec's groups are working on the X-Killer app. Exec now has privy information and is "in charge of X"; oh, noes! Now exec goes to X. Is he automatically in violation of what he signed?

Substitute your top company in your favorite niche market for X. Sony and PlayStation, Skype and VOIP, et cetera ad infinitum et ad nauseam. Eventually, you might see that non-compete agreements are more nebulous things than those who live by them would have it, and the breadth of interpretation enters very much into it. Only a Sith deals in absolutes, as they say.

Second, a lot of people are painting a thick line between NDAs, which don't say "I won't go work for your competitor", and non-compete agreements (NCAs), which explicitly do. "You signed on the dotted line and got paid for it." The reality of the matter is that:

  • 1. In the general case, NDAs have thus far been actionable only if already violated.

  • 2. Companies generally don't go after NCAs unless an employee is poached for specific knowledge, technology, or unique skills covered by the contract.

  • 3. In this case, as Frauenheim reported from the court filings in his August 2nd article, the crux of the debate is whether what KFL would do at Google, and in founding the facility in Beijing, China, actually overlaps enough with what he did at MS to count as an impending NCA violation.


Put these three together and you can see what's going on. It's a legal dragnet for MS: first, they claim it does overlap enough; if that doesn't fly (which it actually doesn't), they have the safety net of "inevitable disclosure" with which to muddy the waters. So, something that should concern you if you are a CS researcher, or indeed an IT professional of any kind, is that companies such as MS are willing to set up legal precedents that could mean NDA == NCA, just in case the things on your personal no-fly list aren't enough to keep you out of the enemy's camp.

Third, a lot of people don't seem to fully grasp that like any competition, hiring away R&D people is itself a legal contest between the old and new company where the NCAs are the playing field. The new hire's job description are the pieces. The old company inflates the job description as much as it can convince the court to accept and has to demonstrate that this violates the NCA, and the new company defends against these allegations. Both sides can offer claims about what the defector will actually do. Because it's essentially trying to prove a negative, noncompetes are not gone after often; sometimes they are upheld, though, because the terms are of limited duration.

Fourth, some people who aren't reading the news (or history) carefully enough keep saying things like:

  • "He went to work for the identical department at the competitor." Um, no. This is called believing everything you read. KFL had several groups under him, speech rec being his primary area and his primary responsibility. Google's claim is that the actual responsibilities and technical knowlege of MS's search technology that KFL had were of tangential relevance to its search strategy. The bar is pretty high on what consitutes the same technology.
    Being a machine learning researcher, I'm aware of the relationship that ASR and NLP in general, have to search. Nobody's claiming they are disjoint. But "exactly the same"? Learn about the technology and judge for yourself.

  • "Oh, it'd be {hard | easy} for KFL to find another job." KFL wants to work for Google, and Google wants him. Nobody has ever denied that, to my knowledge! Much as I'd like the comments implying that Google scooped him up after he went out the door to be true, KFL did leave MS to go to Google.

  • "He's privy to MS's confidential strategy!" Repeat after me: knowing something confidential doesn't violate NDAs; disclosing it does. Similarly, one has to disclose or act on privy knowledge to violate an NCA. The reason NCAs are actually broader in some sense, and say you just can't go work for the competitor if you're going to be doing the same thing, is precisely because the presumption is that in that case, you're going to use that privy information, advertently or inadvertently. So, the court provides for a "reasonable waiting period" for the old company to revise its strategy, or (more likely and more effectively in the internet search industry) for the highly dated insider information to expire.

  • "So he takes a vacation for a year - so what?" Yes, Google's willing to pay him anyway. No, my cousins won't go hungry. So what? As I started to explain above: A year is a very long time in search. The problem is that KFL wasn't just in charge of search, he wasn't the only one in charge of search, and most important, he wasn't mainly doing search. So Google's argument is that holding him for this waiting period isn't reasonable because managerial and administrative tasks such as hiring for the new lab are going to go without KFL's leadership in the meantime.

  • "Who could possibly be worth all this legal hassle?" You'll permit me a bit of family pride: KFL is. Objectively speaking, as a CS researcher, I can tell you that his reputation and his track record are really that strong.


There you have it: my $0.03 worth.

Thanks for reading, and please feel free to post relevant links here, respond with your opinions or questions, and forward this link wherever you like.

--
Banazir
Tags: google, kai-fu lee, microsoft
Subscribe
  • Post a new comment

    Error

    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.
  • 45 comments