On his first day of work at SpeechWorks International, in 1995, Matt Marx was greeted with the usual stack of forms to fill out. Amidst the healthcare paperwork was an agreement that, if he ever left the Boston start-up company, he wouldn't work for a competitor for two years.
"I was a naïve college student," says Marx, who had just earned a graduate degree from MIT. "I didn't want to put up a fight."
Adding his John Hancock to the two-year noncompete agreement - and the effect it would have on his career options - later led Marx to investigate how the contracts can restrict employees' mobility, which is exactly what they're designed to do.
Marx is now a doctoral student at Harvard Business School, and his research is being cited as part of a campaign to eradicate such agreements in the Bay State.
In 1999, Marx left SpeechWorks to enter an 18-month MBA program at Harvard. Midway through the program, he realized that the lengthy noncompete agreement he had signed left him with two choices once he finished the program: either go back to SpeechWorks or go into another industry.
But a third option presented itself when Marx was recruited by another speech-recognition company, TellMe Networks, of Mountain View, Calif. Courts in California and nine other states, as a rule, don't consider noncompetes enforceable. So Marx decided to leave Harvard to take the job with TellMe.
Nearly five years after he moved west, Marx boomeranged back to Massachusetts, mainly because if he didn't finish his MBA, his course credits at Harvard would have expired.
As Marx was wrapping up his MBA and thinking about earning a doctorate, he was approached by one of the founders of SpeechWorks, Mike Phillips, who wanted to know if Marx might do some consulting for a new start-up he was cultivating, Cambridge-based Vlingo Inc.
Interestingly, even though noncompetes are considered unenforceable in California, TellMe had asked Marx to sign a one-year agreement - a fairly common practice - which could have been enforced if he chose to start working in Massachusetts again.
"I told Mike that I had to wait that out," Marx says.
But Marx started consulting for Vlingo after the prescribed year had passed.
Noticing that the legal environment surrounding noncompetes in California had made it easy for him to take a job in Silicon Valley - but that the enforceability of the agreements here had hampered his return - Marx decided to dedicate his doctoral research to the topic.
Marx found that one state, Michigan, had reversed its policy on noncompete agreements through what seems to have been a legislative oversight.
From 1905 to 1985, noncompete agreements were illegal in the state, but in 1985 Michigan reversed its position.
Using patent records to study the job-hopping patterns of inventors in Michigan, Marx found that, relative to other states that hadn't changed their stance on noncompetes, inventor mobility dropped 19.8 percent in Michigan after the law was changed.
For more specialized inventors, the plunge was even more pronounced - as much as 32 percent.
"What I wasn't expecting to find was that really specialized people - a nanotechnology engineer, for example - get bitten the hardest by noncompetes," Marx says.
To many employers, that's great news: The data clearly show that being able to enforce a noncompete makes your most creative employees less likely to jump ship.
But keeping inventors and entrepreneurs frozen in place is detrimental to a region's innovation economy.
Ideas that are stifled, or don't fit, at one company can often create incredible value elsewhere. Intel was founded by refugees from Fairchild Semiconductor; closer to home, Data General splintered off from Digital Equipment Corp., whose chief executive, Ken Olsen, "chose not to enforce noncompetes," Marx says.
It isn't hard to quantify how many smart, frustrated people throughout Massachusetts would leave their current jobs and start a company - or take their talents to a company where their ideas would be embraced - if taking the leap didn't require taking a year's unpaid vacation, or running the risk of being sued.
Attorney Gabor Garai says the impact of noncompetes is like an iceberg: A few cases attract media attention. But those cases, says Garai, a partner at Foley & Lardner LLP in Boston, are just the tip of the iceberg.
"The bulk of it is under water," he says. "That's where you have all the people who don't even try to leave or try to start a company because of their fear of being sued."
The partners at Spark Capital, a Boston venture capital firm, began a campaign this month to get rid of noncompetes in Massachusetts. They sent a letter to Governor Deval L. Patrick in which they predicted that the result would be "more start-ups originating in the Commonwealth, a reduction in the exodus of talented people, and the ability for Massachusetts to better compete nationally and globally as a hub of innovation."
But a legislative campaign to make noncompetes unenforceable, even if it gains momentum in the face of almost certain resistance from the state's big employers, could take years.
In the short term, employees can get better educated about noncompetes and ask about the terms before they accept a job. (The two-year noncompete that Marx signed with SpeechWorks is longer than usual, but not unheard of, lawyers say. )
Employers who believe noncompetes aren't the best way to foster employee loyalty can simply toss such agreements overboard.
Google's Cambridge office hires people without chaining them to noncompetes, and Spark Capital no longer asks its portfolio companies to have their employees sign on the dotted line.
Companies willing to try to retain their employees with incentives rather than threats could start winning the talent wars, forcing more companies to give up on noncompetes.
As for Marx, he says his next research project is focused on whether enforcing noncompetes in a state can spur inventors, engineers, and entrepreneurs to move elsewhere to seek employment - as was the case in his own career, when he went to Silicon Valley.
Scott Kirsner can be reached at firstname.lastname@example.org.