Like a lot of silicon valley rich folk, John Koza is bewitched by numbers. Unlike other Silicon Valley rich folk, though, Koza has applied his numeromania to a singular obsession: the U.S. Electoral College. Yes, the Electoral College, that soul-deadening fixation of generations of droning civics teachers. Koza knows everything about it, can talk about it for hours, and, if he has his way, will do away with it altogether - or at least render it as useless as a sixth toe. "I guess you could say I'm an Electoral College hobbyist," Koza says.

But not just a hobbyist. Koza, who earned millions from the development of ruboff lottery tickets, is also a professor of biomedical informatics at Stanford, a holder of 25 patents in fields ranging from genetic programming to videogames, and a venture capitalist. In Silicon Valley, hobbies often quickly become vocations - that's one of the secrets of the Valley's success - and Koza has spent well over $1 million of his own money to realize his dream of upending the electoral system. With a pair of colleagues he founded National Popular Vote in 2005. It's a nationwide nonprofit that champions Koza's ingenious scheme to end-run the state-centered Electoral College and allow Americans to elect their President by - hence the name - a national popular vote.

To make room for his hobby, Koza says he has cut back his VC activity to one startup a year. Now he devotes three-quarters of his time to getting his plan enacted. Against all odds, it shows signs of catching on. Koza has had bills introduced in 47 states. Thanks to his vigorous lobbying, four states have enacted the plan so far, including such electoral-vote behemoths as Illinois and New Jersey. Another 18 state legislative chambers have passed it, and it has fetched endorsements from the Boston Globe, the Los Angeles Times, the Hartford Courant, and the New York Times, which called Koza's idea "a truer and more certain path of democracy."

The scheme does have a certain comprehensive elegance - like an algorithm that solves several problems at once. The big problem with the Electoral College, Koza says, is that it's undemocratic. Every state but Maine and Nebraska awards its electoral votes on a winner-take-all basis to whomever a majority of its residents voted for. If you were unlucky enough to have been awake in 2000, you know what that means: A majority of voters nationwide can vote for one presidential candidate yet see the other candidate declared the winner by the Electoral College.

In 2000, Republicans thanked the election gods for the arrangement even as Democrats wailed, but 2004 must have given them pause. That year a shift of a mere 60,000 votes in Ohio would have awarded the state's 20 electoral votes - and the election - to Democrat John Kerry, even though President Bush beat Kerry by 3.5 million votes nationwide.

Even so, Koza encounters a lot of skepticism. "I'll be lobbying," he says, "and I'll go to a state senator's office, and they all say the same thing: You've got to amend the Constitution to change the Electoral College!" He smirks slightly, bemused by the depths of human folly. "I say, Show me where in the Constitution it says states must apportion their electoral votes by winner-take-all. It's not there."

This basic error has led most Electoral College opponents to struggle vainly to pass a constitutional amendment - a cumbersome, nearly impossible task. Koza's idea is much more plausible, rooted in the reality that the Constitution empowers states themselves to decide how to allocate their electoral votes and to whom.

So NPV would work like this: A state government passes a law requiring it to award all its electoral votes to whichever candidate wins the nationwide popular vote. The law would take effect only when other states representing a majority of the Electoral College had passed laws making the same commitment. After Election Day the committed states would combine to award a majority of electoral votes to the candidate who got the most popular votes.

It's simple in outline, but the ramifications are elaborate enough to have inspired Koza, along with five co-authors, to write and self-publish a 619-page NPV guide called Every Vote Equal. It's not quite as complicated as his four earlier books on genetic programming. But close.

Koza's peculiar hobby first showed itself publicly in 1966, when as a college student he invented and marketed a board game called Consensus, exploring various mathematical permutations presented by the Electoral College. It was a commercial flop. Imagine that.

"It was way too complicated," he says now. "Graduate students loved it. But no one else did."

After getting his Ph.D. in computer science at the University of Michigan (thesis title: "On Inducing a Non-Trivial, Parsimonious Grammar for a Given Sample of Sentences"), Koza co-founded Scientific Games (SGMS), a company that patented and popularized the ruboff lotto ticket. He cashed in his stake in the company in the late 1980s and moved to California to pursue venture capital projects, as well as his twin fascinations: genetic programming, a field in which he holds 16 patents, and, of course, the Electoral College.

Now he oversees his nonprofit's small office in nearby Mountain View, travels to state capitals to lobby legislators - he has testified before legislative committees from North Dakota to New Mexico - and hires lobbyists to stoke fires in states he hasn't gotten to yet. He's donated hundreds of thousands of dollars to Democratic candidates and committees over the years, but he swears his cause is nonpartisan.

And it's easy to believe him when he condemns another of the system's injustices. It's not about partisanship - it's about fairness for America's deprived, neglected ... um, venture capitalists, among others.

"One main problem with the system is that the party nominees totally ignore two-thirds of the states," he says. "All the money, all the attention goes to battleground states. The candidates spent zero on advertising in California in 2004 because everybody knew it was safely Democratic."

The result, he says, distorts the political agenda. Trivial issues take center stage - "Who cares about the Cuban embargo except swing voters in Florida?" - while critical issues are ignored.

"Where's the computer business located?" he asks. "California, Massachusetts, Texas - none of them swing states. So they're ignored, and the problems of an entire sector of our economy are never debated, even though it created much of our recent prosperity.

"We hear nothing about H-1B visas for foreign students. Nothing about the R&D tax credit, funding for the National Science Foundation, the tax treatment of stock options. Nothing. Biotech has the same problem. Its issues aren't relevant to the presidential campaign because the industry's not in a swing state.

"Venture capital has been the engine of our economy, and when it comes to economic issues, who does the White House listen to? General Electric. The Rustbelt. Because those are the swing states."

Wised-up politicos in Washington will see Koza's idea as a phantasm; in an informal canvass recently, none had even heard of it. But Koza has something politics seldom sees: the passion of a capitalist fused with the passion of a hobbyist. Silicon Valley has shown that it can be an unbeatable combination.