February 2009A lot of cities look at Silicon Valley and ask "How could we makesomething like that happen here?" The organic way to do it is toestablish a first-rate university in a place where rich people wantto live. That's how Silicon Valley happened. But could you shortcutthe process by funding startups?Possibly. Let's consider what it would take.The first thing to understand is that encouraging startups is adifferent problem from encouraging startups in a particular city.The latter is much more expensive.People sometimes think they could improve the startup scene in theirtown by starting something like Y Combinator there, but in fact itwill have near zero effect. I know because Y Combinator itself hadnear zero effect on Boston when we were based there half the year.The people we funded came from all over the country (indeed, theworld) and afterward they went wherever they could get morefunding—which generally meant Silicon Valley.The seed funding business is not a regional business, because atthat stage startups are mobile. They're just a couple founders withlaptops. [1]If you want to encourage startups in a particular city, you haveto fund startups that won't leave. There are two ways to do that:have rules preventing them from leaving, or fund them at the pointin their life when they naturally take root. The first approachis a mistake, because it becomes a filter for selecting bad startups.If your terms force startups to do things they don't want to, onlythe desperate ones will take your money.Good startups will move to another city as a condition of funding.What they won't do is agree not to move the next time they needfunding. So the only way to get them to stay is to give them enoughthat they never need to leave.How much would that take? If you want to keep startups from leavingyour town, you have to give them enough that they're not temptedby an offer from Silicon Valley VCs that requires them to move. Astartup would be able to refuse such an offer if they had grown tothe point where they were (a) rooted in your town and/or (b) sosuccessful that VCs would fund them even if they didn't move.How much would it cost to grow a startup to that point? A minimumof several hundred thousand dollars. Wufoo seem to have rootedthemselves in Tampa on $118k, but they're an extreme case. Onaverage it would take at least half a million.So if it seems too good to be true to think you could grow a localsilicon valley by giving startups $15-20k each like Y Combinator,that's because it is. To make them stick around you'd have to givethem at least 20 times that much.However, even that is an interesting prospect. Suppose to be onthe safe side it would cost a million dollars per startup. If youcould get startups to stick to your town for a million apiece, thenfor a billion dollars you could bring in a thousand startups. That probably wouldn't push you past Silicon Valley itself, but it might get you second place.For the price of a football stadium, any town that was decent tolive in could make itself one of the biggest startup hubs in theworld.What's more, it wouldn't take very long. You could probably doit in five years. During the term of one mayor. And it would geteasier over time, because the more startups you had in town, theless it would take to get new ones to move there. By the time youhad a thousand startups in town, the VCs wouldn't be trying so hardto get them to move to Silicon Valley; instead they'd be openinglocal offices. Then you'd really be in good shape. You'd havestarted a self-sustaining chain reaction like the one that drivesthe Valley.But now comes the hard part. You have to pick the startups. Howdo you do that? Picking startups is a rare and valuable skill, andthe handful of people who have it are not readily hireable. Andthis skill is so hard to measure that if a government did try tohire people with it, they'd almost certainly get the wrong ones.For example, a city could give money to a VC fund to establish alocal branch, and let them make the choices. But only a bad VCfund would take that deal. They wouldn't seem bad to the cityofficials. They'd seem very impressive. But they'd be bad atpicking startups. That's the characteristic failure mode of VCs.All VCs look impressive to limited partners. The difference betweenthe good ones and the bad ones only becomes visible in the otherhalf of their jobs: choosing and advising startups.[2]What you really want is a pool of local angel investors—peopleinvesting money they made from their own startups. But unfortunatelyyou run into a chicken and egg problem here. If your city isn'talready a startup hub, there won't be people there who got richfrom startups. And there is no way I can think of that a city couldattract angels from outside. By definition they're rich. There'sno incentive that would make them move.[3]However, a city could select startups by piggybacking on the expertiseof investors who weren't local. It would be pretty straightforwardto make a list of the most eminent Silicon Valley angels and fromthat to generate a list of all the startups they'd invested in. Ifa city offered these companies a million dollars each to move, alot of the earlier stage ones would probably take it.Preposterous as this plan sounds, it's probably the most efficientway a city could select good startups.It would hurt the startups somewhat to be separated from theiroriginal investors. On the other hand, the extra million dollarswould give them a lot more runway.Would the transplanted startups survive? Quite possibly. The onlyway to find out would be to try it. It would be a pretty cheapexperiment, as civil expenditures go. Pick 30 startups that eminentangels have recently invested in, give them each a million dollarsif they'll relocate to your city, and see what happens after a year.If they seem to be thriving, you can try importing startups on alarger scale.Don't be too legalistic about the conditions under which they'reallowed to leave. Just have a gentlemen's agreement.Don't try to do it on the cheap and pick only 10 for the initialexperiment. If you do this on too small a scale you'll just guaranteefailure. Startups need to be around other startups. 30 would beenough to feel like a community.Don't try to make them all work in some renovated warehouse you'vemade into an "incubator." Real startups prefer to work in theirown spaces.In fact, don't impose any restrictions on the startups at all.Startup founders are mostly hackers, and hackers are much moreconstrained by gentlemen's agreements than regulations. If theyshake your hand on a promise, they'll keep it. But show them alock and their first thought is how to pick it.Interestingly, the 30-startup experiment could be done by anysufficiently rich private citizen. And what pressure it would put on the city if it worked.[4]Should the city take stock in return for the money?In principle they're entitled to, but how would they choose valuationsfor the startups? You couldn't just give them all the same valuation:that would be too low for some (who'd turn you down) and too highfor others (because it might make their next round a "down round").And since we're assuming we're doing this without being able topick startups, we also have to assume we can't value them, sincethat's practically the same thing.Another reason not to take stock in the startups is that startupsare often involved in disreputable things. So are establishedcompanies, but they don't get blamed for it. If someone getsmurdered by someone they met on Facebook, the press will treat thestory as if it were about Facebook. If someone gets murdered bysomeone they met at a supermarket, the press will just treat it asa story about a murder. So understand that if you invest in startups,they might build things that get used for pornography, or file-sharing,or the expression of unfashionable opinions. You should probablysponsor this project jointly with your political opponents, so theycan't use whatever the startups do as a club to beat you with.It would be too much of a political liability just to givethe startups the money, though. So the best plan would be to make it convertible debt, but which didn't convert except ina really big round, like $20 million.___How well this scheme worked would depend on the city. There aresome towns, like Portland, that would be easy to turn into startuphubs, and others, like Detroit, where it would really be an uphillbattle. So be honest with yourself about the sort of town you havebefore you try this.It will be easier in proportion to how much your town resembles SanFrancisco. Do you have good weather? Do people live downtown, orhave they abandoned the center for the suburbs? Would the city bedescribed as "hip" and "tolerant," or as reflecting "traditionalvalues?" Are there good universities nearby? Are there walkableneighborhoods? Would nerds feel at home? If you answered yes toall these questions, you might be able not only to pull off thisscheme, but to do it for less than a million per startup.I realize the chance of any city havingthe political will to carry out this plan is microscopicallysmall. I just wanted to explore what it would take if one did.How hard would it be to jumpstart a silicon valley? It'sfascinating to think this prize might be withinthe reach of so many cities. So even though they'll all stillspend the money on the stadium, at least now someone can ask them:why did you choose to do that instead of becoming a seriousrival to Silicon Valley?Notes[1]What people who start these supposedly local seed firms alwaysfind is that (a) their applicants come from all over, not just thelocal area, and (b) the local startups also apply to the other seedfirms. So what ends up happening is that the applicant pool getspartitioned by quality rather than geography.[2]Interestingly, the bad VCs fail by choosing startups run bypeople like them—people who are good presenters, but have noreal substance. It's a case of the fake leading the fake. Andsince everyone involved is so plausible, the LPs who invest in thesefunds have no idea what's happening till they measure their returns.[3]Not even being a tax haven, I suspect. That makes some richpeople move, but not the type who would make good angel investorsin startups.[4]Thanks to Michael Keenan for pointing this out.Thanks to Trevor Blackwell, Jessica Livingston, RobertMorris, and Fred Wilson for reading drafts of this.
