Reaction Wheel

Yesterday an entrepreneur asked me what I thought about AngelList’s syndicates. I said something like “the best founders will still go straight to Sequoia and USV.” He said he had recently been on Sand Hill Road and that’s exactly what the the VCs he had talked to out there had said: that the best entrepreneurs would still come straight to them. “Why,” he asked, “are they so complacent?”

The question he was asking, tactfully, was why was I so complacent? I had to step back and ask myself the same question–why was I thinking like a VC and not like an entrepreneur?

VCs, myself included, are like accounting firms whose books are always a mess, the classic shoemakers whose children have no shoes. We spend most of our working time thinking about innovation, how to disrupt existing industries, and suspending reflexive dismissiveness in the face of ideas that initially seem ridiculous but later turn out to be genius. But we rarely turn that thought process on our own business–in the same way most mature industries don’t–because we’d rather think about our customers than ourselves. Navel-gazing, especially when you’re in the financial intermediation business, is usually insanely boring.

It’s a little less boring this week.

I should say, first off, that I am uncomfortable with the idea of the syndicates for a few reasons. The biggest is that it creates two classes of investors on AngelList: those who get paid carry and those who pay carry. I am especially sensitive to this because I am one of the handful of professional venture capitalists in the country who is bootstrapping their firm. This is a precarious position. If the venture investing world moved from a spectrum of investors to a two-tiered system, it’s anyone’s guess which side of the chasm I would land on. And if you are paying someone carry then you are no longer a venture investor, you are a venture investor’s customer.

But that doesn’t mean I don’t think syndicates will work. As one of the co-founders of Root Markets–one of the first marketplaces for marketing information–I’ve seen how markets evolve and what they need to do to be successful. And through my investments I’ve been a keen observer of the marketplaces for online ad inventory as they’ve gotten traction and scaled. So I can’t be dismissive. I know that a marketplace for venture investments can work. I also know that the ones we have now have a long way to go before they do.

When VCs say that the best entrepreneurs will still come to them, the implied argument is that by skimming the cream, they lower the average quality of startups in the AngelList pool. As a result, second-tier investors will assume the companies in the marketplace are of less than average quality. That, in turn, means the second best entrepreneurs won’t want to be tainted with the association with the lower-quality startups and will go, instead, directly to investors. That, in turn, lowers the average quality of companies in the marketplace and, well, you get the idea.

At Root Markets we introduced an exchange-like marketplace for leads. Because of the specifics of that market, there were huge efficiencies for all players in the ecosystem to use it. But the best sellers already had buyers for their best inventory, they weren’t going to jeopardize those relationships by moving their inventory to an untested platform. They were happy to put the inventory they couldn’t sell onto the marketplace though. Of course, this inventory–since it excluded their best leads–was not very high quality, and that meant we had a lower average quality of leads than the market as a whole. The buyers, who bought somewhat indiscriminately because there was no way to judge the quality of the leads except by buying them and seeing if they panned out, quickly realized this and lowered their bid prices to adjust for it. Lower bid prices meant that the best inventory we had on our marketplace (which was already missing the best inventory) sold for more off the exchange than on it. So now the second best inventory disappeared from our marketplace. This drove down the average quality even further, and then the prices, etc., in a downward spiral until the only inventory left was the stuff that no one in their right mind would buy.

If you’ve followed the ad exchanges you saw the same pattern in 2004-2007. The constant complaint was that there was too much crap inventory and far too little high-quality inventory. This pattern, in fact, is almost universal during new marketplace formation, so citing it and dismissing the viability of almost any new marketplace is easy to do.

But it’s a solved problem. This is Akerloff’s information asymmetry, aka the Market for Lemons. When the seller of an item in a marketplace knows more about its quality than the buyer, the buyer has to assume each item in the market is of average quality, so they have to pay the price for an item of average quality. This means that any seller who has a good of better than average quality can get a better price outside of the marketplace. The items remaining in the marketplace are then of much lower average quality. Repeat this cycle until only the very worst items are left in the marketplace. The solution, of course, is to level the playing field, to make sure that information needed to determine the quality of the goods in the marketplace is available to buyers and sellers.

What Root Markets did, and what the ad exchanges are doing, to make their markets work is to provide the information the buyers need to be able to make their own judgements about quality. Once this started to happen, the quality (and the prices) in the marketplace started to improve. Some marketplaces, the commodities exchanges, for example, work because the good being sold is always of the same exact quality–the only information needed is price. But if the marketplace is selling something that’s not a commodity, it needs to make sure there is enough information available for the buyer to determine quality. And this information must be equally available to all. Any mature marketplace has mechanisms in place to make this happen. In the market for publicly traded stocks, for instance, we have the SEC causing companies to release the information needed to value them; we also have made insider trading illegal, not because it is unethical, but because if it were allowed it would destroy the market.

What is quality in a startup and what information do you need to determine it? There’s not much information to go on in the venture investing business, and most of it is asymmetrical. If the two things most important about a seed-stage startup are the team and the market, then the founders have a ton more information about the first and the VC has a ton more information about the second. There are different asymmetries going in different directions on the two most important dimensions of startup quality. From this alone, you’d expect each of the two sides to easily and often become frustrated with the other’s opinion of value. You’d expect the market for startup investment to be awesomely screwed up.

And it is. But the point is not that the venture investing market is strange and awful–it always has been, we all knew that already, and somehow we still have a market anyway–it’s that AngelList is not necessarily in an untenable position. I said this is a solved problem, but what I meant was it’s a solved problem in theory. AngelList does not need to solve it any better, it just needs to figure out how to solve it as well as we do currently. And while that might not seem a very high bar–early stage startups have not done much, so there’s not much to know about what they’ve done–the reason top VCs usually have either a competency in the field they’re investing in, a competency in quickly learning and apprising a situation (Michael Moritz was a journalist), or a large and expert network is that the information needed to judge a startup’s quality is obscure and diffuse, so it’s difficult to find and assemble. Judging whether a startup might make it or not requires a large number of datapoints, each of which is by itself of little value. A good proxy for the value of a public company can be hacked together from just a few datapoints–earnings, assets, etc. This is not true of early-stage startups. There is a lot of work to be done figuring out what the right information that needs to be disseminated is and how to make sense of it before AngelList or any other marketplace for startups will start to rival the old ways of venture investing.

But thinking like an entrepreneur instead of an embattled venture investor, this is not a problem, it’s an opportunity. How big is the public stock-market ecosystem? A trillion dollars a year? (I don’t know, but I assume this isn’t a bad guess.) Of this the NYSE/Euronext Group, the owners of the New York Stock Exchange, had $2.4 billion in marketplace revenues in 2012 (they had another $1.3 billion in other revenue, selling information and technology services, mainly.) However big the ecosystem is, the exchanges are a tiny piece of it. The marketplace is not where most of the value in the market ecosystem is. Most of the value is in the businesses built around the marketplace, the ones providing information, helping make decisions and facilitate transactions. The aggregate business opportunity around AngelList is bigger than the one in it. And building that supporting infrastructure is vital to AngelList prospering. But if it’s built, my guess is that AngelList, or some startup financing marketplace, will indeed prosper.

Clearinghouses, like the old AngelList, are aggregation technologies. Marketplaces are information technologies. It makes sense for VCs to believe that the old AngelList would never be able to compete with them. But there is a possible future AngelList ecosystem that could. It would be stupid to be dismissive of that.

[Edit: Figured I might as well walk the walk. Started a syndicate on AngelList. This is not a solicitation to invest–I’m not even sure that would be legal–just a POI.]