So you do like "lab" companies - you just don't like calling them labs?
I'm on the "call it like it is" side of the equation.
I actually like what Milk is doing, and I like "lab" companies in the mobile space. I think, outside of Angry Birds" its going to be very difficult for an "iPhone App" company to exit based on a single app. I think the "Tapulous" model, where a single brand makes a whole bunch of Apps is much more attractive. I think these kinds of companies are going to have a much harder time getting early seed funding unless you're a Kevin Rose or Bill Nguyen (BTW I think Color became a lab, I don't think it started that way). But on the flip side going to be much easier to bootstrap these companies.
So you do like "lab" companies - you just don't like calling them labs?
No, there's a distinction trying to be made here. A startup is a temporary lab. The goal is: Do lots of experiments, but generally only a few at a time, and then when you find something that works even slightly stop experimenting and start iterating.
It's true that the word startup connotes experimentation, for the same reason that the word baby connotes being small, helpless, and illiterate. But just as the purpose of a baby is to stop being a baby as quickly as possible and grow up to be an adult, so the purpose of a startup is to find a repeatable business model and then crank on it.
Whereas a true "research" company doesn't do the cranking part. Their entire product is research reports, and their customers are the people who give them grant money.
Now, mobile-app companies with constantly-churning portfolios probably don't fit cleanly into this dichotomy, which seems to be what you're sensing as well. And it's also fair to ask if the "lab" metaphor really fits, e.g., Milk. Maybe the Milk folks have decided to label themselves a "lab" in the short run but will eventually pivot to being "the X company" once they find the value of X, just as Rovio used to be "a company with a portfolio of small games" but is now "the Angry Birds company".
I think the second part of your comment hits the nail on the head: labs might actually be safer because of that lack of total commitment to a single idea. This lets them seek out value in areas that might be too risky, or with product/market fit that's hard to predict in advance. Essentially, it lowers the chance of failure for the startup.
The business model you're describing is YCombinator's business model: Invest in ten or twenty experimental businesses at a time, to hedge against each individual business's probability of failure.
The problem with this model was described by the OP: It's hard to sell "Product/Market Fit in a Box" because so much of product/market fit is tied to specific products built by specific teams. YCombinator finds product/market fits, but they don't just sell the ideas: They sell the specific companies and products along with the ideas, because these things are not separable. Indeed, PG's explicit belief is that even a valuable idea like, say, Dropbox is less valuable than the specific team of people who built Dropbox.
Because the value is in the people and the brands and code they create – these things are your real product – your "labs business" will need a constant stream of fresh recruits to replace the veterans who graduate. You're basically running a school. (Incidentally, it's no coincidence that, even in actual laboratory science, so much funded research gets done at schools.) However, in case you haven't noticed, it's not that easy to find a constant stream of smart founders, each with the skills to find a product/market fit and grow it into a business.
And there's a big cash flow problem. Poor people do not get to be angel investors, no matter how good the angel-investing business model is. Perhaps funding ten startups at once is less risky than funding just one, but it also costs ten times as much up front, and it will take years to find out if you have a winner. Better have years worth of living expenses in the bank. Better have a lot of money you can afford to lose.
This is actually not true. They started as a way to transfer money or IOUs between palm pilots. The web portion of the product got so much more traction that they pivoted.
I agree with most of the points in your blog post and would also not recommend to take the 'labs' approach. My team and me have dropped our 'labs' company just recently. We have built extremely innovative algorithms to tackle a huge problem but completely failed to define and follow an actual product vision and strategy. Our motivation went down over time and our individual visions became more and more misaligned.
What you are saying is right indeed, but most of the problems that you talked about are people problem rather than having to do something with the tag of "labs". Otherwise how different is the failure of two consecutive products by the same team under the same "labs" banner from the failure inside two entirely different banner!
So you do like "lab" companies - you just don't like calling them labs?
I'm on the "call it like it is" side of the equation.
I actually like what Milk is doing, and I like "lab" companies in the mobile space. I think, outside of Angry Birds" its going to be very difficult for an "iPhone App" company to exit based on a single app. I think the "Tapulous" model, where a single brand makes a whole bunch of Apps is much more attractive. I think these kinds of companies are going to have a much harder time getting early seed funding unless you're a Kevin Rose or Bill Nguyen (BTW I think Color became a lab, I don't think it started that way). But on the flip side going to be much easier to bootstrap these companies.