Comment by EGreg | Hacker News Reader

Comment by EGreg | Hacker News Reader

Parent Full thread

There is another way to get the hit rate higher, just like we got better at stuff we couldn't so in the past. It's called SELECTION. That's why all our industrial processes today are better than 100 years ago. They aren't "safer", or performing worse, they perform better than EVER because we have figured out what works and standardized it.

In accelerators, that means having more consistency in what factors work, and more selectiveness. Take things that repeatably work and use that.

In my opinion, companies like Google build one hit after another, internally, by re-using their internal platform and users and virality. There are failures but not as much as with startups. But it all benefits from internal resources, server farms, infrastructure, user base etc.

If you are outside such a company, use an open source platform that worked for others. We are working on such a platform and will probably take partners in a couple years, to build apps for our several million users, and provide them with all the infrastructure to try things out and take a % if it works out.

> In my opinion, companies like Google build one hit after another

Wait, what? Do you have any idea how many ideas they've churned through to get the relatively few hits they've had? Wave didn't do all that well. Orkut had a decent run before they dissolved it (almost 11 years) G+ is still in question.

Yes, they've had some success with some products. And made some excellent purchase decisions (you didn't think they wrote everything they released, did you? Urchin? Writely?).

They've also killed software instead of fully developing it: notebook, reader.

And that is to say nothing of the large numbers of 20% projects. I've personally heard engineers (as a whole) at Google being chastised because the success rate of 20% projects was too high.

That may be true of industrial systems, but human systems bite back. If you devise a strategy for the stock market or the startup market to make more money that anyone else, people who get wind of that strategy can (and will) devise a counter-strategy to eat your lunch.

Innovation is not zero-sum, but if you create a monoculture of "what works", then what will almost certainly eventually work in the market is something different than the monoculture. An interesting analogy here is Feyerabend's take on creativity in science, "Against Method."

Also, I strongly disagree that Google has figured out a process for innovation. Google does not have "one hit after another." They have a hit here and there in a sea of failure, just like Microsoft did. IBM, on the other hand, had one hit after another, until the market figured out how to "disrupt" them, and did.

As an investor I've spent much of the last twenty years trying to figure out how to make a better risk-adjusted return. Either I'm a complete idiot, or it's not as easy as it sounds.

As an investor I've spent much of the last twenty years trying to figure out how to make a better risk-adjusted return. Either I'm a complete idiot, or it's not as easy as it sounds.

These are not mutually exclusive :)

More seriously, you do make a really good point. Innovation is hard and making money from it is even harder.

>In my opinion, companies like Google build one hit after another,

There is a huge difference though. Google has the resources and network to push things towards success. There are so many different reasons why a startup may fail, but Google is able to avoid many of these issues because they are google. Not a good comparison IMO.

Reply on news.ycombinator.com