My startup's secret recipe

I am sorry. I think this is a dumb idea that will set you up for failure.

This makes one person the bottleneck, and the startup very rigid and inflexible.

Just go get a real team, convince great coders of your idea--people you can trust--and share everything with each other.

Its hard enough to do a startup as it is.

my father told a similar story. his friend, a camera man, wanted to assemble a film camera from scratch. This is Soviet union in the 1960s. cameras were pretty impossible to "own". Studios owned by the State owned cameras. But he had access to the cameras at the studio. He was not a machinist, but could take them apart and service them. If he went to a machinist to copy the camera privately it would cost him a lot of money. possibly a year's salary. So he would take a part at a time. he went to different ones and told them that it is apart that broke on his movie studio camera and that getting a part from the factory would stall the production. So they did it sometimes free of charge and sometimes he paid them with money, they would charge him about a worth of a quarter for a small part.

I'm not sure I understand entirely - it sounds like MVC architecture to me. I'm not sure what layers he's handling, but seems like he's at least abstracting the database away into some simpler calls. As long as he separates everything in a reasonable way, it could even help the stability of his program. The part I don't quite understand is how it adds any significant amount of "protection" from his code being stolen. Programmers forced to stick to a documented API would be able to reverse engineer it, more or less. If he has a secret algorithm in his secret sauce, it's possible that his workers wouldn't be able to duplicate it, but as he's not a passionate programmer, I have my doubts. That said, it seems very closed to me, and I would feel a bit nervous working for him. But it's all just a perception from a single blog post, so I could be way off-base.

I just posted this comment on my blog in response to another developer/entrepreneur with their doubts about my strategy:

Thank you. Your comment summarizes the response I've had from a lot of developers and you are right. However there are a few details I should have but didn't explain. For me it's only temporary to get the foundation ready - the startup itself will have a full team. I am also doing this to give myself time to continue to talk to possible investors and decide where I want to locate.

Will I then throw away all the code and start over?

No and this is why: I have two user interfaces, one for the web and one as a browser extension, in any case they would have been developed by different people. They're two completely independent parts. As last I have the backend, a complex database structure but is setup to only accept a few easy and standard queries. The rest of the queries are handled on database level.

Looking at one of these parts you can't find out my completive advantage, looking at them all together you have my startup. I am incredibly aware of the risks I take with this and especially if I switch developers the problems that can (probably will) come up. I find it harder to read other people's code then to start from scratch and make it myself. Therefore all the complicated parts occur on database level, the code I wrote myself. So no matter what happens there will always be someone in the team (myself) who can work on that, or explain it to others.

This screams naivete to me (but maybe perhaps I'm naive myself ...). I can't imagine how this system would possibly scale or how integration with all the small parts would work. Software is unique in the world of engineering. People know how to build bridges. We, for whatever reason, just don't know how to build software on that scale at the same reliability and with the same consistency that we build bridges.

Software is not made up of interchangeable parts in the sense that you can't outsource each individual function in a giant library and hope for seamless integration. This guy's working at a lesser granularity than that, but I'm just waiting for the day when some unexpected behavior from his controller causes massive failure in his backend and no one knows how to fix it because no one understands the whole system (or at least a big enough chunk of it).

Edit: Just saw his response:

Am doing this to give myself time to continue to talk to possible investors and decide where I want to locate.

If you're prototyping for investors, that may be reasonable, as long as you throw it away and start with a fresh coherent team at some point.

I messed up and I updated my post saying so: http://eisokant.com/2008/07/26/my-startups-secret-recipe/