[davidbirk]

I think you're right about the cost & time increases for building hardware, but I'm not sure many fundamental changes would be required for hardware startups.

If the hardware mod. can be approximated fairly well by computer simulations then the same model should work fairly well, IMHO. The goal would still be 'build what people want' and 'excite the Angels' right?

For any machine that is currently being built, someone, somewhere is an expert in whatever part the startup is seeking to improve. If the machine is being sold, buyers know what they are looking for and can give meaningful feedback. So the startup simply needs to convince the experts / investors of the value of their new technology, market size & demand, their ability to execute, etc.

For Tony Wright's “car that runs on lemonade” example, the startup would have 3 months at YC HQ to convince enough experts in the field of internal combustion engines that lemonade is a sufficiently better fuel than gasoline to justify further R&D work. Then, Angel investors can make a relatively informed decision about project risk vs potential rewards.

Increased risk would need to be balanced by increased upside, but I'm thinking that with computer-based tools, hardware hacking is not that different than software hacking.

What do you think?