Comment by rogerkirkness

In my totally uneducated opinion as a founder working on winding down my software company that can't differentiate anymore and considering finding someone in hard tech to partner with, I think problem first is a mistake. Problem first is appropriate for a business, but for a lab, I think it should be organized around the art of possible, deep specialization and capabilities first. You never know what valuable things will come out of that process. Once you have something promising, you can match it to problems and make a business case working backwards from you. Generally startups are the opposite, you want to organize them around a customer and a problem, but fundamental R&D != startups.