This article is not nearly the first time I thought about such things. His reason for getting a technical Ph.D. was essentially the same as mine. And I conclu

This article is not nearly the first time I thought about such things. His reason for getting a technical Ph.D. was essentially the same as mine. And I concluded the same as he did, that to use such education in business, basically have to start a business and own it. He omitted my basic rule: For getting venture funding for such a company, the rule is that some successful research promising as the core 'secret sauce' of a new business together with a dime won't cover a 10 cent cup of coffee, that some solid research for a great technological advantage with some traction is significantly less good than just the traction. For more, such discussions miss some big points likely especially relevant now. First, let's consider the 'big problem': Since we've already decided that we have to be in business, the 'big' we might as well take as promising to make a lot of money, e.g., build a company worth $100 million times some factors of 10. Second, for building this company, we come up with two lists: The list in column A is for technology we have or might invent. Column B is for problems people have and are or at least will be quite willing to pay to get solutions (if there is a big question about 'product/market fit', then leave the problem on the cutting room floor -- we want no questions about eager customers). Yes, it can be that we deliver the solution for free on a Web site and then get paid by advertisers eager for the eyeballs of the users -- that's a small detail we all understand. We want no technology sold before its time, no newly invented technology looking for suitable real problems, no products without eager customers. Then we pick a 'crucial core' technology from column A and a corresponding problem we will solve from column B. Third, we already know how we will deliver the solution -- over the Internet -- and we already know how the solution will be generated for the users/customers -- via software on computers. Why do we know these things? Because of the astounding, historic improvements in price/performance of bandwidth and computing, and platform software, the fact that these improvements have not yet been fully exploited, and the fact that in the current economy information generated from software can be among the most valuable, and most profitable, results to bring to the market. Biotech? Have to wrestle with the FDA, at $100 million or so a round. Energy? It's an old field, has been slow to exploit nuclear energy which was and remains one of the greatest opportunities in the ascent of man, and, thus, really is stuck in the mud. Materials science? A sheet of carbon atoms is not yet an actual product that can be sold to solve a real problem. Flying cars? I won't comment! Fourth, here's a secret (make a billion): When we look at column A of the technology, note carefully the phrase "or might invent". No joke, guys. It's too easy and common to laugh at original research. Most of the people laughing have done no original research. And it's too easy to notice that the guys in universities that do original research usually get maybe papers, professional reputation, and tenure but no business revenue. Why? Because heavily that research is funded for US national security or US health, and in either case the problem sponsors in the funding agencies don't want just 'research' to make good products to solve real problems but another E = mc^2 or actual good progress on a cure for a major disease, especially cancer. Considering that they are spending my tax money, I'm essentially pleased. NSF/NIH money is too short now to fund the best, good, far out, fundamental research; there's no hope for it funding good original research to make money. But such research can be done. Trust me! Then with the research, old or original, from column A and problem from column B where the research provides a good solution to the problem, convert the research to the corresponding software, load it onto a server, put a Web site or an app in front of the server software, please users/customers, and make money. In this, research just such as I described is crucial for much of the progress people are not seeing but have been expecting. Silicon Valley VCs, listen up: You guys will have to start taking research, original and/or old, seriously or risk funding just more social, mobile, locational, sharing apps or trying to buy tickets on airplanes that have already left the ground and are not taking on any more passengers. Entrepreneurs, listen up: Pick a pair from columns A and B, write the software, get the server up, and go live. The software? Mostly the time is just 'overhead' that you have already covered as part of your day job. The server might cost you less than $1000. The bandwidth might cost you less than $100 a month. Or just put your software in the cloud (at maybe some risk of your crucial, core research). That's the opportunity.