Some of my arithmetic is off; I typed too fast, and it's too late to edit now. But broadly the lesson remains: 100,000 uniques can yield enough revenue so that a one person company will be comfortable financially, i.e., enough to pay the rent, buy groceries, take a pretty girl to a movie, and have cash enough for more servers, and if those servers stay busy, then he is on his way to more servers and revenue and, then, hiring.

I, too, did mention that there are cases where have to be in a hurry so that then some equity funding could be crucial.

For the young, inexperienced readers of HN, I'm sorry: HN is helping them get educated, and there is a lot on the Internet, e.g., at Fred Wilson's AVC.com. But help for the young or not, the projects VCs are interested in are exceptional which, we have to expect, filters out a lot of people for various reasons. Tough to make a $1 billion; the economy is not big enough for everyone to make $1 billion.

For being slow to market, my approach is to have done some original research that few others will want or be able to do, and until I am successful nearly no one will try to do. Then if I am successful, they will be too late.

But I've had a long time, e.g., mud wrestling with Microsoft's software, learning that VCs don't look at projects anything like I do, etc., without any significant progress by competition.

A common claim is that any idea entrepreneur A thinks of some entrepreneur B thought of before. Of course, this is silly because among all people who thought of the idea, there has to be a first person and, then, for that first person there wasn't anyone before. Again, that first person is exceptional, as is usually necessary to make $1 billion.

In picking projects, I would recommend picking one where can do some original research that few in IT are able to do and have that research be an 'unfair' advantage and a technological barrier to entry.

Of the "tons of ways to approach a problem", I would urge others to pick some ways that help make the path to success easier. E.g., now and for a few years, exploit an 8 core server for $1500. E.g., if have to be in a mad rush to get to market, then pick another project. If have much doubt about 'product-market fit', then pick a project that right from the beginning has very little such doubt, e.g., like a safe, effective, cheap one pill cure for any cancer. If need a big team long before revenue enough to pay them, then pick another project. My project, about to have all the software needed to go live, has several simple Web pages, two more complicated ones, and two internal server programs with a total of 24,361 source code lines and 6,615 source code statements. Not a lot of code. There is a little more code that runs maybe once a day in 'batch'. So, it's a one person project.

There is an old piece of advice that a key to success in research is good project selection -- similarly for entrepreneurship.