I think one aspect of this that deserves more discussion is the difference between simply being self-employed and having a business model. Discussion around just "starting a company" in a generalised sense doesn't really capture this critical nuance very well, though PG's chronic highlighting of the distinction between being consulting companies and product companies does so very well.<p>I had to learn this the hard way, and still am.<p>Being self-employed ("working for yourself") intrinsically is not too different from a glorified J.O.B. It's just a bit more flexible in terms of location, working environment and, depending on what you're doing, some degree of creative control and, if you can afford it, perhaps some level of freedom to reject uninteresting or otherwise non-compelling projects. Instead of having one boss and having to do everything demanded of you, you have 8 different bosses and have to do most of what they demand.<p>Yes, it's definitely not the same relationship, but, if you need money to survive, you don't really have nearly as much freedom to not do what they want - on their terms - as one might imagine in the abstract. And, to be honest, it doesn't much matter what the hourly rate is; supply and demand works against you. Insane multiples of your hourly equivalent wage on salary don't offset that.<p>I encounter a lot of people with the misconception that self-employment, contracting and/or business ownership is somehow inherently lucrative just because it's not being a "wage-slave"; it's really not. I don't make nearly as much as I did at my last job, and I'm quite a ways off from that. Even if revenue doubles, most of that is going to be sunk into reinvestments, paying other people more money, etc.<p>The other problem - this one being the more central preoccupation of PG's advisory writing - is that there's just a very hard limit to the amount of money you can make this way. You're not working 40 billable hours a week, and you can only do so much work as an individual, especially given that you also have to tend to all the auxiliary aspects of running a business and, of course, the selling, marketing and project management part that ensures that you have more work in the pipeline after the current project is over.<p>And if you're like me and don't have much money while having significant living expenses that you got yourself into while living the good life employed, you can't afford to float lengthy projects either; you pretty much need things that can turn around within a month from conception to final delivery and collection because of cash flow imperatives. Your bills are due monthly, but that timing resolution has no imaginable correlation to how your customers are going to schedule and implement anything. The small guys suffer from a shortage of time as much as a shortage of cash; you may get them the final product, but then they'll drag their feet for a month before doing any kind of acceptance testing and sign-off, which is required before you get paid.<p>Working out some sort of pay-as-you-go terms can solve this problem to some degree, as I've had to figure out, but there's only so far that's going to carry you because from a customer's point of view, an 80%-complete project is just a lump of useless code; there's a synergy that happens at the 100% point that gives it orders of magnitude more utility (specifically, the kind that's worth paying for) than when it's less than 100% complete. So, for the most part it's some kind of relatively small down-payment at the beginning, and a big lump sum at the end, whenever they feel like declaring it's done and paying you. And maybe another 10-45 days from there if they have a lengthy AP process. There's no reason the customer is going to want to pay you 7/8ths of the money by the time the project is 7/8ths done.<p>Besides, they all suffer from the same pervasive fear (or "fear," if it's just a matter of principle that they take this posture, which for many is the case) that you're going to screw them, by which it is meant take the money and run - without the necessary "follow-through" and post-sales support and tying up loose ends. So, customers often believe it necessary to "manage" you with carrots, and unless you're just drowning in work and positively don't need anyone who gives you the slightest bit of untoward attitude, you grin and bear it. Facing that kind of condescension and prejudice - if you choose to look at it that way - can be more than a little bit reminiscent of how you felt as an employee before "fashionably cynical" managers and/or businesspeople, with all their accounts of how lazy employees steal/embezzle/fuck them all the time and how they don't trust anyone too much and all that good stuff. It's a pretty crappy place to be in when you just need to be paid cash, like, tomorrow.<p>In a linear business model like that, it's like a barber shop; you want to make twice as much money, you have to cut twice as much hair. The only way to make any serious money this way is to have a nontrivial amount of other people working for you, and that is a whole new set of problems, especially if you're doing something highly specialised and yet can only afford to pay fairly entry-level salaries. There's just no way that the knowledge transfer can be made effective without talent transfer, which, as we all know, is not really possible in the balance of things.<p>That doesn't mean you can't be a worse ("Here! <i>throws manual at employee</i> YOU figure it out!") or better ("Let me sit down and explain and show you everything you need to know about this new system") manager, but if it's not their thing, it's not their thing.<p>Having a product puts you in a very different place, especially if it's a take-it-or-leave-it shrinkwrap product. That's something around which you can build reusable business processes that you can replicate and scale at _decreasing marginal cost_. That's very important from a hiring standpoint, if nothing else; if something is narrow enough, you can specify enough process that more people can do it to a higher - and even more importantly, a more consistent - standard. It also lowers the per-unit price and makes your product more attractive, and so on and so forth. The most important point about product is that the marginal cost on making another deployment/adding another user/whatever is almost negligible, at least in the software world. Doing another one-off consulting project very much lacks that indispensable quality, and thusly limits your profitability.<p>In short, being self-employed in the former sense can easily be twice the work for half the money, and there's absolutely nothing about it inherently that's going to liberate you from the essence of what you most hated about being employed, whether generally or in a particular place. It's something you have to approach very pragmatically and with the right goals.