More than a few people have observed that entrepreneurship is extremely simple: all it takes to build a successful company is to make something people want, then sell it to them.

Of course, there’s a difference between simple and easy. After all, 90% of new businesses fail. So entrepreneurs lay awake at night, thinking about how to grow their companies. But they tend to worry about the wrong things: how to make something, and how to sell it. In my experience, those parts aren’t actually the problem. Sure, getting the making and selling right requires ungodly amounts of hard work. As Paul Graham has described it, a startup is a bargain in which you squeeze a lifetime’s worth of work into three to five insanely hard years, in exchange for receiving a lifetime’s worth of salary at the end of that time. It’s tough. Very tough. But that work, the making and selling, is rarely where companies actually go off the rails. Indeed, it turns out both parts tend to yield eventually to smart, focused, head-down busting ass.

The thing that really kills companies is the part that founders worry about less: making something that people want. I’ve screwed that up in a bunch of ways in the past myself, and I’ve seen literally thousands of current and prospective companies do it, too.

Figuring out what people want is hard. And it’s hard for a lot of reasons. In the tech world, for example, it’s hard because builders tend to forget they’re different from regular users; hackers argue about the relative merits of Emacs vs VI, while according to recent research 90% of ‘regular people’ don’t use keyboard shortcuts. And it’s hard because, as Steve Jobs famously observed, those people don’t even know what they want until you show it to them.

So figuring out what people want is tough. During the first Internet bubble, VCs ‘solved’ that problem in a standardized way: by hiring MBA CEOs. Find someone with an HBS diploma and some biz dev / sales experience, put him (or her, but probably him) in charge, and task him with figuring out what users want, then with explaining it to the engineering team. While that sounds excellent, unfortunately, it doesn’t actually work, as the subsequent implosion of the internet sector demonstrated. In short, it turns out it’s nearly impossible to figure out what you should build, if you have no idea what you can or can’t build.

In today’s Internet Bubble 2.0, VCs read the lesson of the MBA CEO debacle as: put the tech guys in charge. Now, everyone wants teams of ‘technical founders’. But, in my estimation, that’s a bit of an oversimplification, as not all tech founders are the same. As Geoffrey Moore observed in his excellent (albeit slightly dated) Crossing the Chasm, bleeding edge types actually break into two, very distinct sub-groups: technologists, who are excited about technology for technology’s sake, and visionaries, who are excited about what technology can do for people, about how it might change real, day-to-day lives.

Both types these days pass themselves off as technical founders – that’s what gets funded. But when it comes to actually writing code, the visionaries tend to be more or less crap. Consider Foursquare founder Dennis Crowley – ‘technical’ visionary to Naveen Selvadurai’s legitimately technical technologist; while the two wrote the first version of the app together, their first hire, Harry, was initially tasked with rewriting all of Dennis’ code. At the same time, it’s the visionary who shoulders that crucial question of what people want. Hacking skill aside, it’s good news that Dennis squeezed his quasi-technical way to the helm, as Foursquare would never have grown to what it is today without his lead.

Even if we also call them ‘technical founders’, visionaries aren’t exactly tech peeople, nor exactly business people, but some weird hybrid, some kind of geek ambassador, living in the world between. As a result, the ideal technologist/visionary startup pairing is easy to miss – or, at least, to mischaracterize. Some would see the pair as a tech guy and a business guy, while others would see two tech guys. Neither is quite right. Because what, exactly, was Steve Jobs? Tech guy? Biz guy? Neither and both. He was the prototypical visionary to Woz’s prototypical technologist.

Recently, in an effort to re-secure the US’s place on the world innovation and economic stage, there’s been a strong push to increase the number of engineers coming out of America’s colleges and universities. But if we believe startups are a real driver of innovation and growth, I worry that education push will miss half of the founder equation. Our education system tends to divide students binarily into ‘art people’ and ‘science people’, giving short shrift to those in-between geek ambassadors.

Computer Science departments, for example, are notorious for disdaining ‘dilettantes’. If you’re not hacking compilers in assembly language, you might as well head back to the theater department, because most CS profs have little patience for or interest in anyone who isn’t at least willing to pretend they’re chasing a CS PhD down the line. Still, from what I’ve observed, at least a small number of budding visionaries manage to find ways to build the education they need, often hiding out in the slew of new ‘cognitive science’ majors that have popped up in the last decade – a spot that allows them to balance CS classes with psychology, philosophy, neuroscience and linguistics.

To grow the next generation of startups, we need to grow the next generation of both geek ambassadors and top-notch hackers, then to find smart ways to pair off the two. A technologist and a visionary. It’s the best way to build a startup that makes something amazing – something that people really want.