In his recent book Thinking Fast and Slow, behavioral economist Daniel Kahneman tells the story of observing army recruits out on exercises and his belief that he could spot the potential leaders amongst them. Years later, it turned out he'd been almost entirely wrong. His confident judgment had been a morass of bias, heuristics, and narrative fallacies.

We like to imagine that our powers of discernment are greater than the evidence demonstrates. Similarly, investment banks like to imagine that their fund managers are impeccable pickers—even though the data shows almost no one is over a significant period of time.

We like to think that we know what is not knowable. I was reminded of this when I was asked what had made me so good at picking talent. Many of my ventures—broadcast productions as well as businesses—had involved individuals who went on to great things: stardom, fame, or success running their own businesses. How, I was asked, had I been able to identify ability to early?

On reflection, I knew I hadn't. Some of the entrepreneurs gave no sign that one day they'd be their own boss. Some of the TV talent I most liked went nowhere while others have developed rich public personae. I don't think I picked them; rather, I created the conditions in which they could grow.

So what were those conditions?

An open atmosphere. Every team I've ever run has, I realized, been profoundly democratic—which is to say that any good idea got attention. I've never cared about status except in disliking those who do. I take huge delight in good ideas from anywhere for anything: making the lunch room more pleasant or improving the core technology. Ideas didn't have to fight their way through a structure. They could be blurted out.

Extravagant diversity. It wasn't just that we employed Russians, Taiwanese, Indians, Americans, Brits, and Italians. Our youngest employee was eighteen and our oldest was 68. They'd worked in different countries, states and industries. Some demonstrated enormous emotional intelligence; some had virtually none.  We hired more women than most technology companies—it wouldn't be hard—but you'd be hard pressed to make any generalization about the kinds of people we employed.

Time didn't matter. I've never cared when the work got done—as long as it does get done and in a way that doesn't cause co-workers undue inconvenience. Hours are not what count; productivity is. The same applied to maternity, paternity, and family leave. Individuals figured out what worked for them, and for their families—and that worked for the business. Too many companies seem to feel they have to fight for priority over family life and in that jealous feud, vast amounts of energy and goodwill are dissipated.

Stretch goals were just the start. Almost everything we tackled was hard. And if we cracked it, work got harder. We liked big challenges and cheered anyone who met them. I think it was clear when people came for interviews that the company was fun but not an easy option.

Was this enough to attract and develop wonderfully talented people? I was probably also lucky to hire a few exceptionally creative individuals who then attracted others like themselves. But none of that would have mattered if the work itself hadn't been joyful and hard. It may sound like a contradiction, but doing great work is hard—that's why great people love it.