2 ms·

Companies are still fighting yesterday's war. Traditional industrial labor has a concave relationship between input (effort, skill) and output (diminishing ret

Companies are still fighting yesterday's war. Traditional industrial labor has a concave relationship between input (effort, skill) and output (diminishing returns) where the difference between mediocrity and excellence is minimal to zero. Modern technological work has a convex curve (accelerating returns) where the difference between mediocrity and failure is minimal while that between mediocrity and excellence is massive. When the work is concave, you want to minimize variance, which means you eliminate individuality and manage toward mediocrity. Control-freak managers and petty tyrants are good at that, because you can trust them to tighten down bolts and yell at people. ("Ahem! There's SAND in my boots!" -- Kefka.) You build out a hierarchy, give managers total control over their reports, and even though there's a lot of loss in the form of attrition (good people fired by bad bosses) and squandered capability (strong people doing mediocre tasks) that's treated as a rounding error. People are fungible and, if they're not, then something is wrong with your process. Your goal isn't excellence. It's repeatable mediocrity. Convex work is an entirely different game. It's much more like R&D. The problems being solved are a lot harder, although the upside of a success is much greater. Technological work is becoming increasingly convex with time. When you have concave work, your management strategy is to beat up on the slackers. If the best people are 1.5 times as productive as the average, then one slacker cancels out 2 excellent people, so rooting them out and disciplining them is the right strategy. With convex work, the danger isn't having a few slackers. It's that you don't have any excellent people, or that the excellent people you do have are unmotivated and underperforming. The best way to "manage" convex work is to hire the best people and get out of their way. Managerial extortion (i.e. the manager's use of his unilateral ability to damage an employee's career to put the employee toward his career goals, rather than company goals or individual growth) is ruinous when one is attempting convex work, but large companies don't see it that way. Their processes are oriented completely toward concave, commodity labor. So, "employees leave managers" is only half the story. Managers go bad because companies allow them to do so. A company that doesn't want managerial extortion can implement Valve-style open allocation, but few do. Companies allow them to do so because they're fighting the last century's war. They're industrial machines, and anachronistic in a technological era.

tokenizer 14y ago
So how would you recommend fixing this? You mentioned hiring excellent workers and getting out of their way, but I'm hoping you could elaborate.
- michaelochurch 14y ago
  It's hard. Startups begin in a state of constrained open allocation. It's not "work on whatever you want" because the company needs laser focus, but it's also not big enough for managerial extortion/closed allocation to set in, because the nominal social distance between any two people is still 1 edge. Small startups can have problems, but the loss-of-culture usually happens when companies get larger. Around 15 people, that one-edge factor becomes less true and if you don't explicitly manage the culture, you'll start having political problems. People won't turn malevolent right away, but your process becomes biased in favor of losing good people and attracting bad ones, especially as the TechCrunch coverage mounts and you become attractive to narcissists who want to be your next generation of transplant executives. You have to be an exceptional judge of character to avoid hiring bad executives (remember: many of these people lie and schmooze for a living) and judgment-of-character does not scale. Most companies lose startup culture rapidly (like, before they get to 50) in the process of hiring executives, because traditional dinosaur managers assess their self-worth based on metrics like "number of reports" and want to be handed control of other people. Also, once you have a world in which engineers are getting 0.08% and VPs get 0.5%, the "Uncle Tom Engineer" effect sets in and you lose your engineer-centric culture. Valve did a couple things differently. First of all, my understanding is that they use profit sharing instead of equity. There are advantages and drawbacks of that model, but at the sub-0.5% order or magnitude, "ownership" is meaningless and a more generous profit-sharing system is going to be a lot better for most employees than a meager equity/stock-option system. Second, they made open allocation a cultural pillar that no would-be executive can challenge. Valve trusts before it hires, and that's different from these get-big-or-die VC-istan companies that hire people in bulk but don't implicitly trust most of them with their own time.

tokenizer 14y ago

So how would you recommend fixing this? You mentioned hiring excellent workers and getting out of their way, but I'm hoping you could elaborate.

michaelochurch 14y ago
It's hard. Startups begin in a state of constrained open allocation. It's not "work on whatever you want" because the company needs laser focus, but it's also not big enough for managerial extortion/closed allocation to set in, because the nominal social distance between any two people is still 1 edge. Small startups can have problems, but the loss-of-culture usually happens when companies get larger. Around 15 people, that one-edge factor becomes less true and if you don't explicitly manage the culture, you'll start having political problems. People won't turn malevolent right away, but your process becomes biased in favor of losing good people and attracting bad ones, especially as the TechCrunch coverage mounts and you become attractive to narcissists who want to be your next generation of transplant executives. You have to be an exceptional judge of character to avoid hiring bad executives (remember: many of these people lie and schmooze for a living) and judgment-of-character does not scale. Most companies lose startup culture rapidly (like, before they get to 50) in the process of hiring executives, because traditional dinosaur managers assess their self-worth based on metrics like "number of reports" and want to be handed control of other people. Also, once you have a world in which engineers are getting 0.08% and VPs get 0.5%, the "Uncle Tom Engineer" effect sets in and you lose your engineer-centric culture. Valve did a couple things differently. First of all, my understanding is that they use profit sharing instead of equity. There are advantages and drawbacks of that model, but at the sub-0.5% order or magnitude, "ownership" is meaningless and a more generous profit-sharing system is going to be a lot better for most employees than a meager equity/stock-option system. Second, they made open allocation a cultural pillar that no would-be executive can challenge. Valve trusts before it hires, and that's different from these get-big-or-die VC-istan companies that hire people in bulk but don't implicitly trust most of them with their own time.