Gervais / MacLeod 16: Healthy culture vs. “Why you?”

I’ve discussed a number of problems that businesses face, and started work on at solutions. There’s one major issue that I still need to address. A good organizational culture is expensive. It’s not enormously so, and it pays for itself over time, making it far cheaper than the alternative, but one has to make the conscious decision to pay for culture, or most often it won’t exist. MacLeod pathologies, most pronounced in the stable but undesirable corporate rank culture, seem inevitable because, without ongoing investment in culture, they are. One has to knowingly stand apart from such pathology to prevent it, at least at scale.

Here are a few major ways that it is more expensive for a company to have a healthy organizational culture than the default, broken one. These points are inspired by technology, because it’s what I know. Most of these pertain to risk rather than expense, but the former is generally perceived as the latter, since the real business of business often tends to be risk transfer.

  • Thoughtful and strategic growth. VC-istan startups collect smart people and leave them to fend for themselves, as the company grows ambitiously but not strategically. Healthy culture requires personnel growth in tandem with the legitimate workload (essential or interesting work; not fourth-quadrant executive nice-to-haves). If the workload grows, you must hire more people. If it doesn’t, you shouldn’t. Slow growth might seem less risky, but in the context of VC-istan, it’s much more risky; it’s seem as appropriate for niche “lifestyle businesses” but likely to fail in winner-take-all “red ocean” markets. 
  • Progressive hiring. Most technology companies look for “plug and play” hires who already know the technologies they have and can turn a profit over salary in 1 month instead of 6 months. The tight deadlines of a VC-istan startup seem to necessitate this adversity to ramp-up time. If you’re hiring for culture, though, you need to take account of future potential and you can’t, in practice, be selective for cultural coherence and plug-and-play. You have to hire the people who will make your company great in the long term, rather than for immediate technical-stack fluency.
  • Mentoring. Most companies talk about this lofty ideal, inherited from the guild cultures of old, but few actually do it. One negative side effect of convexity is that, because the time of a seasoned veteran has an order of magnitude more short-term economic value than that of a competent intermediate, mentoring is generally seen as too expensive by executives. Demands placed on the most productive people (by senior people, with power) are already so high that mentorship of new hires (with no power) invariably gets the shaft.
  • Open allocation. Employees are directly responsible for making their work useful to the company, without managerial interference. This is more managerially challenging because it relies genuine motivation, rather than extortion. The upshot of it is that even undesirable work will be done well, because if it’s genuinely important, someone will want to do it after some time. The drawback is that, with work direction coming from the demand rather than supply side, it tends toward “eventual consistency” rather than having the quick-but-sloppy immediacy of managerial edict.
  • Innovation time. So-called “20% time” is not the same thing as open allocation. A healthy company needs both. Open allocation means that a person has the right to move to another sanctioned project without requiring permission, but it’s not “work on whatever you want”. Innovation time means that the employee can work on anything, as part of a team or entirely self-directed, that benefits the company. It enables people to work on 3rd quadrant (interesting but discretionary) work that might have a major payoff (convexity) in the future, but the limited amount of innovation time keeps divergent creativity (which might never pay off) from going off into the weeds.
  • Severance. You’ll need to fire people who just don’t work out. If you fire someone without a severance package, you’re gambling with your reputation. Startups don’t fear termination lawsuits, knowing they’ll either be big or dead by the time that one would conclude– it’s tomorrow’s problem. But severance is also about PR. Reputation risk is more immediate. People talk. Internet happens. If you’re in dire financial straits and everyone knows it, you can lay people off and they probably won’t expect a large package, and the good faith coming from mutual suffering will keep them from disparaging you. If, however, you’re flush with cash and you fire a basically decent “no-fault lack of fit” employee without severance, you’re an asshole and deserve what happens to your reputation.
    • Oh, and don’t even think of using “Performance Improvement Plans”, which allow HR departments to claim they “saved money” on severance while externalizing costs to the team and manager. The morale toxicity of having a “walking dead” employee in the office for one month (hell, even one week) is more expensive than a 3-month severance. Also, most “low-performer initiatives” are dishonest layoffs that turn into politicized witch hunts. You’ve been warned.
  • Firing and demoting toxic high-performers. People can be individual high-performers but damaging to the group. If you can isolate them and demote them out of managerial authority, then fine. Often, the only separation that will work is termination. Toxic people tend to have a desire for control over others that exceeds their leadership ability. They need to be fired, even if they seem “essential”. They aren’t. No one is essential. If someone is insistent on controlling others or, worse yet, bullies or harasses them, you must get rid of that person. You’re a business, not a day care.

All of these efforts pay off in the long term, but are costly enough in the short run to introduce risk. VC-istan, with its disposable-company attitude and obsession with fast growth, is rarely going to pay for any of those. This might seem contradictory: isn’t VC-istan all about embracing risk? It’s not that simple. Organizations tend toward “risk-against-risk compensation”, where increasing risk of one variety requires a zero-tolerance crack down on the other forms of risk, in order to keep total risk below some accepted level (“risk budget”). VC-istan loads up on one kind of it– business-model risk– while being extremely risk-averse with regard to the rest, explaining why most of these “VC darling” startups have horrendous corporate cultures. Messianic founders (often, people with VC contacts who are also too narcissistic to be anything but “serial entrepreneurs”, because they can’t keep normal jobs for longer than 3 weeks) have a tendency to take all the creative risk for themselves. At the interface level of the company, they exhibit an extreme (and not always undesirable) affinity for rapid, sudden changes (pivots) in business model and vision. However, the firm’s entire risk budget is allocated to people at the interface. The interior (where engineers live) is neglected and gets no risk budget (read: only what one can hide). While the company is swift and small, it’s more like a tough culture in which it can still be enjoyable to be a low-level employee– one thrives by hiding risks, working very hard, and getting lucky. Once professional managers (who reduce risks, even of the good kind, because it’s what they’re trained to do) are hired, rank culture sets in and the company is no longer with caring about, except for true shareholders (and not people sacrificing their careers to vest tiny slices of equity).

Good and bad risks– and why the distinction used to not matter, and now does

Companies exist to shift around risks, but this raises a question. Is moving risk the only thing we should care about?

Clearly, there are some good and some bad risks. Most business risks that companies take are beneficial to them and, sometimes, to society at large. Those are good risks. Funding basic research is a good risk; the worst-case scenario is a well-understood financial loss, and the upside is immense. Playing Russian Roulette is a bad risk: it has no upside for anyone, and there’s a 1-in-6 chance of a bullet in the head. Risk can have an irremovable moral character, and the business world tends to ignore that.

Financial risk can be commoditized and transferred, due to separability. This leaves the risk (which can be traded on a market) without a directional moral character or color. All that matters is the (explicitly quantifiable) amount of it that there is. With separability, you can take the attitude that there’s a fixed, quantifiable “pool” of risk that may be taken, and risk allowance will be allocated according to political standing. When you’re dealing with separable commodity risk, it doesn’t matter who has the allowance or what kind of risk it is. As an owner or top executive, you set a maximum amount, let the politically empowered or daring take risks (for personal and corporate benefit) until that limit is reached, and hope for the best.

The problem, in a fully convex technological economy, is that most risks are no longer separable, meaning raw amount of risk (e.g. statistical variance) isn’t the only vital concern. Why? First, there are too many important risks to set up a market for transfer. With industrial commodity labor, individual efforts were concave. Now, each employee is a source of convexity. Creative risks, in the technological world, are individualistic and non-fungible. The payoff distributions are not Gaussian. Old models break down, and management according to risk allowances and principled reduction result in lost upside. In the concave, industrial world, this was tolerable. Concavity, which favors risk aversion, means there’s little value to extreme high performance. Taking a haircut on the upper end was fine: a “Maserati problem”. Convexity’s different. Without that “fat tail” upside, one cannot compete. Losing the upper end means losing almost everything.

Good risks generally involve growth and building: “blue sky” R&D is an example. Bad risks usually involve damage and harm: “low performer” witch hunts might reduce costs, but can demolish morale forever. Bad risks tend to be concave (downside-heavy) and good ones convex (upside-heavy) but that isn’t strictly or uniformly true. In any case, typical industrial-era, MBA-toting management never bothered to learn the difference between good and bad risks because, until recently, it didn’t matter. Risk was a measurable but fungible (thus, always financial) quantity to be sloshed around. Loss induced by sloshing costs was minimal: a rounding error. With inseparable risks, sloshing is infeasible. Risk must be “allocated” and executed where it “naturally” lives. The concrete result of this, amid widespread convexity, is that employees must be trusted with their own time and risk. Not taking that approach will hamstring a business.

What do the players want?

MacLeod organizations exist to transfer certain kinds of risk– especially the personal risk of income volatility that has little to do with business, but is a motivating factor for people to go to work, even under disadvantageous (MacLeod Loser) conditions. If one wanted to see it this way, one could perceive the (idealized) corporation as a purification plant that takes peoples’ personal income risks (bad risk) and turns it into an engine that can provide them steady employment while delivering high average returns for those who can tolerate volatility (good risk). This is the sort of thing that becomes possible in a world of separable risks.

Regarding risk, individual people generally don’t want sudden losses of income or painful or disruptive changes in their daily routines. They especially hate involuntary geographical mobility, one of the strongest predictors of mental illness. The first (and legally inviolable) provision of the corporate social contract is that the employee gets paid speedily for work furnished. Implicitly, they also harbor expectations regarding career management, fair warning of job loss, and fairness– those are delivered with less of a scrupulous reliability, because they can’t be legally enforced. Ultimately, however, most people are looking to be separate from the potentially life-ruining risks that they’d face on a daily basis if they interacted directly with the market. These are the MacLeod Losers. They take a steady wage that falls short of their expected productivity (and that they will lose if severely unproductive) and the difference is the risk premium they pay. The risk-seeking and entrepreneurial MacLeod Sociopaths collect these risk premiums and often get rich.

Intermediate management (which, in a risk-analytic perspective, includes executives, insofar as they are “upper management” but sit between risk-exposed owners and risk-selling workers) is a disease and a treatment. The problem with such people is that they often find ways to take upside risks while externalizing the downside (“heads, I win; tails, you lose”). Executives combine the ambition of ownership and the risk-aversion of management, and often the most fit personality type for this is a thief with a mature and nuanced understanding of risk and a preternatural skill for externalizing and hiding risks. At some point, an Effort Thermocline forms and the true executives, collecting only upside, are less accountable and less productive than the downside-laden chumps below them. Those who succeed in the trade of risk and credibility (the right to take organizational risk in one’s own direction) become the MacLeod Sociopaths. Those who fail become the Clueless, who inadvertently serve as a countervailing force to the mounting pathology and sociopathy of the shell-gaming Sociopaths. The Losers, on the other hand, are aware of the risk transfer that’s going on and, as long as their personal risk is reduced, they don’t care who wins or loses.

The new fourth category of the Technocrat has a different attitude. Clueless are laden with bad risk and unaware of it, thinking they’re doing right by their companies. Losers want to get rid of personal income, location, and condition-change risk. Sociopaths try to take existing good risk for themselves and externalize bad risks, but their main goal is their personal balance (good risk, minus bad risk). Technocrats actively seek good risks, biased toward the convex-friendly opinion that taking desirable risks (rather than reducing disliked ones) is the optimal strategy. They want improvement, hard problems to solve, and creative endeavor.

The Miser’s Question: Why you?

When does emotionally neutral (and justifiable) corporate risk aversion turn into resentment, bad faith, and moral corruption? The answer is the Miser’s Question.

When a person tries to pursue creativity that entails risk (especially, the financial kind) for others, he’s going to run to into the Miser’s Question. Why you? It’s not rejection of the idea. It is to say: the idea sounds like it has merit, it could be a good one, but what makes you the one to execute it? What’s your competitive advantage over the other guys? Shouldn’t we bring in an expert to make those calls?

For a brilliant cinematic example of the Miser’s Question, there’s a scene in Fargo where Jerry Lundegaard– the protagonist, an emasculated and fairly stupid man turned to crime in desperation– discusses a business proposal with his father-in-law, a wealthy banker. For maximal humiliation, the banker recognizes the deal as a good one, takes it for himself, and offers Jerry a trivial finder’s fee. The banker didn’t perceive Jerry as having the competence to execute it. (To his credit, the banker was probably right. The movie is about Jerry’s incompetent execution at, well, a lot of things.)

Worse is a move that I call the Miser Bomb. It’s when a boss takes a subordinate’s idea and gives it to someone else he perceives to be more credible. That’s evil. Once the Miser Bomb falls, the relationship between that manager and the employee is over. Having an idea rejected is just business. The Miser Bomb is rejecting (and insulting) a person. It’s a good idea, but we don’t trust your judgement. That wound never heals. It leads irreversibly to resentment, adversity, and sabotage.

A true-blue Sociopath would fire a subordinate as soon as he drops the Miser Bomb. At that point, it’s probably the only reasonable thing to do: summarily terminate this guy who will never be invested in his work, and will almost certainly desire to undermine his superiors.

I don’t like “why you?” and I especially dislike “why you?” cultures. Having grown up in blue-collar Pennsylvania, I can say there’s clearly a set of hard-working, intelligent people who end up not achieving much because they feel that ambition and upper-tier achievement “just aren’t for our kind”. (Of course, that’s bullshit.) In the Philippines, this is referred to as the “crab mentality“, which refers to the tendency for captured crabs in a bucket (that, individually, one could escape) to pull each other down, so that none get out and all die. It’s militant mediocrity. “Why you?” is the crab-mentality conviction that anything interesting (executive-level business problems, hard-core machine learning, self-executive direction of one’s own career) can only be performed by anointed “special” people, rather than learned through trial and error. Startups are supposed to be beyond that, but I find the opposite often be true. Often, a good idea will be met with, “That would require hiring a real X with production experience at scale.” Never is approached the idea that “real X”es didn’t descend from heaven, but learned those skills by, you know, doing X without asking for permission from risk-averse, emasculated imbeciles who use words like “scaling” without knowing that they mean. This “real X” obsession is often related to a disgusting, social-climbing “our people aren’t good enough” attitude that I’ve seen in many startups, and it must be run away from with extreme prejudice.

Given the toxicity of “Why you?”, why does it still exist? There’s an amazing saying (often falsely attributed to Eleanor Roosevelt) that explains it:

The best minds discuss ideas, middling minds discuss events, weak minds discuss people.

Apply this maxim to business and investment. The most progressive thinkers want to participate in human creativity, so their goal is to validate new concepts in a calculus that balances their divergent creative needs (exploration) with the convergent, pragmatic motive of turning a profit (exploitation, here used non-pejoratively). Middling businessmen want to see numbers about the market, some projections and charts, and fetishistic buzzwords that make them feel safe. The small-minded and deficient operate based on emotion, superficial assessments of character, and credibility. They’re the ones who ask “Why you?”

VC-istan, as I see it, is still a “Why you?” culture. Investors are looking for “track record”. What galls me is when they say, “we don’t invest in ideas; we invest in people”. That’s supposed to sound agile and progressive. Actually, the people who say that sound like small-minded dipshits. If you’re an investor or executive, then your goal should be to invest in human creativity (not “people”, meaning resumes or superficial reputations) and, while pragmatic compromise is always necessary, if creative excellence isn’t your aspiration as an executive/investor, you’re a supernumerary, conformist bag of waste and you should just sit this life out. If you’d rather invest in “track record” than potential, then I’m sorry but the future just isn’t for you.

The answer to “Why you?”: Why not?

The problem with “why you?” is that people internalize it, especially after 20 years of disempowerment. If you’re obsessively questioning whether you’re “good enough” to try something, you’re wasting time that could be spent either learning the requisite skills, or just going the fuck out and doing it. There are a million things worth doing, and if we leave them to “special” or “credible” people, most of them will never be done. There are only about 23 people in the world who are perfectly and implicitly credible (Google’s Jeff Dean is one) at any given time, and each can do a maximum of maybe 6 things-worth-doing at a time, which leaves 999,862 things worth doing that won’t be attended.

The only business organizations that are worth caring about have “why not?” cultures where people focus on doing rather than jockeying for permission to do things.

“Why not?” is not about irresponsible permissiveness. It’s a real question, not rhetoric. There are often good reasons not to take risks. That question must always be examined, and any reasons considered. If, however, the worst-case scenario is merely an affordable expense of time, people should go for it. That should be encouraged in all levels of an organization. People should be implicitly trusted with their own time, and encouraged to take beneficial risks.

The reason I don’t see a future in VC-istan is that it’s doomed to continue along with its “Why you?” mentality, if for no other reason than its centralization of power. VC may be a slight improvement over the traditional corporate culture that peaked in the 1970s, but it’s every bit as doomed to MacLeod stratification, upper-crust entitlement, and pervasive mediocrity. It’s so far along that path that it can’t come back. We need a genuine “why not?” culture; that, more than anything, is going to define creative health over the next several decades.

What will overwhelm VC-istan and drive it into obsolescence? I’d put my money on an armada of 50,000 or more small companies, not focused obsessively on rapid growth. These are derided as “lifestyle businesses”, but I trust them more than anything else that is out there to build out the future. The lifestyle business is viewed as a failure because, in the current regime, it’s too small to be safe. A competitor can kill it, and the owners’ lives are probably ruined. That’s a real problem. It makes a lot of great people not want to do lifestyle businesses, because there is this serious risk. VC-istan is there to provide a safety hatch for good-faith business failure, which lifestyle businesses don’t have. Good-faith failure at a lifestyle business still fucks up your life. If we can provide a reliable, working path for talented entrepreneurs to start lifestyle businesses without egregious personal risk, we’re headed in the right direction. Nothing can (or should) protect businesses that fail in the market from dissolution and reallocation of the (underused) resources, but we should make it easier for the people who fail in good faith to try again. We need to make “yeoman capitalism” a legitimate and sustainable mode of existence.

Moreover, a fleet of 50,000 strong but not gigantic businesses is, in my opinion, a hell of a lot more robust than VC-istan’s few hundred red-ocean “X killers”, where X is some giant corporation that, while sluggish and mediocre in the development of internal talent, still has the means and will to fight viciously against (and possibly demolish) anything with any real chance of killing hit. VC-istan despises lifestyle businesses not because there’s anything wrong with them, but because it favors get-big-or-die “X killer” gambits that are continually reliant on external capital.

The solution to our problem is coming into view, but we still have questions to answer. It would be a great thing to have 50,000 lifestyle businesses building real technology. How on earth are we, as a society, going to pay for it? We come back to age-old economic problems of risk and finance. I’ve painted the broad strokes; the finer ones are where I intend to go next.

Gervais / MacLeod 13: Separability, work and play

We are close. Close, that is, to having done sufficient exploration into the role and operations of economic organizations (such as the corporations or nation-states we love and hate) to start solving some of these organizational problems– to attempt to invent the future.

My role now, in analyzing the organization, is to watch for The Shark. Creative divergence is fun, but at some point, convergence becomes necessary. One needs to make a coherent whole– to solve the problems one has set forth. This can be brutally difficult to do. It’s why the sixth and seventh books of the Harry Potter series are so long and complex, and why some doubt that any person (even a clearly strong storyteller like George R. R. Martin) can handle the monstrous convergence task existing after the first 5 books of A Song of Ice and Fire. To his credit, I think Martin has the right idea– to put the continuing world exploration into a companion book, and finish the damn story in the next two books. Divergence is the fun, playful, erotic part: branching. Convergence is the difficult, thanatoptic pruning task. Writers refer to it as “killing your darlings”. Steve Jobs said it succinctly, for consumer technology: real artists ship. If you diverge for too long without tying everything back into a coherent whole, you Jump The Shark. That’s not something that happens when people are “out of ideas”, as if there were a finite pool of them. When a narrative reaches The Shark, changing writers and hiring an untapped, relief writer won’t help. Shark-jumping occurs when further new stuff (e.g. new characters, motifs, and ideas) can be added, but there’s no good place to put them without loss of clarity. After the shark is jumped, more neat stuff can be added, but the whole (degrading in quality) no longer provides its own reason to care about that stuff.

This isn’t just “meta” wankery. I wouldn’t pollute this series (already tens of thousands of words) with my own problems as a writer. Creative divergence and convergence are at the core of the topics I intend to discuss today. This is the fun topic. It’s why work can, if conditions are right, not suck. I discussed, in Part 9, the role of the organization of a computation problem for an optimization problem, while the “fitness landscape” grows increasingly complicated. In Part 12, I tackled chaos and risk. Financial risk has been commoditized and can generally be measured, sold and bought. Performance risks, at concave labor, tend to be normalized by the Central Limit Theorem when there is a large team of people. Companies want to be able to do this with all kinds of risk, but creative and chaotic risks are harder to manage. The concept involved is separability.

If I had an asset (a security) that, in one hour, would either be worth $10 million or zero based on a coin flip, I’d be inclined to sell it for a middling value. As I alluded in Part 12, this is identical to having $5 million (expected value, my financial wealth) and a zero-mean random variable: 50% chance of +5 million, 50% chance of -5 million. I’d do best to keep the former but have the latter risk off my books.

I might find a wealthy, risk-neutral person and sell him the security. Since it’s not worth it for him to buy it at its expected value, I’d probably sell around $4.99 million, the other $10,000 being a risk premium I pay to him to have that coin-flip out of my portfolio. That’s a case of separable risk. In that scenario, I have financial wealth of $5 million and pay $10,000 to cancel out a risk source threatening a sudden (and possibly catastrophic) swing in income. Here, my making the trade has no effect on the outcome of the coin flip. Risks can be moved around (and, from a cynical perspective, hidden) when they are separable.

Let’s change the game. I’m building a company, or writing a novel, or engaging in some creative effort that involves a lot of work. Let’s keep the same payoff structure, 50/50 chance, $10 million but include individual performance. If I’m diligent, it might be higher than 50%. If I’m lackadaisical, it will be less. If I’m noncompliant (don’t do it at all) then it’s zero. It would no longer be wise for someone to buy at the “expected value” of $5 million.  If I get all the risk off my books, I have no incentive to perform. I might be able to sell that “chance at $10 million” for $100,000 if I’m lucky, in which case, I’m an employee.

There’s a partial solution, which would be to sell some of my risk. I might reach an agreement with a counterparty where I take a $2 million advance and offer him 50% of my upside. He loses $2 million (paid to me, either way) if I fail, and nets $3 million if I succeed. From my perspective, I’m paying a risk premium of $500,000– hefty, but the fact that I’m selling my own performance risk (read: hedging against myself) doesn’t entitle me to the best terms. From his perspective, he’s making a winning deal so long as he believes that, even with my reduced incentive, there’s better than a 40% chance that I succeed. This seems like a win-win.

Here’s the problem: let’s say I make two deals of that sort, getting $4 million with no risk, then I don’t perform. That’s fraudulent and undesirable. I could potentially go further and make twenty such deals. I collect $40 million but am negatively exposed to my own performance (with $10 million coming in, and $100 million owed to counterparties). Not only would that leave me without incentive to perform, but I’d be unable to afford success. So I don’t perform and rob my counterparties blind (and, one hopes, end up in jail). Sold in these large, incentive-affecting blocks, my performance risk isn’t commoditizable.

Clearly, financial markets have regulations in place to prevent those types of abuses: for an obvious starter, executives can’t short-sell their own companies. With commoditized risk, the corrupt cases are few enough that laws can be written to preclude abuses. “Insider trading” can be defined and banned because the set of people who have privileged information is sparse and that set of people isn’t hard to define. Creative and chaotic risks don’t work that way; information asymmetries and incentive effects are much more brutal and fairness is impossible to define. A sane, just regime of risk commoditization is hard to put in place for chaotic risk. It’s an unsolved problem.

Why is risk commoditization so important? It keeps society and the economy fresh. Left to their own devices, economies will converge to a “power law” arrangement where a few have the bulk of the wealth and most have none. It’s not about talent, nor is it about intentional malice or greed. It’s just the result of a mixture of random drift and the feedback cycle of wealth that exists when its many forms (social status, financial capital, access to jobs) are transferrable. It becomes pathological because elites tend toward entrenchment. Then, the executive nerve center (of an organization, or a society) is characterized by complacent mediocrity. Talent desires to break through, but is increasingly far from the resources (capital) needed to put it to use. Risk commoditization is the only mechanism that will connect talent and resources, which are often far away from each other due to the tendency of both to have lopsided (non-democratic) distributions. With commoditized risk, talent (traits that confer favorable odds regarding chaotic and performance risks) and capital (put at risk for mutual benefit) can find each other.

This has always been a hard thing to get right. How is good-faith business failure separated from negligence, incompetence, or outright fraud? It’s not as easy problem to solve at scale. Additionally, talent often has no collateral. Bank loans require personal liability and some capital investment, as such deals probably require, but that limits access in a major way. One needs money to play, and can’t go into high-risk sectors. Venture capital doesn’t require personal financial liability, but views its (otherwise benighted, and somewhat illegal because of the collusion) small-town reputation economy as the machinery that keeps entrepreneurs honest. All of these mechanisms have scaling limitations. They’re only available to a small set of talented people who are in access to some resources; the financial transfer just gets them more. They can connect resources with talent but in a limited, much-to-few, way. A small number gold-stamped “worthy individuals” get the credibility and right to some risk. The rest of the talent is seen as unneeded and, therefore, not developed. Such people are consigned to implement the ideas of others with more social access and credibility. This could be tolerated in a time when most of the work that society needed to be done, and done by humans because there was no alternative, was rote, mechanical, and uninspiring. That was the case for thousands of years, and isn’t anymore. Let’s focus on the history of work– and, later, its purported opposite, play.

Some people fetishize the hunter-gatherer existence, whether they’re talking about prehistoric humans, nomads who coexisted with (or parasitized) agrarian societies, or even contemporary cultures of that kind. One good thing about such lifestyles is that the toxic separation of work and play doesn’t seem exist. People do, and the things they do to subsist don’t seem painful. They enjoy hunting, gathering fruit, teaching others how to do these things, and learning about the natural world. People in such societies seem to spend about 50 hours per week on productive or resource-gathering activity, but without the clock-punching, mind-numbing monoculture of activity, or the required unhealthy arrangements (10-hour block, ass-in-chair, constant visibility). I don’t mean to imply that the more primal lifestyle is superior. It ended for many reasons. It terms of the ability to support a large population, it’s not nearly as fit as agriculture. There’s no way the Earth could support even a tenth of its current population as hunter-gatherers. A stable hunter-gatherer world would entail control of population growth, which requires the dominance of some humans over others. Since it’s mostly subordination (and not productive activity, which humans seem to enjoy) that makes Work such a life-ruining, millennia-long, species-wide clusterfuck, I don’t see primality as any solution. The evidence is strong that Work sucks not because agriculture or technology are unnatural, but because of who we are. The dreamed-of primal utopia is quite flimsy against the same greed that ruins work.

If primal humanity can be described in moral terms, it’s sociopathic– certainly in the MacLeod sense, and often in the truer sense. Lives were nasty, brutish, and short. Positional violence among men, in order to increase social status, was common, with a per-year death rate of men in violence being comparable to those experienced by modern gang members, prisoners, and soldiers. Female sexuality was under male control, with the strongest men having exclusive sexual rights to tens of women– their consent being irrelevant– and the shut-out men angry and tempted toward positional murder. This was the state of affairs in which psychopathy conferred an individual advantage. Most modern people wouldn’t consider it desirable. Yet, there was a way in which the primal and nomadic people were, for lack of a better phrase, more alive than the sedentary, often servile, and often less healthy agricultural people. I’m going to crib the some observations from Paul Graham, in “You Weren’t Meant To Have a Boss“:

I was in Africa last year and saw a lot of animals in the wild that I’d only seen in zoos before. It was remarkable how different they seemed. Particularly lions. Lions in the wild seem about ten times more alive. They’re like different animals. I suspect that working for oneself feels better to humans in much the same way that living in the wild must feel better to a wide-ranging predator like a lion. Life in a zoo is easier, but it isn’t the life they were designed for.


Having seen that happen so many times is one of the things that convinces me that working for oneself, or at least for a small group, is the natural way for programmers to live. Founders arriving at Y Combinator often have the downtrodden air of refugees. Three months later they’re transformed: they have so much more confidence that they seem as if they’ve grown several inches taller. Strange as this sounds, they seem both more worried and happier at the same time. Which is exactly how I’d describe the way lions seem in the wild.

“More worried and happier” is a great observation. The YC founders left their corporate factory farms for the MacLeod-Sociopathic world of business formation, and found that it’s more fun out there. It’s also much harder, much higher in risk, and much more chaotic than the monoculture of bare-minimum subordinate busywork. Is it better than corporate junk work? That depends.

First, longevity is a problem. Almost no one can commit a 480-month block of time (with no departure longer than 2-3 weeks) to that kind of intense work, or even one-fifth of that amount without a break. Second, it would be unwise for most of us 99-percenter poors to bet our incomes on that kind of work, when we need fairly regular paychecks. Even people who have enough savings to last a few months need to worry about their resumes. Paul Graham’s thesis is that neither of these concerns is in force anymore. With the returns available from technology startups (rapid generators of value) being what they are, longevity isn’t an issue (cash-out, travel the world, come back fresh) and neither is the day-job economically competitive. That’s the theory.

There’s much truth in Paul Graham’s ideological commitment to the technology startup as the organization of the future. The uncertainty pertains to the timeframe and also the logistics– it would be good for society for the world’s smartest 10 million (or more) people to have implicit autonomy over their own time, but who will pay for that? It will pay for itself over time, but who takes the initial risk? An investment model (with returns enriching the backers) is obviously requisite, but how is that going to be set up at scale? VC-istan takes a much-to-few model: a lot of resources go to a small number of in-crowd-approved, people deemed actually to have the right to be founders. The advantage it gets out of a much-to-few topology is that it generates a reputation economy discouraging defection. Founders are kept in check by the investors’ ability to lock them out of ever again receiving investment (which is also used for extortion).

VC-istan, in 2013, has met congestion, as seen in the sobering market performance of the genuine concerns (e.g. Facebook) and the proliferation of batty, nonsensical operations (e.g. “Groupon for pets”). VC-istan is now a big company, bereft of real vision, as high-profile investors have congealed into one executive suite. It will be obsoleted by a fleet of smaller, long-term-oriented lifestyle businesses not focused on get-huge-or-die gambits in deep-red oceans. How this will be funded– i.e. what process will discover and enable the relevant talent, and connect it with appropriate risk and capital allocations– is a completely open question. Perhaps that is the problem of the early 21st century. It’s clear that work and money require redefinition, as machines take over the grunt work to which semi-coercive labor was originally directed.

Returning to the primal world in which the evil of institutionalized, subordinate Work had not been invented, we see that it was still sociopathic. Positional violence and warfare among men were common, for one thing. The primal state had its appeals– living under the sun, not caring whether it’s a Tuesday or a Saturday– but people had strong reasons to move away from it as they discovered agriculture (a process that, in truth, occurred gradually over thousands of years) and, later on, writing, law, and the division of labor. There was one inherent problem with this new world. Something horrible became possible: slavery.

When the idea of slavery emerged, it was probably seen as merciful and humane. A fully primal tribe, if victorious in war, would be inclined to kill off the defeated group– at least the men, and probably the women, who were also quite capable of being deadly– to avoid retribution. Primal societies couldn’t use captives as slaves, because they engaged in work activities (such as hunting) that were dangerous, as a defector could murder his masters out in the wild. Agricultural societies, however, could delegate work that was undesirable and free of risk to the superior to a class of permanent subordinates. Primal and agricultural societies both began partaking in one of the most profitable, yet morally rancid, businesses ever devised: the capture and sale of humans.

This created a stratification of work. Most slave-owning societies had multiple tiers of slave. Above them were a class of servants given the work that could only be trusted to free people, followed by financially independent yeomen, and finally the slave-owning leisure class. With productive activities being intertwined with social class– some being desirable, others being dishonorable– and therefore often divorced from intrinsic motivation, it was necessary to separate work from leisure.

As the agricultural era evolved into the industrial one, slavery fell in favor of semi-coercive wage labor. Conditions like morale mattered enough in the industrial context that such economies could not make use of enslaved people. Companies needed to provide financial risk reduction and uniformity in working conditions, setting up the MacLeod Loser trade (facilitated by the Clueless) with people who had a very limited ability to refuse. The industrial era’s sunset will see the semi-coercive model go into obsolescence, pulling society toward fully non-coercive self-executive labor. That, however, is still in the future for most.

Even for high-status people, work became a psychological monoculture (making it unhealthy and, worse yet, boring) in the agrarian and industrial eras. Slave sellers had to run auctions, keep books, and haggle. Monarchs had to hear petitions and pay attention to castle intrigue. Low-status people were subsumed in painful and rote physical labor, while high-status people found their lives devoured by a 24/7 job of social maintenance and posturing. At all levels of society, there emerged a common and unifying thought: I kinda hate this shit. So what did people do when they got away from it? Something that is often called play.

I can’t do justice to play in a few thousand words. As a game designer in addition to the other stuff I do, I know better. It’s emergent activity that is not generally oriented toward production, although there are often echoes of productive activity within it. For most of our history, the ultimate form of upper-class play was hunting— work, in a primal context; play, in one where the activity was irrelevant to the lord’s soul-destroying job of maintaining social position. Play is work-like, but helps people escape into something real that provides a genuine sense of accomplishment.

Indeed, for a lot of people, their play is more like work– open-source software, altruistic travel, amateur art– than the subordinate junk activity that they’re paid to do. Play seems frivolous and indulgent, but it actually leads somewhere. It goes into (as I defined the concept in the last essay) chaos. I mentioned in the previous essay that chaos can be viewed as a creative emptiness. In the vacuum, something new emerges. Play is the chaos that exists when extrinsic direction is removed form activity, often enabling creation that would never occur in a subordinate or mercenary context. Sometimes, that creation is much greater value than anything produced in typical “work”.

Heading into chaos is not profitable, most of the time, but when something novel and useful is discovered, the rewards are immense because so few people can navigate any specific neighborhood of chaos. In this metaphor of chaos as “like” a physical space, we can get a physical sense of the divergent and convergent aspects of the creative process. Divergent creativity (branching) is heading deep into chaos. The issue is that most paths “into chaos” don’t lead anywhere. One needs, at some point, to pop back out and explore a new path. Convergent creativity (pruning) is an anchoring process that extends order and refines the understanding of what kinds of chaos are potentially desirable. Convergence kills off the explored but useless branches, and it reminds us why “chaos” is viewed as a desolate void by most. With divergence only, one will explore limitlessly without return. But if convergence is over-emphasized, one will not go far enough into chaos to find anything that is of value.

Large organizations eventually get to a point where they view their role as keeping chaos out. Law and order are the business of business. Salaries must be paid, deliverables must be met, and work must be defined to leave as little room for variation as is possible. This might induce a psychological monoculture that is unhealthy, bizarre, and probably causative of early cognitive decline, but that’s what the money’s for. This variance reduction serves these firms well in a concave world. Reining in slackers and incompetents compensates for the management’s (counterproductive) interference with the stars, who might only be 1.25 times more productive than the average. To be competitive and functional, organizations in concave labor must drive out variance as much as they can. That means they push away chaos, and drive out play.

Divergent creativity might be called free play. That’s Calvinball. The rules don’t exist yet. Things should be explored. At some point, however, limitless divergence fails to satisfy certain needs. Players want feedback on performance and, often, direction and rules. This necessitates structure that kills off some of the less beneficial or important fruits of chaos. Convergent creativity is discipline. It directs play and makes it more beneficial and focused. The “darling killing” convergence of the novelist produces a coherent story instead of an orderless array of ideas, characters, and settings. Convergence is less “fun” than the divergent part of creativity– it’s demanding and feels more like work– but it’s equally important.

Organizations typically allocated the iota of divergently creative work that they needed done to the caste of people called executives. A very limited role in the convergence process (related to trimming away pesky human chaos, but involving no real control or creative input) went to managers. Workers got no room for risk, no play, no creative control. That model worked for more than 200 years, and it took something radical (widespread, commoditized, and extremely cheap computing) to kill it.

In the concave world, risks are separable and can be commoditized. The top executives can bicker over who gets how much risk to play with, and manage accordingly. The convex world’s risks are non-separable. Move the rewards to one party, the idea generation to another, implementation to a third, and responsibility to a fourth, and you’ll get deceit and dysfunction. If anything worthwhile is being done, there’s just too much chaotic risk that can’t be moved around. People need the right to risk, and they need to be trusted with their own time, or what they produce will be of so low value as to render the enterprise unable to compete. Convexity mandates that people become more self-executive, and that their companies let them do so. The semi-coercive wage labor of the industrial world is dying out. What’s replacing it is fully non-coercive work: disciplined play.

If disciplined play and self-executive, well-treated labor shall carry us forward, then what is the role of the organization? Given how exceedingly difficult it seems to be for a corporate organization to avoid falling into dysfunction, are they desirable in the first place? Why not have some utopian “market state” of self-executive free agents, with no need for these stodgy and often corrupt corporate employers? The answer is that it won’t work. So long as people need an income to survive– and a high one, thanks to corrupt influences on the housing market, a transportation regime that stopped improving in the 1960s, the 9/11 every 24 days that we call our healthcare system, and the expensiveness and exclusivity of the educational institutions that can still place children decently in this imploded economy– they live in a tough-culture reality. The 99%, out of panic, make disadvantageous Loser/Clueless trades that hurt not only them but, in a convex world, society by depriving the world of talent’s proper applications.

The obvious solution to that is a guaranteed basic income. One major advantage of the technological era is that autonomy with one’s own time is often enough “capital” to build something great, computing costs being minimal. But we’re about as likely to see a basic income in the U.S. as we are to change the weather with complaint. Europe’s recent economic crises also show us that, while such states have desirable features, the bigger problem remains unsolved. Europeans have a far better healthcare system and humane vacation allotments (20 days, generous by U.S. standards, is an EU minimum) but their work life isn’t exactly a self-executive paradise. The ideal of a welfare state, to me, is not that it enables people not to work, but it that liberates them to work.

Europe’s systems, with harsher personal bankruptcy laws and more resistance to business formation, haven’t achieved self-executivity much better than we have. With sweeping social reforms (such as basic income) off the table– not impossible, and certainly not unreasonable ideas, but far out of my power– we need to focus on what technocratic individuals can actually do. We can harness individual, mostly localistic, energies. We have one source in those who are starting companies and genuinely want to build great organizations: companies actually worth caring about. That may be a “selfish” motive, and it’s certainly a pragmatically localistic one, but it’s what will save us: people who (for a mix of selfish and altruistic reasons) want to build excellent businesses.

Most of VC-istan, in my mind, doesn’t qualify. These build-to-flip gambits are mostly marketing experiments designed to exploit existing technological trends. Since most of them aren’t building real technology (which requires investment) and measure their own health by “virality”, the get-big-or-die mentality is appropriate for them. We can preach the virtues of open allocation, but cultural health just isn’t important to executives of a company who plan to sell it in 3 years. We need to talk to the people building lifestyle businesses expecting to last 20 years or more. Even if the material ambitions of such firms seem lower– a steady few million per year, instead of corporate Big Swinging Dickery– it’s from a fleet of some 50,000 (plus or minus) small technology companies that we’ll probably get the first successful approach to convexity. In technology, there aren’t as many of those. No one is funding lifestyle businesses yet. One of the biggest financial questions of the 21st century is how to get money into that market. Now that securities markets have been shown to be vulnerable to manipulation, with the housing market continually exposed to execrable corruption, is there really a good reason not to find a way for a much larger pool of capital (not only venture) to connect with talent? While an individual business is certainly too risky for less sophisticated, middle-class investors (hence, the regulations pertaining to accredited investors) there should be a broad-based, relatively de-risked way for them to put money into independent talent.

Financial risk is, of course, one of the foremost problems with the concept of a “self-executive utopia”. There’s another important issue to address, however. While self-executive disciplined play can handle convexity, where does that discipline come from? When and where do people learn the skills necessary to perform convex work at a level even close to what people will pay for? A fully self-executive world still leaves unanswered the question of progress. Who pays people to learn how to become great at things?

Concave labor seems dull; convexity seems sexy and exciting. There is, however, one virtue in concavity. It favors equality. If the best people are only 1.5 times as productive as the mediocre, there’s benefit in bringing the laggards up to speed and giving the competent less attention. The contemporary Theory-Z cult of teamism makes a lot of sense. Convexity, however, seems to reward doubling down on successes and discarding failures. In the short term, that’s correct. That’s exactly what one should do, if optimizing for immediate payoff. Having the best and oldest people mentor the new and young inflicts an opportunity cost– a loss. Guild cultures (lawful good) tolerated that, expecting to be repaid in loyalty from developed talent, but self-executive cultures (chaotic good) struggle with it.

Becoming great at something requires a balance of law and chaos. For example, there’s no educational program in existence that can train someone to be at the forefront of technology. It requires a lot of independent learning (self-execution; disciplined play). Such a person needs enough fluency with law (and humility) to stand on the shoulders of giants, but enough chaotic capability to get out there into chaos where no one will tell a person to go.

There are a number of interesting subproblems that come out of this understanding of convexity. How do we get the desired productivity out of disciplined play? How do we get the right kind of play? Where does the discipline come from? Finally, from a manager’s standpoint, how do we handle convexity’s risks, given the innate non-separability that hasn’t been seen at such scale before? After 13 very long essays, it looks like we finally have the tools to solve some of these problems and, one hopes, the insights necessary to make it possible for business organizations not to suck.

Don’t look now, but Valve just humiliated your “corporate culture”.

The game company Valve has gotten a lot of press recently for, among other things, its unusual corporate culture in which employees are free to move to whatever project they choose. There’s no “transfer process” to go through when an employee decides to move to another team. They just move. This is symbolized by placing wheels under each desk. People are free to move as they are capable. Employees are trusted with their time and energy. And it works.

Surely this can’t work for larger companies, can it? Actually, I’d argue that Valve has found the only solution that actually works. When companies trust their employees to self-organize and allocate their time as they will, the way to make sure that unpleasant but important work gets done is to provide an incentive: a leadership position or a promotion or a bonus to the person who rolls up her sleeves and solves this problem. That’s “expensive”, but it actually works. (Unpleasant and unimportant projects don’t get done, as they shouldn’t.) The more traditional, managerial alternative is to “assign” someone to that project and make it hard for her to transfer until some amount of time has been “served” on the shitty project. There’s a problem. First, the quality of the work done when someone tells a newcomer, “Complete this or I’ll fire you” is just not nearly as good as the work you get when you tell someone competent, “This project will be unpleasant but it will lock in your next promotion.” Second, it tends to produce mediocrity. The best people have better options than to pay 18 months of dues before having the option (not a guarantee, but the right to apply) to transfer to something better, so the ones who remain and suffer tend to be less talented. Good people are always going to move around in search of the best learning opportunity (that’s how they got to be good) so it’s counterproductive to force them into external transfer with a policy that makes it a lot harder to get onto a decent project than to get hired.

Valve actually Solved It with the elegance of a one-line mathematical proof. They’ve won the cultural battle. The wheels under the desk are a beautiful symbol, as well: an awesome fuck-you to every company that thinks providing a Foosball table constitutes a “corporate culture” worth giving a shit about. They’ve also, by demonstration of an alternative, shown a generation of technology workers how terrible their more typical, micromanaged jobs are. What good is an array of cheap perks (8:30 pm pizza!) if people aren’t trusted to choose what to work on and direct their own careers?

I think, however, that Valve’s under-desk wheels have solved an even deeper problem. What causes corporations to decay? A lot of things, but hiring and firing come to mind. Hiring the wrong people is toxic, but there are very few objectively bad employees. Mostly, it comes down to project/person fit. Why not maximize the chance of a successful fit by letting the employee drive? Next, firing. It always does damage. Even when there is no question that it’s the right decision to fire someone, it has undesirable cultural side effects. Taking a short-term, first-order perspective, most companies could, in theory, become more productive if they fired the bottom 30 percent of their workforce. In practice, virtually no company would do this. It would be cultural suicide, and the actual effect on productivity would be catastrophic. So very companies execute 30 percent layoffs lightly. Nonetheless, good people do get fired, and it’s not a rare occasion, and it’s incredibly damaging. What drives this? Enter the Welch Effect, named for Jack Welch, inventor of stack-ranking, which answers the question of, “Who is most likely to get fired when a company has to cut people?” The answer: junior people on macroscopically underperforming teams. Why is this suboptimal? Because the junior members of this team are the ones least responsible for its underperformance.

Companies generally allocate bonuses and raises by having the CEO or board divide a pool of money among departments for the VPs to further subdivide, and so on, with leaf-level employees at the end of this propagation chain. The same process is usually used for layoffs, which means that an employee’s chance of getting nicked is a function of his team’s macroscopic performance, rather than her individual capability. Junior employees, who rarely make the kinds of major decisions that would result in macroscopic team underperformance, still tend to be the first ones to go. They don’t have the connections within the company or open transfer opportunities that would protect them. It’s not firing “from the top” or the middle or the bottom. It’s firing mostly new people randomly, which destroys the culture rapidly. Once people see a colleague unfairly fired, they tend to distrust that there’s any fairness in the company at all.

Wheels under the desk, in addiction to creating a corporate culture actually worth caring about, eliminate the Welch Effect. This inspires people to genuinely work hard and energetically, and makes bad hires and firings rare and transfer battles nonexistent.

Moreover, the Valve way of doing things is the only way, for white-collar work at least, that actually makes sense. Where did companies get the idea that, before anyone is allowed to spend time on it, a project needs to be “allocated” “head count” by a bunch of people who don’t even understand the work being done? I don’t know where that idea comes from, but its time is clearly long past over.