I won’t say when or where, but at one point in time, a colleague and I were discussing our “red button numbers” for the organization under which we toiled.
What’s that? The concept is this: a genie offers you the option to push a “red button” and, if you do so, your company will go bankrupt and cease to exist. Equity will be worthless, paychecks will bounce, and jobs will end. However, every employee in that company gets the same cash severance. (Let’s say $50,000.) The stipulation that every employee gets paid is important. I’m not interested in what some people might do if they had no moral scruples. Some people would blow up their employers for a $50,000 personal payoff, with everyone else getting nothing, but almost no one would admit this, or do it if it were to become known. If everyone gets paid, it pushing the red button becomes ethically acceptable. At $50,000 for a typical company? Hell yeah. Most employees would see their lives improve. The executives would be miserable, getting a pittance compared to their salaries, but… seriously, fuck ‘em. A majority of working people, if their company were scrapped and they were given a $50,000 check, would be dealt a huge favor by that circumstance.
The “everyone gets paid” red-button scenario is more interesting because it deals with what people will do in the open and consider ethically acceptable. When I get to more concrete matters of decisions people make that repair or dismantle companies, the interesting fact is that most of those decisions happen in the open. Companies are rarely sabotaged in secret, but disempowered and decomposed by their own people in plain view.
The “red button number” is the point at which a person would press the button, end the company, have every employee paid out that amount, and consider that an ethically acceptable thing to do. It’s safe to assume that almost everyone in the private sector has a red button number. For the idealists, and for the wealthy executives, that number might be very high: maybe $10 million. For most, it’s probably quite low. People who are about to be fired, and don’t expect a severance, might push the button at $1. Let’s assume that we could ask people for their red button numbers, and they’d answer honestly, and that this survey could be completed across a whole company. Take the median red button number and multiply it by the number of employees. That’s the company’s defensibility. We can’t actually measure this number directly, but it has a real-world meaning. If there were a vote on whether to dissolve the company and pay out some sum D, divided among all employees equally, the defensibility is the D* for which, if D > D*, the motion will pass and the company will be disbanded, and if D < D*, the company will persist. It’s the valuation the employees assign to the company (which is, often, a very different number from its market capitalization or private valuation).
Of course, such a vote would never actually happen. Companies don’t give employees that kind of power, and there’s an obvious reason why. Most companies, at least according to the stock market or private valuations, assigned values much greater than their defensibility. (This is not unreasonable or surprising, if a bit sad.) I can’t measure this to be sure, and I’d rather not pick on specific companies, so let me give three models and leave it to the reader to judge whether my assessments make sense.
Model Company A: a publicly-traded retail outlet with 100,000 employees, many earning less than $10/hour. I estimate the median “red button” number at $5,000, putting the defensibility at $500 million. A healthy valuation for such a company would be $125,000 per employee, or $12.5 billion. Defensibility is 4 cents on the dollar.
Model Company B: an established technology company with 20,000 employees. Software engineers earn six figures, and engineers-in-test and the like earn high-five-figure salaries. There’s a “cool” factor to working there. I’d estimate the median “red button” number at about 9 months of salary (and, for some of the most enthusiastic employees, it might be as much as five years, but at the median, it’s 6-9 months) or $75,000, putting the defensibility at $1.5 billion. A typical valuation for such a company would be $5 million per head, or $100 billion. Even though this is a company whose employees wouldn’t leave lightly, its defensibility is still only 1.5 cents on the dollar.
Model Company C: a funded startup, with 100 employees and a lot of investor and “tech press” attention. Many “true believers” among the employee ranks. Let’s assume that that, to get a typical employee to push the “red button”, we’d have to guarantee 6 months of pay ($50,000) and 250 percent of the fully-vested equity (0.04%) because so many employees really expect the stock to grow. The valuation of the company is $200 million (or $2 million per employee). We reach a defensibility of $250,000 per employee, or $25 million. That’s a lot, but it’s still only 12.5% of the valuation of the business.
None of these companies are, numerically speaking, very defensible. That is, if the company could be dissolved to anarchy with its value (as assessed by the market) distributed among the employees, they’d prefer it that way. Of course, a company need not be defensible at 100 cents on the dollar for its employees to wish it to remain in existence. If a $10 billion company were dissolved in such a way, there wouldn’t actually be $10 billion worth of cash to dish out. To the extent that companies can be synergistic (i.e. worth more than the sum of the component parts) it’s a reasonable assumption that a company whose defensibility was even 50 percent of its market capitalization would never experience voluntary dissolution, even if it were put to a vote.
In real life, of course, these “red button” scenarios don’t exist. Employees don’t get to vote on whether their companies continue existing, and, in practice, they’re usually the biggest losers in full-out corporate dissolution because they have far less time to prepare than the executives. The “red button number” and defensibility computations are an intellectual exercise, that’s all. Defensibility is what the company is worth to the employees. Given that defensibility numbers seem (under assumptions I consider reasonable) to consistently come in below the valuation of the company, we understand that companies would prefer that their persistence not come down to an employee vote.
That a typical company might have a defensibility of 5 cents on the dollar, to me, underscores the extreme imbalance of power between capital and labor. If the employees value the thing at X, and capital values it at 20*X, that seems to indicate that capital has 20 times as much power as the employees do. It signifies that companies aren’t really partnerships between capital and labor, but exist almost entirely for capital’s benefit.
Does defensibility matter? In this numerical sense, it’s a stretch to say that it does, because such votes can’t be called. Fifty-one percent of a company’s workers realizing that they’d let the damn thing die for six months’ severance has no effect, because they don’t have that power. If defensibility numbers were within a factor of 2, or even 4, of company’s market capitalizations, I’d say that these numbers (educated guesses only) tell us very little. It’s the sheer magnitude of the discrepancy under even the most liberal assumptions that is important.
People in organizations do “vote” on the value of the organization with what they do, day to day, ethically (and unethically) as they trade off between self-interest and the upkeep of the organization. How much immediate gain would a person forgo, in order to keep up the organization? The answer is: very little. I’d have no ethical qualms, with regard to most of the companies that I’ve worked at, in pressing the red button at $100,000. Most employees will be ecstatic, a few executives will be miserable; fair trade.
Thus far, I’ve only addressed public, ethical, fair behavior. Secretive and unethical behaviors also affect a company, obviously. However, I don’t see the latter as being as much of a threat. Organizations can defend themselves against the unethical, the bad actors, if the ethical people care enough to participate in the upkeep. It’s when ethical people (publicly and for just reasons) tune out that the organization is without hope.
The self-healing organization
Organizations usually rot as they age. It’s practically taken as a given, in the U.S. private sector, that companies will change for the worse as time progresses. Most of the startup fetishism of the SillyCon Valley is derived from this understanding: organizations will inexorably degrade invariably with age (a false assumption, and I’ll get to that) and the best way for a person (or for capital) to avoid toxicity is to hop from one upstart company to another, leaving as soon as the current habitat gets old.
It is true that most companies in the private sector degrade, and quite rapidly. Why is it so? What is it about organizations that turns them into ineffective, uninspiring messes over time? Are they innately pathological? Is this just a consequence of corporate entropy? Are people who make organizations better so much rarer than those who make them worse? I don’t think so. I don’t think it’s innate that organizations rot over time. I think that it’s common, but avoidable.
The root entropy-increasing cause of corporate decay is “selfishness”. I have to be careful with this word, because selfishness can be a virtue. I certainly don’t intend to impose any value, good or bad, on that concept here. Nor do I imply secrecy or subterfuge. Shared selfishness can be a group experience. Disaffected employees can slack together and protect each other. Often this happens. One might argue that it becomes “groupishness” or localism. I don’t care to debate that point right now.
Organizations decay because people and groups within them, incrementally, prefer their interests over that of the institution. If offered promotions they don’t deserve, into management roles that will harm morale, people usually take them. If they can get away with slacking– either to take a break, or to put energy into efforts more coherent with their career goals– they will. (The natural hard workers will continue putting forth effort, but only on the projects in line with their own career objectives.) If failures of process advantage them, they usually let those continue. When the organization acts against their interests, they usually make it hurt, causing the company to recoil and develop scar tissue over time. They protect themselves and those who have protected them, regardless of whether their allies possess “merit” as defined by the organization. These behaviors aren’t exactly crimes. Most of this “selfishness” is stuff that morally average people (and, I would argue, even many morally good ones) will do. Most people, if they found a wallet with $1,000 and a driver’s license in it, would take it to the police station for return to its owner. However, if they were promoted based on someone else’s work, and that “someone else” had left the company so there was no harm in keeping the promotion, they’d keep the arrangement as-is. I’m not different from the average person on this; I’ll just admit to it, in the open.
People in the moral mid-range will generally try to do the right thing. On the “red button” issue, most wouldn’t tank their own companies for a personal payout, leaving all their colleagues screwed. Most would press the button in the “every employee gets paid” scenario, because it’s neither ethically indefensible nor socially unacceptable to do so. Such people are in the majority and not inherently corrosive to institutions– they uphold those that are good to them and their colleagues, and bad to those that harm them. However, they hasten the decay of those organizations that clearly don’t deserve concern.
Let’s talk about entropy, or the increasing tendency toward disorder in a closed system. Life can only persist because certain genetic configurations enable an organism, taking in external energy, to preserve local low-entropy conditions. A lifeless human body, left on the beach to be beaten by the waves, will be unrecognizable within a couple of days. That’s entropy. A living human can sit in the same conditions with minimal damage. In truth, what we seem to recognize as life is “that which can preserve its own local order”. Living organisms are constantly undergoing self-repair. Cells are destroyed by the millions every hour, but new ones are created. Dead organisms are those that have lost the ability to self-repair. The resources in them will be recycled by self-interested and usually “lower” (less complex) organisms that feed on them as they decay.
Organizations, I think, can be characterized as “living” or “dead”, to some degree, based on whether their capacity for self-repair exceeds the inevitable “wear and tear” that will be inflicted by the morally acceptable but still entropy-increasing favoritism that its people show for their own interests. The life/death metaphor is strained, a bit, by the difficulty in ascertaining which is which. In biology, it’s usually quite clear whether an organism is alive. We know that when the human heart stops, the death process is likely to begin (and, absent medical intervention, invariably will begin) within 5 minutes and, after 10 minutes, the brain will typically be unrecoverable and that person will no longer exist in the material world. Life versus death isn’t completely binary in biology, but it’s close enough. With organizations, it’s far less clear whether the thing is winning or losing its ongoing fight against entropy. To answer that question involves debate and research that, because the questions asked are socially unacceptable, can rarely be performed properly.
“Red button” scenarios don’t happen, but every day, people make small decisions that influence the life/death direction of the company. Individually, most of these decisions don’t matter for much. A company isn’t going to fail because a disaffected low-level employee spent his whole morning searching for other work. If everyone’s looking for another job, that’s a problem.
In the VC-funded world, self-repair has been written off as a lost cause. It’s not even attempted. The mythology is that everything old (“legacy”) is of low value (and should be sold to some even older, more defective company) and that only the new is worth attention. Don’t try to repair old organizations; just create new shit and make old mistakes. It’s all about new programmers (under 30 only, please) and new languages (often recycling ideas from Lisp and Haskell, implementing them poorly) and new companies. This leads to massive waste as nothing is learned from history. It becomes r-selective in the extreme, with the hope that, despite frequent death of the individual organisms, there can be a long-lived clonal colony for… someone. For whom? To whose benefit is this clonal organism? It’s for the well-connected scumbags who can peddle influence and power no matter which companies beat the odds and thrive for a few days, and which ones die.
In the long run, I don’t think this is going to work. Building indefensible companies in large numbers is not going to create a defensible meta-organism. To do so is to create a con (admittedly, a somewhat brilliant and unprecedented one, a truly postmodern corporate organism) in which enthusiastic young people trade their energy and ardor for equity is mostly-worthless companies, and whose macroscopic portfolio performance is mediocre (as seen in the pathetic returns VC offers for passive investors) but which affords great riches for those with the social connections (and lack of moral compass) necessary to navigate it. It works for a while, and then people figure out what’s going on, and it doesn’t. We call this a “bubble/crash” cycle, but what it really is, at least in this case, is an artifact of limitations on human stupidity. People will only fall for a con for so long. The naive (like most 22-year-olds when they’re just starting the Valley game) get wiser, and the perpetual suckers have short attention spans and will be drawn to something shinier.
Open allocation
What might a defensible company look like? Here I come to one my pet issues: open allocation. The drawback of open allocation, terrifying to the hardened manageosaur, is that it requires the organization be defensible in order to work, because it gives employees a real vote on what the company does. The good news is that open allocation tends naturally toward defensibility. If the organization is fair to its employees and its people are of average or better moral quality, then the majority of them can be trusted to work within the intersection between their career interests and the needs of the company.
Why does open allocation work so well? Processes, projects, patterns, protections, personality interactions, policies and power relationships (henceforth, “the Ps”) are all subjected to a decentralized immune system, rather than a centralized one that doesn’t have the bandwidth to do the job properly. Organizations of all kinds produce the Ps on a rather constant basis. When one person declines to use the bathroom because his manager just went in, that microdeference is P-generation. The same applies to the microaggression (or, to be flat about it, poorly veiled aggression) of a manager asking for an estimate. Requesting an estimate generates several Ps at once: a power relationship (to be asked for an estimate is to be made someone’s bitch), a trait of a project (it’s expected at a certain time, quality be damned), and a general pattern (managerial aggression, in the superficial interest of timeliness, is acceptable in that organization). Good actions also generate Ps. To empower people generates power relationships of a good kind, and affords protection. Even the most superficial interactions generate Ps, good and bad.
Under open allocation, the bad Ps are just swept away. When one person tries to dominate the other through unreasonable requests, or to socially isolate a person in order to gain power, the other has the ability to say, “fuck that shit” and walk away. The doomed, career-ending projects and the useless processes and the toxic power relationships just disappear. People will try to make such things, even in the best of companies, and sometimes without ill intent. Under open allocation, however, bad Ps just don’t stick around. People have the power to nonexist them. What this means is that, over time, the long-lived Ps are the beneficial ones. You have a self-healing organization. This generates good will, and people begin to develop a genuine civic pride in the organization, and they’ll participate in its upkeep. Open allocation may not be the only requisite ingredient for success on the external market, but it does insure an organization against internal decay.
Then there’s the morale issue, and the plain question of employee incentives. Employees of open allocation companies know that if their firms dissolve tomorrow, they’re going to end up in crappy closed-allocation companies afterward. They actually care– beyond “will I get a severance?”– whether their organizations live or die. In closed-allocation companies, the only way to get a person to care in this way is either (a) to give him a promotion that another organization would not (which risks elevating an incompetent) or (b) to pay him a sizable wage differential over the prevailing market wage– this leads to exponential wage growth, typically at a rate of 20 to 30 percent per year, and can be good for the employee but isn’t ideal for the employer. Because closed-allocation companies are typically also stingy, (a) is preferred, which means that loyalists are promoted regardless of competence. One can guess where that leads: straight to idiocy.
Under closed allocation, bad Ps tend to be longer-lived than good ones. Why? Something I’ve realized in business is that good ideas fly away from their originators and become universal, which means they can be debated on their actual merits and appropriateness to the situation (a good idea is not necessarily right for all situations). Two hundred years from now, if open allocation is the norm, it’s quite likely that no one will remember my name in connection to it. (To be fair, I only named it, I didn’t invent the concept.) Who invented the alphabet? We don’t know, because it was a genuinely good idea. Who invented sex? No one. On the other hand, bad ideas become intensely personal– loyalty tests, even. Stack ranking becomes “a Tom thing” (“Tom” is a made-up name for a PHP CEO) because it’s so toxic that it can only be defended by an appeal to authority or charisma, and yet to publicly oppose it is to question Tom’s leadership (and face immediate reprisal, if not termination). In a closed-allocation company, bad ideas don’t get flushed out of the system. People double down on them. (“You just can’t see that I’m right, but you will.”) They become personal pet projects of executives and “quirky” processes that no one can question. Closed allocation companies simply have no way to rid themselves of bad Ps– at least, not without generating new ones. Even firing the most toxic executive (and flushing his Ps with him) is going to upset some people, and the hand-over of power is going to result in P-generation that is usually in-kind. Most companies, when they fire someone, opt for the cheapest and most humiliating kind of exit– crappy or nonexistent severance, no right to represent oneself as employed during the search, negative (and lawsuit-worthy) things said about the departing employee– and that usually makes the morale situation worse. (If you disparage a fired and disliked executive, you still undermine faith in your judgment; why’d you hire him in the first place?) No matter how toxic the person is, you can’t fire someone in that way without generating more toxicity. People talk, and even the disaffected and rejected have power, when morale is factored in. The end result of all this is that bad Ps can’t really be removed from the system without generating a new set of bad Ps.
I can’t speak outside of technology because to do so is to stretch beyond my expertise. However, a technology company cannot have closed allocation and retain its capacity for self-repair. It will generate bad Ps faster than good ones.
What about… ?
There’s a certain recent, well-publicized HR fuckup that occurred at a well-known company using open allocation. I don’t want to comment at length about this. It wouldn’t have been newsworthy were it not for the high moral standard that company set for itself in its public commitment to open allocation. (If the same had happened at a closed-allocation oil company or bank, no one would have ever heard a word about it.) Yes, it proved that open allocation is not a panacea. It proved that open allocation companies develop political problems. This is not damning of open allocation, because closed allocation creates much worse problems. More damningly, the closed allocation company can’t heal.
Self-healing and defensibility are of key importance. All organizations experience stress, no exceptions. Some heal, and most don’t. The important matter is not whether political errors and HR fuckups happen– because they will– but whether the company is defensible enough that self-repair is possible.
The complexity factor
The increase of entropy in an organization is a hard-to-measure process. We don’t see most of those “Ps” as they are generated, and moral decay is a subjective notion. What seems to be agreed-upon is that, objectively, complexity grows as the organization does, and that this becomes undesirable after a certain point. Some call it “bureaucratic red tape” and note it for its inefficiency. Others complain about the political corruption that emerges from relationship-based complexity. For a variety of reasons, an organization gets into a state where it is too complicated and self-hogtied to function well. It becomes pathological. Not only that, but the complexity that exists becomes so onerous that the only way to navigate it is to create more complexity. There are committees to oversee the committees.
Why does this happen? No one sets out “to generate complexity”. Instead, people in organizations use what power they have (and even low-level grunts can have major effects on morale) to create conditional complexity. They make decisions that favor their interests simple and those that oppose them complicated enough that no one wants to think them through; instead, those branches of the game tree are just pruned. That’s how savvy people play politics. If a seasoned office politicker wants X and not Y, he’s not going to say, “If you vote for Y, I’ll cut you.” He can’t do it that way. In fact, it’s best if no one knows what his true preference is (so he doesn’t owe any favors to X voters). Nor can he obviously make Y unpleasant or bribe people into X. What he can do is create a situation (preferably over a cause seemingly unrelated to the X/Y debate) that makes X simple and obvious, but Y complex and unpredictable. That’s the nature of conditional complexity. It generates ugliness and mess– if the person’s interests are opposed.
In the long run, this goes bad because an organization will inevitably get to a point where it can’t do anything without opposing some interest, and then all of those conditional complexities (which might be “legacy” policies, set in place long ago and whose original stakeholders may have moved on) are triggered. Things become complex and “bureaucratic” and inefficient quickly, and no one really knows why.
The long-term solution to this complexity-burden problem is selective abandonment. If there isn’t a good reason to maintain a certain bit of complexity, that bit is abandoned. For example, it’s illegal, in some towns, to sing in the shower. Are those laws enforced? Never, because there’s no point in doing so. The question of what is the best way to “garbage collect” junk complexity is one I won’t answer in a single essay, but in technology companies, open allocation provides an excellent solution. Projects and power relationships and policies (again, the Ps) that no longer make sense are thrown out entirely.
The best defense is to be defensible
The low defensibility of the typical private-sector organization is, to me, quite alarming. Rational and morally average (or even morally above average) don’t value their employers very much, because most companies don’t deserve to be valued. They’re not defensible, which means they’re not defended, which is another way of saying self-repair doesn’t happen and that organizational decay is inexorable.
