Over the past couple weeks, I’ve delved into organizational health and the processes that compromise it. (See: Part 1, Part 2, Part 3, Part 4, Part 5, Part 6.) Typically, organizations tend toward a MacLeod hierarchy with three tiers: the Sociopaths, the Clueless, and the Losers, in that order from top to bottom. Sociopaths, who take an up-or-out strategy and either end up at the top or fired, are strategic and dedicated but not subordinate. They’ll never sacrifice their individual career goals for the benefit of the organization, although they may work to improve the company if there is gain for them. These aren’t always bad people. Of the three personalities, I have most in common with the Sociopaths, but I’m not a bad (much less psychopathic) person. Clueless, on the other hand, are dedicated and subordinate but not strategic. They’ll work hard, and take orders, but they don’t have a good intuition for what is worth working on. They’ll eagerly follow or lead pointless, “death march”, projects. They tend to be shunted into middle management where their dedication and eagerness are an asset but their lack of strategy does minimal damage. Losers are subordinate and strategic but not dedicated. They know what’s worth working on and what not, and will follow orders– they’re subordinate because it makes their lives easier– but rarely put forth more effort than is required of them. (As already said, they’re not actually “losers”. More on that, later.) They’re discomfort-minimizers, while the equally strategic Sociopaths are yield-maximizers.
The MacLeod process exists because most organizations make it inconsistent for a strategic person to also be subordinate and dedicated, meaning that the confluence of these three traits won’t occur. Strategic people tend to be rational and selective with their efforts. Some will aim to minimize change and discomfort, willing to sacrifice compensation and career progress in exchange for an easy job where they won’t get fired. That favors subordinacy: make the boss happy, get away with more and reduce risk. If such a person can leave work at 3:00 and collect the same salary, he will. (Why not? If the firm remains happy with his work ethic, is it unethical?) These are the Losers. It’s important to note that they’re not losers in the sense of being disliked, undesirable, or defective. They only “lose” in the first-order economic sense, since their position is one in which they receive less (usually, a steady pittance) from the organization than they put forth. They trade expectancy (long-term average compensation) for the risk-reduction offered by a large, stable company in which they can retain good standing by appeasing a small number of important people. There’s another crowd who are also strategic, but who have more of an appetite for risk and want to capture some of the surplus value generated by the Losers’ trade (of expected value for risk reduction). They make the equally rational decision to put forth a lot of effort, and to take on a lot of risk, in exchange for rapid career growth. They tend to be dedicated, hard workers, but they’re not subordinate. This doesn’t mean that they’re ideologically or obnoxiously insubordinate, but they’ll never buy into the fiction that people must put company goals above their own career needs. Ultimately, such a person will gladly work a 50 or 70-hour week for the right price, but will never sacrifice her own career goals for the company or “the team”. Such people can’t stay in one place for very long. It’s not that they make themselves disliked, but there is a certain “offness” about them. They’re ambitious. They have “an agenda” (as if that were a bad thing). They’re only loyal to companies that treat them well and give their career needs a special value that can’t be given to all. They’re mercenary. It’s up-or-out for them. Companies promote as many of these as they have room for, but must discard the rest. This is actually an impersonal and necessary process– when you have more ambitious people than there is room at the top, you must make a way out for some, even if they are good people that you personal like. However, companies often have a psychological need to create a mythology for hard decisions, and this leads into the “not a team player” epithet that exists to justify this process (that, in my opinion, companies are not required to justify, because it’s not immoral to fire people; it’s just the way human organizations work). These strategic, ambitious people are the Sociopaths. Finally, the non-strategic who are dedicated and subordinate must, almost by definition, be Clueless. Those are the “true believers” who give to the company without expecting much in return.
The MacLeod organization is very stable according to its own internal metric. The Clueless provide a barrier between Losers and Sociopaths on the rank spectrum, while Losers (who give little, and get little) provide one between (value-capturing) Sociopaths and Clueless (who give a lot, and get little) on the hedonic spectrum. The effort thermocline and differential social status hold everything together. It’s an envy-reducing structure. Clueless are oblivious to the effort thermocline and don’t want the jobs immediately above them. Losers don’t want to become Clueless middle-managers– with only token power and slight improvements to compensation, but substantial increases in responsibility and discomfort. Thus, very few people want the jobs immediately above them, and those who clearly do want to ascend (Sociopaths) are fast-tracked either up or out. The MacLeod organization is, on its own terms, fairly peaceful. On the external market, however, such organizations are not always able to compete. MacLeod hierarchies keep peace within the firm, but often make the organization slow to adapt to the world outside.
The problem with the MacLeod organization is that it’s driven by three tiers for which each has a critical defect. Insubordinacy (of the Sociopaths) is not always bad, but it is a defect from an organization’s point of view, and with no one to audit the Sociopath tier here is no way to exert control over the moral character of the leadership. If “good Sociopaths” (Technocrats) are in charge, it will be a good organization. If bad people get into power, they will drive out the good. The problem isn’t that such ambitious people are uniformly or even often bad, but that organizations can almost never self-regulate in such a way as to audit or change the moral character of their leadership– and once an organization goes bad, it will rarely revert, because bad people have an innate competitive advantage; true psychopaths are more agile in social competition than the irreverent-but-decent “good Sociopaths”. In short, MacLeod Sociopaths are not necessarily bad people, but they’re impossible for an organization to control. The Clueless, lacking strategy, are unable to function without good people above and below them. They produce most of the tactical effort, but require management from above and below in order to retain focus on useful stuff. This dependence on others is their shortcoming. Finally, there are the Losers, who do most of the work and will directly affect the company’s ability to execute as well as its reputation for quality of service but who are, rationally, only dedicated enough to remain in good social standing. It’s not true that they “do the minimum not to get fired” (that’s a common misunderstanding of Loser-ism) so much as they manage their performance to the Socially Acceptable Middling Effort (SAME). They want neither the reputation for being a slacker, nor that for working too hard. They aim for the middle. While less important than the strategic competency of the leadership, the SAME level is a major cultural factor in a company’s macroscopic performance.
The organizational issue with the SAME is that it drifts over time. A high SAME will keep the working people of the company, who are often very competent despite their lack of ambition, highly productive. On the other hand, when the SAME falls to zero, the company ends up with a lot of deadwood; Losers stop working, because there’s no reason to do so. In this light, we can understand Marissa Mayer’s recent decision to crack down on work-from-home employees. This effort is about raising the SAME, which is hard to manage in a distributed setting. The problem isn’t that WFH employees are likely to be lazy. That’s often not the case. In fact, the best WFH-ers are far more productive than any in-office employee. The issue is that the good WFH-ers have no effect on the SAME– they get a lot done, but the rest of the office doesn’t see them working hard– and the bad WFH-ers have fallen to zero. An in-office policy will, at least at first, reduce bulk productivity (by hurting the good WFH-ers more than it brings up the bad ones, most of the latter being incorrigible people who will need to be fired) but the expected second-order effect is to raise the SAME. Mayer’s changes may kick Yahoo into the ugly state of a tough culture, but that’s likely to be less dysfunctional than the stodgy rank culture into which MacLeod corporations devolve. In rank culture, getting along well with management is more important than effort itself, which pushes the SAME down as obedient non-performers get a pass and weak managers become entrenched.
If each of the 3 MacLeod tiers is defective in one of these three desired organizational traits, we must ask the question. Can people be strategic, subordinate, and dedicated? It only happens in the context of a mentor/protege relationship. In truth, people who are truly strategic are never fully subordinate or insubordinate. While I originally approached the MacLeod hierarchy and this three-trait decomposition, I assumed the existence of all three traits to be impossible. In reality, those who are strategic are subordinate or insubordinate based on context. Strategic and dedicated people will subordinate, in the short term, if the leadership (management, advisors, investors) shows a long-term interest in their careers. True loyalty is not incompatible with the insubordinate tendency of the highly capable mind. There’s a pay-it-forward mentality and a personal affinity that enables such people to live in harmony with the organization. This is observed in the rarest of the four work cultures: the guild culture, which creates balance in the relative importance of dedication, subordination, and strategy. The other cultures tend focus on one of the 3 traits at the expense of the others: rank culture, on subordinacy; tough culture, on dedication and sacrifice; self-executive culture, on being strategic. Guild cultures encourage balance. Why, then, are guild cultures rare (and dying)? The answer is two-fold. First, they’re hard to maintain, because the leadership must continually refresh its skill base in order to mentor new people, which means that the teachers must “moonlight” as students and study new methods as well as the external market. Second, guild cultures cannot grow fast. If they take on too many new people at subordinate ranks, it becomes clear that the not all of their careers will succeed (there isn’t enough room higher on the ladder) and the whole thing falls to pieces. While guild culture may be “good”, it is not fit (as an evolutionary term) insofar as it does not allow fast growth of its organizations.
A guild culture that collapsed recently is that of large-firm law (“biglaw”). Originally, it was difficult to get an associate position at a “white-shoe” law firm, but one who had one stood a very high chance of making partner. Not making partner (after seven to ten years) was the exception reserved for bottom-5% performers. In the 1980s, this changed, due to the increased workload generated by the private equity boom. The firms began taking on large numbers of associates without allowing the partnership ranks to grow in tandem, the result being that making partner is now the rarity. Now, it’s about 5 percent who get partnership, rather than 5 percent being denied it. The guild culture died horribly, being replaced by a catastrophic tough culture where 70-hour work weeks and “4:30 work drops” (late-day assignments with next-morning deadlines) are the norm. New York attorneys frequently describe the state of their (ex-)profession as “banking without the upside”.
So what defines a healthy working culture? We now have the vocabulary to address this. I’m going to elaborate six statements about a healthy workplace culture, and then arrive at a 7th (The Mike Test) which is the most important of all. I derive these six assertions from the three workplace traits– strategy, subordination, and dedication– in addition to the need for moderation in each.
- (Dedication, or DED) People can be relied upon to do their jobs well, and to treat their work as important. Ideally, this trust and reliability should exist at all levels of the organization. The company functions best if everyone is willing to do the hard jobs.
- (Moderation in dedication, or M-DED) People do not engage in unnecessary sacrifice for social or subordinate reasons. Pain does not become a measure of a person’s work or value.
- (Subordinacy, or SUB) People take a “pay-it-forward” attitude toward their colleagues and the company. Junior employees take direction from mentors; seniors invest in their reports for the long term. People invest in their relationships with the company, because it is worthwhile to do so.
- (Moderation in subordinacy, or M-SUB) Doing the right thing is more important, to employees and the company, than following orders. This requires a culture that enables people to speak up without fear of retribution.
- (Strategy, or STR) Individual employees take ownership of their own work and have the autonomy to place their efforts where they perceive the most value. There’s no need to ask permission or “apply for transfer” to work on something that appears important. This is where open allocation shines.
- (Moderation in strategy, or M-STR) Allowance is made for exploratory work that might pay off only in the long-term. Work need not deliver short-term dividends to be acceptable. People are allowed, at all ranks, to invest at least some of their time into R&D that is abstractly beneficial to the organization, even if it doesn’t produce an immediate benefit.
If I were to grade each of the four cultures (0 to 8, 8 best) for its typical level of success on each of these traits, here’s how I would assign them (based on their observed performance, not on their purported values).
Trait Rank Tough Guild Self-exec.
DED 2 6 5 8
M-DED 3 0 5 6
SUB 4 1 8 4
M-SUB 0 2 5 7
STR 1 4 4 8
M-STR 2 0 5 3
Avg. 2.0 2.2 5.3 6.0
Rank cultures are across-the-board weak. They excel in superficial subordinacy, but the lack of true loyalty earns them only a ’4′ grade for that trait (SUB). The only good thing about them is their internal stability, which is why they are the eventual state of a hierarchical (MacLeod) organization. Still, they tend increasingly toward macroscopic underperformance due to their corrosive mediocrity. When the only thing that will prevent laziness is aggressive management, effective managers leave (wanting better reports) and the company ends up getting stuck with lazy managers who tolerate crappy employees. Tough cultures are slightly better– people actually give a shit, if only for selfish reasons such as aggressive performance reviews– but still fail in most critical areas, and behavior tends toward moral degradation. In tough cultures, people are as subordinate as they need to be to survive, but deeply disloyal. Two examples of tough-culture degradation are Enron and Google after the introduction of “calibration scores”, but the tracks of the organizations are different. Enron maintained its tough culture by executive fiat and experienced top-to-bottom ethical corrosion. Google’s tough culture (introduced by a play-for-play copy of Enron’s performance review system) reverted quickly to a rank culture, because many managers had no desire to enforce it and began agreeing to “peg” calibration scores in return for loyalty. (Google is an unusual example, having a healthy self-executive culture above the Real Googler Line– enabling it to maintain strong macroscopic performance– and a necrotic tough-turned-rank culture below the RGL.) The latter path is more common. Tough cultures usually return to rank cultures; those who have the power to protect people from the harsh review system become the new holders of rank.
Let’s examine the healthy cultures, in the light of the grades above. Note that self-executive cultures get top marks for dedication and moderation in dedication. When people are trusted to direct their own efforts, and rewarded for good work, they tend to put their effort levels at the right level. They also excel at moderation in subordinacy and (of course) strategy. What self-executive cultures tend to be bad at is rewarding long-term investment, including mentoring new hires. Guild cultures, although not especially fit due to their intolerance of fast growth, are well-rounded and healthy. When they work well, they excel in terms of a “pay-it-forward” attitude that replaces subordinacy (or insubordinacy) with genuine altruism. They’re also, perhaps surprisingly, the most prone to long-term investment (moderation in strategy). The central planning inherent in guild culture can be a weakness, but it can also allow the firm to “future-proof” itself consciously– if its leadership wants to do this.
Both cultures struggle when it comes to growth, but in different ways. Guild cultures can grow vertically– bringing in people of lower or higher skill levels– because they have the machinery for assimilating them into mentor/student relationships, but tend to fail at horizontal growth, because guild cultures require a carefully managed balance of work quality, skill level, and rewards. If that gets out of whack due to rapid growth, some form of scarcity (e.g.,of good reports, of high-quality work, or of room at the top) will turn the firm into a zero-sum slugfest that destroys the mutual trust of the guild culture. (It becomes a tough or rank culture.) On the other hand, self-executive cultures can grow horizontally– taking on more people at the same skill level– but tend to be incapable of assimilating people of superior or inferior skill levels to those who are already there, because self-executive cultures rarely provide the incentives for managed growth (as opposed to going out and putting forth work that brings immediate results). An example is Valve, an ideologically self-executive culture that simply does not hire junior-level people– it has enough self-awareness to know that it can’t maintain its self-executive culture in the face of the skill inequality that vertical growth creates.
In truth, self-executive cultures tend to downplay differences in skill– everyone has basic autonomy, there are no real “bosses”– and can’t hire below the prevailing skill level and maintain their culture. They could hire above the prevailing skill level and survive– that would be desirable– but have a hard time getting such people to work for them without offering some authority to them. People who are used to be entitled leaders (executives) do not usually want to work on equal terms with people they consider of inferior skill. Thus, self-executive cultures can only hire near the prevailing level and, since they only succeed if the prevailing skill level is fairly high, the scarcity of such talent means they cannot grow fast even if they would want to.
Tough and rank cultures, on the other hand, grow much faster. Why is this so? Tough cultures do not focus on growth so much as churn. They hire a lot of people, fire a lot of people, and are responsive to market conditions but generally agnostic on the direction of headcount numbers. Tough culture, being the absence of a work culture, has no growth management. A tough culture can hire a person of low market value, accepting the very high chance (in some cases, over 50% per year) that it will have to fire him. What causes directional growth (i.e. more hires than fires) in tough cultures is the long-term tendency toward inefficiency (necessitating greater headcount) and the emergence of pockets of rank culture as the tough culture decays. Tough cultures tend to grow in spite of themselves, because the cultural corrosion reduces individual performance despite the culture’s intent of doing the opposite. Rank cultures, for their part, actively encourage headcount growth. Hiring binges mean that there are more subordinates to go around, which makes managerial decision-makers happy.
In other words, neither of the desirable cultures (self-executive, guild) excels at horizontal growth and vertical growth. This makes them less fit, at least in terms of the ability to subsume people, than the pathological rank or tough cultures of typical organizations. There might be as many of these desirable cultures (self-executive and guild) as there are pathological cultures, but the latter house more people. It may be that most businesses have good cultures, but most employees work in dysfunctional businesses, if only because the dysfunctional work cultures are most able to grow quickly.
This conclusion is depressing: most people will work in dysfunctional cultures. Can we change that? If we want to do so, we need to answer some questions? What is the best workplace culture? Should one aim for the guild culture, or the self-executive one? I think the answer is obvious: one needs a hybrid, with ideas from both. I would say that one should aim for a culture that is mostly self-executive, and especially so for senior hires, but that creates an internal market for mentoring new employees, and for long-term investments, in which those efforts are equally valued. In doing so, it would manage to capture some of the assets of the older guild cultures, thus enabling some degree of vertical growth.
On the topic of growth: in addition to the six cultural evaluations above, I’ll add a 7th that is, above all, most important. The other six pertain to the present-time cultural health, but this 7th determines whether an organization will improve or decay over time. It’s The Mike Test. It has one criterion. The Mike Test is…
…would you, if no one would know or give you credit for doing so, want to hire someone better than you? Would it be rational to do so?
Assume that there’s no hiring bonus, and she’s not your best friend. You won’t get credit, in any form, for a great hire, and she won’t feel preternaturally loyal to you if she rises fast. The only benefit you get from hiring this great person is that you improve the company. The risk you take on is that your relative status in that company will decline. In a large company, the risk is low– you’re unlikely to compete directly with her– but so is the reward, making the balance zero. (It’s not worth it to hire her even in the large firm, not because there’s a risk of competition, but because you could spend time on other things.) I would argue that most companies fail The Mike Test. In the vast majority of companies, it is not rational to hire a person of superior ability.
This hits on the (slightly altered) Jack Welchism that “A players hire A+ players; B players hire C players”. In my use, however, “A” and “B” orientation pertain to context and security rather than innate competence. A players are secure enough that their interests (at least, in hiring) align with the company’s. They want to work with great people, and it has nothing to do with coarse personal benefits (hiring bonuses, “finder’s” credit). They make their companies better. B players, who are insecure because they are unremarkable, want to be safe and manage themselves to the middle. Hiring incompetents minimizes their risk of drifting out of the middle (and into the bottom). Good cultures discourage this kind of B-ness. They make sure that competent people are secure, and they rid themselves swiftly and fairly (preferably with severance; it reduces drama) of incompetents. Sadly, that’s rare. Most companies fail the Mike Test, and MacLeod hierarchies form because of organizations’ tendencies to hire people inferior to the (semi-insecure) real decision makers and, when superior people are brought in, to try to make them inferior using their control over the division of labor. Those tiers emerge because people in an insecure context have the incentive to hire only inferior copies of themselves.
Let’s discuss each of the four cultures in the context of the Mike Test.
Tough cultures fail the Mike Test in the worst way. In a tough culture, no one is secure– that’s the whole point of tough culture– so everyone is a contextual B-player (i.e. incentivized to hire mediocrities and make oneself look better). No rational person living in a tough culture would hire someone of superior skill– that person becomes an immediate threat. Rank cultures don’t fare much better. In a rank culture, one tends to have one of two kinds of managers. The first is the checked-out (i.e., also lazy and mediocre, as befits rank culture) boss who just doesn’t care. There isn’t a major loss inherent in hiring strong people, but it’s not a worthy use of time. The second is the hard-ass, careerist boss who creates a pocket of tough culture under him. (The company might still have a rank culture, but this boss holds high expectations.) That type of boss only cares about hiring insofar as it builds his team, and the tough-culture rules (don’t hire someone better than you, lest you be thrown under the bus) apply. Bosses in rank cultures might want to hire people of superior skill, but only if they could lock in permanent rank superiority. Rank cultures especially stigmatize reporting inversion (i.e. having a younger, less senior, or less educated boss) to the point that it can be career-ending to be at the butt of one. The benefits of having a strong subordinate are offset, in a rank culture, by the risk of reporting inversion. Therefore, both of the dysfunctional cultures (rank and tough) fail the Mike Test, explaining why they hire worse people with each generation of growth.
Guild cultures pass The Mike Test. If you hire someone better than you are, he might end up in a higher place, but that’s okay. You get a mentor, not a threat. A well-managed guild culture can assimilate someone at a higher skill level without eroding the well-being of the rest of the organization. This is the symbiotic nature of the guild culture. Self-executive cultures also pass, because their “bossless” nature means that the prospective hire’s superiority of skill is not a major concern. It’s a good thing to hire a better person, because you’ll have a strong colleague.
To summarize this briefly for each culture:
- Tough cultures fail the Mike Test catastrophically. You endanger yourself directly if you hire someone better than you are.
- Rank cultures fail the Mike Test by apathy. People only hire strong people if they can guarantee that person’s subordinate status. Rank cultures tend toward inferior hiring– strong people are less likely to be subordinate– but not as fast as tough cultures.
- Self-executive cultures pass the Mike Test, as most people would prefer stronger colleagues, but the equality inherent to a self-executive company is a hard sell to one who might expect (because of a higher skill level) a leadership role.
- Guild cultures pass the Mike Test, if they can convince the superior to take the role of a mentor rather than a manager.
Perhaps surprisingly, most VC-funded startups fail the Mike Test. They pass it for founders and real owners, who will have a more successful company, but they fail for engineers. Founders and investors control the dilution process and, even if they lose relative share, they’ll only make deals that are beneficial to them (50 percent of something is better than 100 percent of nothing.) If you own 20 percent of the company, you’re ecstatic if you just hired someone better than you are. What about engineers, however? What are their incentives?
Let’s consider a typical software engineer at a 50-person technology startup with 25 software engineers. His compensation is $25,000 per year lower than the market level, offset by equity equal to 0.05 percent of the company, vesting over 4 years. (In practice, he’d more likely have options at a low strike price, but I’m going to simplify here.) Salary raises are rare in startups, and equity improvements (without a promotion) are almost unheard-of. When things are going badly, that’s not a time to ask for anything; when they’re going well, the appreciation of equity is the raise. In addition to the $100,000 lost over four years at a low salary, let’s value the lost salary growth at $50,000 over 4 years, plus another $50,000 to account for future income lost to being at a low salary level when he exits. So he’s paying $200,000 for 0.05% of the company, implying a valuation of $400 million. This is, most likely, much higher a valuation than the one at which investors would buy into it. However, I can make a strong case that the engineer’s valuation should be lower than that given by investors, especially in a VC-feeding frenzy. When risk, liquidation preferences, cliffs and the lack of control are included, an engineer doesn’t do well to accept this typical startup offer ($25,000 salary drop, no raises; 0.05% equity) unless he can realistically value the company at approximately $1 billion. With the shoddy IPO climate, not many startups deserve that kind of valuation. (In fact, the companies that do are no longer really startups.) So, the equity offered by a startup is not, for most mere engineers, a good reason to be there.
So why do software engineers work for these VC-funded startups? There are many Clueless in the mix who massively overvalue their equity consideration, but most engineers believe their initial grants to be “teasers”. They know that their starting allocations are low, but expect that they’ll get something closer to a “fair” share when they grab those “inevitable” executive positions at which real equity allotments (0.25 to 0.5 percent for Directors, 0.5 to 1.0 percent for VPs, 2.0 for C-level, 3.0 for CTO/COO and 5.0 for CEO) are common. That is what keeps engineers in VC-istan motivated; the implicit (and often broken) promise of a higher-ranking role (with investor contact, and real equity) as the company grows. The real concern, when engineers participate in hiring decisions, isn’t about equity dilution associated with strong hires. They just don’t have enough equity for that issue to really matter. Their fear is that, if they hire people stronger than them, they’ll lose out on the executive positions implicitly promised to them as early hires at their startups. If they hire engineers better than they are, they’re doomed to languish in a company that has begun hiring above them. One note about typical startup sociology: once your startup hires someone from outside directly above you, it’s over. It will continue doing so, and your career has stalled out. You’re not a real player in the firm.
There’s an inherent conflict of interest in this style of VC-funded startup. The real owners (investors, founders, top management) get a social-climbing mentality and seek to hire stronger technical talent (even if they don’t need it!) for the sake of “scaling”. Engineers, on the other hand, would do well to subvert this. They don’t actually want to hire bad people (terrible software engineers reduce the productivity of the team) so much as they want to hire just-slightly-inferior people who aren’t so bad as to poison the team, but won’t challenge their chances of getting an executive position. This is more of a case of B players hiring B- players than one of them hiring C players.
However, two to five generations of “just slightly inferior” will turn to bad, and the rapid churn of VC-funded startups means that it doesn’t take that long. VC-istan startups degrade severely and predictably once hiring decisions are made by people with small equity slices, whose concern is not the health of the company (ownership) but the protection of their own inside track to executive positions. Such startups fail the Mike Test, and it doesn’t take long for it to show.
What is VC-istan’s place in the 4-culture taxonomy?
What is the culture of a typical VC-istan startup? Within VC-istan, there are efforts (e.g. Y Combinator) to generate a pay-it-forward guild culture. This is actually quite interesting– entrepreneurship is supposed to be market-oriented, but the most healthy subsector of VC-istan is guild culture– the healthy variety of a command culture. On the other hand, the healthy market culture (self-executive) is shockingly rare. This suggests that VC-istan has become somewhat of a command (not market) culture, and it seems that it would be more of a rank culture than anything else. I tend to believe that this is probably true. VCs talk to each other. They decide as a group who is hot and who is not, and this extortionate power makes them the true holders of rank. VC-istan, a postmodern corporation that manages to transcend mere companies– in VC-istan, companies are disposable– is so amorphous and complex that it cannot be tagged as “only” belonging to one culture, but rank culture is increasingly the dominant force. Unfortunately, I don’t see a solution for this. One possibility would be for the government to interfere with communication among VCs in order to kill off the collusion, herd mentality, and “accept this term sheet or I’ll pick up a phone and no one will fund you” extortions, but a government step-in would probably cause more problems than it would solve.
VC-istan itself may be a rank culture, but an imperfectly formed one. VC-istan exists because a web of socially connected people– reputable investors, and those with welfare-check (err, I mean “acq-hire”) writing ability at large companies– have created a controlled-market zone. Outside of that is regular ol’ business formation: an actual market. By the way, what are free markets? They are like self-executive cultures at best, but tough in their own way. Corporate tough cultures are mean-spirited and driven by intentional human malice (stack ranking, “calibration scores”, transfer blocks). Free markets are tough only because they are indifferent, but self-executive for those who’ve managed to develop a pattern of success. So the “greater world of business” formation is one that contains the market’s mix of self-executive and tough features at its periphery, and becomes more rank-culture-like as one grows closer to the VC-istan power players.
Startups are a reaction against the dysfunctional rank cultures, and would prefer to set themselves up as self-executive enterprises.However, as the company grows, power shifts to the rank culture of the world without (investors, acquirers). Full-on rank culture is usually in force shortly after liquidity, but the phase most typically associated with VC-darling “startups” is a transitional spell of (unplanned) tough culture.
Startups tend to remain “flat” for a long time, but this is usually not out of executive altruism. Self-executive cultures are naturally flat, but so are tough cultures. Tough cultures want everyone in competition with everyone else and use a (pathological) flat model to maximize internal competition. The self-executive model of the VC-istan startup dies as soon as the company stops handing out real equity slices (often immediately after the Series A). At this point, the “colony” mentality (live or die as a group) ends and the true goal of the new hires is to get into the executive positions that make real equity slices possible. The internal competition that emerges causes alliances to form and break, influence to be peddled, and informal management (people who manage to win credibility through illicit trades and manufacture the appearance of high performance) will emerge. The tough culture’s informal management is actually substantially worse than the rank culture’s rigid system, because the former encourages more influential people (managers in practice, if not in title) to compete with their (again, informal) inferiors. Rank cultures, at least, are rendered stable by the absence of competition between managers and subordinates– leading, over time, to the Gervais hierarchy. Tough cultures are just amoral, vicious messes.
When a startup first gets funding, the cultural neighborhood of the founders switches from the tough one of an indifferent market to a more self-executive one. They now have some autonomy, measured numerically in “runway” (how long they can survive with current capital). Unfortunately, this self-executive culture cannot survive the rapid growth typically expected by investors (especially VCs). Self-executive cultures can only hire near the prevailing skill level. VCs, on the other hand, want the company to indulge in social climbing (hiring above that level, which involves enticing people with executive positions that reduce the autonomy of those within) and rapid growth (which mandates hiring below the prevailing skill level to tackle the grunt work generated by sloppy, fast growth). Guild culture is clearly not an option, because tight deadlines leave no time for mentoring. Nor is rank culture, because it tends rapidly toward the underperformance of the MegaCorps that startups exist to destroy. Thus, VC-funded startups tend to degrade into a tough culture by default. This should explain the churn-and-burn behavior for which they are so infamous.
So, what culture is VC-istan? It depends. If you’re lucky enough to win the attention and mentorship of the (extremely rare) well-connected person who’s not an asshole, it’s a guild culture. If you’re a founder who just got funding, or a very early hire guaranteed a strong position, it’s self-executive. For founders and executives fighting mostly external battles (with acquirers, investors, and others with a million times more in the way of connections) it’s a rank culture, but one that does not (at that point) seep into the organization. For typical engineers facing internal battles (for scarce future executive positions that come with real equity) it’s a tough culture of long hours, harsh deadlines, vicious politics, and fast firing.
In other words, startups go through four cultural phases. Before funding, they live in the prevailing, indifferent tough culture of the market. Once they get their initial funding and have some autonomy, they turn self-executive (second phase). The romance of the “startup ideal” is derived from this phase of organizational life. As they grow (often sloppily) and take on more people than they can really provide startup perks (real equity, autonomy, leadership) for, they turn into tough cultures (third phase). Most VC darlings have tough cultures. By the time the company is getting that much attention, the self-executive era has ended. Finally, once the startup reaches liquidity and is a full-fledged corporation, it tends toward typical rank culture (fourth phase).
Mike Testing VC-istan to predict its future
Let’s see if we can apply the Mike Test to VC-istan, noting that Silicon Valley emerged as a reaction to the stodgy rank cultures associated (at the time) with East Coast corporations. The Mike Test tells us that self-executive and guild cultures can improve as they grow, although these low-entropy work cultures are hard to grow fast. Rank cultures, on the other hand, tend to stagnate, and tough cultures actively devolve, as headcount grows. What is the future, then, of VC-istan?
Most VC-istan companies are in active cultural devolution, being tough cultures where people are prone to hire inferior versions of themselves. In software engineering, this process is slowed somewhat by the (genuine) desire not to hire outright incompetents. Software is structurally cooperative enough that the pain (bugs, bad designs, low code quality) associated with hiring an incompetent does not offset the gain in relative position. As I said, this software-specific trait means that B players don’t try to hire C players, but B- players, which slows the Mike-Test decay normally associated with tough cultures. All of this is, in fact, not a major problem for VC-istan. Companies are disposable! The thing should be sold before it falls that far. VC-istan can tolerate the corrosion of the tough corporate cultures it generates, since these companies are just going to be sold to rank-culture corporate behemoths. VC-istan is not designed to build free-standing companies. They exist to be bought or to die.
The tough cultures of VC darlings will bring incompetents into those companies, but not a rate that is fast enough to matter to the corporations themselves. They’re build-to-flips that have no reason to care about slow cultural corrosion. There might be a “littering effect”, if these tough cultures are (a) bringing undesirable people from outside of VC-istan, and (b) leaving them in the VC-istan ecosystem after the company fails or is bought. If both (a) and (b) are true, then we can hold these cultures responsible for lowering the overall quality of people in it. I don’t know if that’s the case. That subject needs further study. Since the most egregious incompetents in VC-istan are not low-level engineers but the supernumerary, non-technical, executives, I tend to doubt that any engineering-specific littering effect is a problem, either in the short or long term. The executive-specific one is well-documented; VC-istan does have a way of turning shitty, failed bankers into even shittier non-technical startup “executives”.
In other words, I don’t know if the tough culture that typifies the standard VC darling is going to fill VC-istan with incompetents. The company itself will take on undesirable people, but those companies aren’t built to last very long. That tough cultures are so common is a symptom of something pathological, but that’s a topic for another essay. Instead, to project the future of VC-istan, we need to look not at the culture of the typical VC-funded company, but at the culture of VC-istan itself: the postmodern corporation in which the VCs are executives, so-called “CEOs” are glorified project managers, and engineers are clueless chumps who don’t realize they work for a big company. What we see is a very typical rank culture, with VCs and well-connected “serial entrepreneurs” (read: people whose connections entitle them to continued funding no matter what happens) in the Sociopath tier and a swollen Clueless tier. It’s a very effective, internally stable MacLeod rank culture.
As I discussed earlier, the stability of the MacLeod organization comes from the lack of envy between Losers and Clueless; and between Clueless and Sociopaths. Differential social status (DSS) and the effort thermocline (the level at which jobs become easier, rather than harder, as one ascends) ensure this stability. Losers prefer things of genuine value over the non-transferrable DSS coveted by the Clueless. Clueless are oblivious to the effort thermocline and consider the Sociopaths above them to be harder working and more capable than they are. Losers who want to become Sociopaths are fast-tracked up or out of the organization. How does this work out in VC-istan? The effort thermocline is blindingly obvious: it’s the distinction between startup employees and powerful investors. Above the thermocline are the reputable VC partners, whose social connections entitle them to an easy life; below it are founders and the people they hire. The founders are the upper-tier Clueless who (just as in a MacLeod organization) don’t want the jobs above them, on the assumption that the Sociopaths “just shake hands and push paper around; we do the real work”. Differential social status, in VC-istan, is access to funding and publicity– something that a generation of idealistic idiots has spent decades of hard work to get, while their peers “sold out” and made actual money that could be used to buy actual houses and vacations.
For all this, I used to think that VC-istan didn’t have MacLeod Losers (with the losses borne by the swollen Clueless tier) but I realized that I wasn’t looking hard enough. Clueless are true believers, while Losers are rationally disengaged people who work hard, but minimize discomfort. (Again, they aren’t actually losers per se. Unlike the Clueless, they know the trade they make.) In VC-istan, MacLeod Losers tend to be consultants and freelancers. They cherry-pick the work that’s interesting to them, can lead quite a nice lifestyle if they’re good, and never work for equity. Since these people can often command $250 per hour or more, it’s hard to call them Losers! What they are is checked out of the Clueless game of VC-istan. They will never get rich– it’s hard to get more than 800 hours of work per year, with a decent rate, as a freelancer– but they will never lose a vacation or a relationship to a get-big-or-die company either. They sell tools and water to the idiots who dig for gold in the 120-degree heat. The MacLeod Losers of VC-istan are the mercenary consultants who manage to get by, contribute some work, but never hitch their fortunes or make undue sacrifices for a specific company.
If VC-istan is a MacLeod organization, then what is its future? MacLeod rank cultures decline, but they do it slowly. That is also what will happen to VC-istan. I couldn’t possibly say whether it will happen over the next 2 years or the next 20, but VC-istan is already in a MacLeod state and, while its decline is likely to be gradual, it’s also inexorable. I’ve started calling many of these companies “ad-banking”; VC-istan is what investment banking was in 1995. That is far from “death”. It does not even mean that it will become impossible to get good people; investment banks are able to get good people even now, but they pay dearly. It only means that good people will become more expensive over time. Only when compensation ceases to grow exponentially (in banking, circa 2009) will the best people start to trickle out and look for something new. VC-istan engineer compensation has quite a few years of 10-20% annual growth before reaching the unsustainable level, so I don’t see its “death” setting in until about 2025. Until then, there’ll be a lot of money to be made by mercenary engineers, so long as they know the market well enough to play it.
What will make VC-istan’s unwinding process interesting is its generation of alternatives. As investment banking grew stodgy, boring, and difficult, banks raised compensation (for genuine talent) into the stratosphere. No one would spend 15 years in New York investment banking for less than $400,000 per year at the end of it. This transfer of wealth generated early retirees and a few hedge funds, but not rival investment banks. People did not take their $6-million nest eggs and launch rivals to Goldman Sachs. VC-istan’s story will be different. As talented engineers wake up to the scam, wages may rise to the same levels that bankers commanded, generating a transfer of wealth toward a set of people not typically associated with richness: hard-working computer programmers. Those among them who have a mind for business will want to participate in investment, and to build something different from VC-istan. Something better. I don’t think anyone knows– yet– what will be built.
So what should be built? At 7.7 kilowords, I can’t hack that now. That’ll be covered in a future essay.
