CEOs Don’t Steer

There is a pattern to the most influential business writing, in The World is Flat league. Especially writing that CEOs seem to like enough to exhort their organizations to read. Every such work offers one big, unqualified, unquantified, universal proposition. Usually with an obvious black-and-white moral assessment attached as an implied parenthetical [and this is a good thing]. The proposition will typically offer a big generalization covering a really vast range of things going on in the environment: An extreme, if very lossy, compression.

There are no if…then…else conditions attached. There are no temporal markers or spatial delimiters like this will be true between 2017 and 2022 in the developed world.


The World is Flat [and this is a good thing] 


Under Certain Assumptions, the World Will Likely Continue Flattening for Approximately at Least Another Decade, and This Is a Probably a Good Thing.

This pattern isn’t mere rhetorical pithiness in the title or a distaste for weasliness. It permeates the entire idea being offered. And it exists as a consequence of a CEO trait:

CEOs Don’t Steer [and this is a good thing].

Big business ideas are the way they are because they are designed to feed and nourish this CEO trait. It’s a proposition that, at first sight, sounds both wildly untrue and something that would be really bad if it were true.

Other kinds of leaders steer. Political leaders steer. Military leaders steer. Investors and central bankers steer. Artistic and creative leaders steer.

But CEOs don’t steer.

When I read or listen to exemplars of other categories of leaders, I am often impressed by the subtlety, nuance, agility, and complexity of their thinking.

This is not true of most things CEOs say or write.

Public displays of CEO thinking are impressive primarily for their sheer banality, far above and beyond the needs of non-offensiveness, political correctness and perception management. You can tell it’s coming from deep down. It’s not an act. I’ve seen it on display in candid, private settings as well, where there’s no particular reason to keep things simple. CEO world views really are that simple. That does not mean they are simplistic or entirely a consequence of survivorship bias and attribution errors.

When CEOs “pop” in the public imagination, it is almost always due to an energizing emotional or aesthetic quality in things they say/write/show/demonstrate. Not cleverness, complexity, or intellectual depth. And again, it’s not marketing. It’s a symptom of how they think: in simpler ways than the rest of us. Other kinds of leaders are generally playing a higher-dimensional kind of chess than the people they are leading or influencing. CEOs are playing a lower-dimensional chess.

Though the banality not an act, neither is it an intellectual or character failing. On the contrary, it is a strength, and possessing it is the primary job qualification.

Resistance Must Be Futile

You don’t want a CEO to be an intellectually stimulating bon vivant who can float like a butterfly, sting like a bee, wafting through rich landscapes of thought, making subtle subcultural references and in-jokesThat’s a job for people like me in the chatteratti class.

You want a grinding, relentless, tank-like drive in a seemingly unshakeable direction that builds shareholder value. If you’re in the way, you get crushed. If you’re off to one side, you are ignored. You’re looking for a resistance is futile affect. Terminator, not Tarantino. Force of Nature, not Beautiful Mind.

CEOs may make many decisions but they are not steering decisions. They may set a course when they first ascend to the job, and temporarily maneuver in unstable ways to restore normalcy if they get knocked off course, but that’s it.

CEOs don’t steer. All apparent exceptions will be explained shortly.

They may help a report solve a problem or remove an obstacle or make suggestions about how someone else should steer. They may even underwrite a steering decision proposed and executed by others. Even ones they don’t believe in, as a hedge against being wrong.

But in their own role, they don’t steer.

By which I mean “execute risk-managed and planned turning maneuvers based on foresight, predictions, and map-like mental models of evolving action in a complex environment.”

This does not mean no steering happens in the company or that steering is unnecessary. If any kind of ongoing steering is obviously necessary, it becomes a function at a lower level than the CEO, associated with a suitable kind of mapping, sense-making, and navigation skill. CFOs navigate the capital markets and steer around financial trends. CTOs navigate technology trends. CMOs navigate a changing media landscape and power-optical conditions. Ideally, steering functions settle as low in the organization as possible. In the best case, in automated, feedback-controlled, self-driving layers.

There is a massive stack of steering going on throughout the organization, except at the very top.

Nor does it mean that CEOs can’t steer. Most CEOs get to their jobs via some good, but not exceptional steering performances along the way. If they are exceptionally good at a particular kind of specialist steering, they generally top out in a role below CEO. What is a strength in leadership roles below the top level is a liability at the very top.

So once they reach the top, if they know what’s good for them, CEOs stop steering. If possible, they break off and throw away the steering wheel in a show of directional commitment. Why?

Because the actual job is the opposite of steering.

The Opposite of Steering

At the level of abstraction at which a CEO views what is going on, direction does not change, and more importantly, the CEO’s job is to make sure it doesn’t.

CEOs are orientation locks. The opposite of steering is orientation locking. To enforce Newton’s first law, the inertia one, on dancing human systems inclined to violate it. Contrary to popular belief, it isn’t inertia that’s the problem for most companies, it is lack of sufficient inertia in the right direction. Enough to punch through any resistance that might be encountered. And the reason they lack the inertia is that CEOs aren’t steady enough in their jobs as orientation locks, providing a steady True North signal to everybody else doing more local kinds of steering work.

The primary CEO function, and the trait the good ones are selected for, is to provide the gyroscopic stability required to keep a company vectored in the chosen direction. They end up in the jobs they do because they counterbalance an organization’s natural tendency towards distraction, ADD and momentum dissipation. A typical company is a wandering, wobbling hive mind, liable to spend all its time chasing distractions if you let it, before dissolving into a bunch of clever tweets about crappy prototypes.

As the orientation lock, the CEO becomes the human locus where momentum compounds; the psychological platform others build on. They are the steward of whatever snowballing network effect or unleashed natural wealth-creating dynamic is the company’s raison d’être. Their primary job, and ideally their only one, is to protect and feed that dynamic, and get everything else out of the way.

Every act of steering leaks or drains at least a small amount of momentum, no matter where in the company it happens. But steering at the CEO locus truly hemorrhages momentum, creating serious, possibly existential, vulnerabilities. Because by definition there are no systems for doing it well. If there’s a system, it would have kicked in before things got to CEO level. Worse, steering at CEO level might actually kill whatever compound-interest dynamic is driving the whole show, killing not just live momentum, but its source of renewal and growth.

If you didn’t have a CEO serving as an orientation lock, you’d have governance by some sort of bickering executive committee. Sometimes star-bellied sneetches would win, sometimes crescent-bellied sneetches would win.

Result: the irresolute Brownian motion of (say) a Parliamentary democracy. No compound interest. No network effect. No platform effect. No force-of-nature unleashed. No dent in the universe.

If the orientation is not locked, no wealth-creation dynamic is unlocked.

This is why CEOs are different from other kinds of leaders. Leaders in other societal roles are typically not the stewards of any source of promethean energy. Instead, they are involved with directing and routing second-order effects that drain it. So they can afford to steer and be clever about it. They are zero-sum carve-the-pie leaders (which is important too, just not the point of this post).

But CEOs don’t steer.

What is good for countries, militaries, and art movements is not good for companies (and conversely, when the Trumps, Modis, and Xi Jinpengs bring CEO-like orientation-locking tendencies to inclusive governance jobs that require non-trivial steering, the results are usually not pretty).

When Companies Must Steer

So what happens when a company encounters a situation outside the domains of functional expertise of the available stack of sense-makers, map-readers, and helmsfolk reporting up to the CEO? What if there is an Maximally Exceptional Steering Event that only the CEO can and must navigate, and cannot delegate. When the company in its broadest, most inclusive sense, needs to change direction?

The answer is probably not what you want to hear if you’re a CEO, but it is more often true than not.

When a company needs to truly steer at the CEO level, it changes CEOs.

A “steering” event at the CEO level is a reorientation (in the OODA loop sense) for the entire company, and for reasons I’ll get to, the best way to accomplish it is to replace the CEO with a new one who is already locked on to a more desirable orientation vector. Like say, Dara Khosrowshahi, who, based on his history, was apparently already locked on to the “don’t be a misogynist asshole” vector that Uber needed before he got the job.

If a direction change event is more complex than substituting one set of foundational orientation/vectoring beliefs for another, you might require some temporary high-agility steering between two locked orientations. Under such circumstances, a company usually hires a specialist turnaround CEO.

“CEO” is probably the only job in the world where any non-trivial steering events are handed off to temporary specialists. The only other such job I can think of, not coincidentally, is ship captain, which is basically a CEO job, so not really a second example. When large, heavy ships enter or leave ports, local harbor pilots must be hired to do the steering, by law. And while they’re on board, they have the authority to overrule the captain on many matters.

Captains don’t steer. Underlings steer during normal times, harbor pilots steer during exceptional times. CEOs don’t steer. Underlings steer during normal times, turnaround/caretaker CEOs steer during exceptional times.

You could even say a CEO is like a railroad engineer driving a train, except the tracks are in their mind rather than in the environment.

When the environment changes in more profound and permanent ways, the steering happens in the guise of an entirely different company: a disruptor company from the margins that is on a better vector. Often one founded and staffed by veterans of the original company.

Incidentally, The Innovator’s Dilemma is among the (very) rare bestselling business books that is not reducible to a single declarative propositional-logic assertion like The World is Flat or IT Doesn’t Matter. And if you think about it, its core message, notwithstanding Clay Christensen’s brave later attempts to offer an “Innovator’s Solution”, is essentially a close cousin of mine:

Companies Don’t Steer.

What looks like steering to people inside the company, in specialized steering roles, is really more like growing or changing shape. Sort of like how snakes slither along approximately a straight line, but individual bits of their bodies look like they’re steering if you zoom in enough.

More precisely, the innovator’s dilemma in its stronger, truer, impossibility/limit form is the following idea:

If there is a circumstance requiring the CEO to steer, the market will almost always do a better job. Either by changing the CEO via the labor market, or energizing a disruptive competitor.

In this view, CEO tenure is purely a function of volatility in market conditions. In calm conditions, the median CEO will last a long time. In turbulent conditions, they will last a shorter time. The tenure of a CEO is like the mean free path of atoms in a gas.

The rare legendary CEO who bucks the trend of tenure being a function of general environment volatility is usually somebody who has found a truly powerful, long-lasting wealth-creation dynamic to defend, and is smart enough not to try to steer it, only fuel it.

But unlike atoms zipping along between collisions, sticking to a straight line path is not easy as a CEO.

Not Steering is a Real Job

Sticking to a straight line is actually a very difficult job for a human.

It’s hard enough to do it when you only have to deal with the distractions generated by your own senses and mind. When you have to do it for the distractions being generated by hundreds or even thousands of curious people, you need a brain with more than tank-like directional inertia. You need aircraft-carrier levels. Starship levels.

Star Trek actually gets this right. Most steering is done by a pilot character. At most the captain might order “evasive maneuvers” or “attack pattern beta 4-7.” Most of the truly captainy decisions are about punching through troubles in a straight line. Not steering when everybody wants to steer.

It takes enormous discipline to not steer. Especially because people who end up in the top job usually do so via lower-level roles requiring some sort of specialist steering or the other (or in the case of startups, the high-pivot pre-product-market-fit phase, where the CEOs are auditioning to be real CEOs the way their companies are auditioning to be real companies).

One pernicious mode of CEO failure is addiction to some sort of steering they learned to be too good at on the way to the top (or less frequently, one that they were bad enough at to develop insecurities about). The temptation is to micromanage or take over the helm at some specialized level that is probably best left to someone else.

This does not mean the CEO never takes over a helm position, but such an override is usually only called for when the report is doing something that might derail the company from its main locked-orientation vector. Since CEOs usually only need to do this 1-2 levels down at most (other CxOs or VPs), it means the override event is either a challenge to authority, serious misalignment on direction with a serious executive, or serious incompetence (at high enough levels where a CEO rather than a low-level team manager or supervisor might be overruling you, you can’t afford to be wrong too often; the room for error is very low; you have your job because you’re “right a lot”).

In all cases, needing to be over-ridden via the CEO taking over a steering function is usually a firing offense. And the CEO only remains at the helm as a pinch-hitter in the lower-level role until they can fill it again with a more simpatico, less threatening, or more competent hire.

CEOs don’t steer, but they do make aiming decisions — usually they get one shot at “aiming” the company during the honeymoon period after they first land the job — but once is usually it. If it looks like they’ll hit the target, they’ll get a chance to set the next one as well.

This is a good thing.

CEOs make acceleration decisions when they can, braking decisions when they must, retargeting decisions after successes, but steering decisions only when the alternative is losing the job (and the company) no matter what they do. And generally they lose the job anyway.

Growth decisions fit within this logic too. Mergers and acquisitions are like adding engines or dropping deadweight, not steering, though they may cause turns as a side effect. Adding new lines of business is about adding to the momentum along the chosen vector, not steering (hence “sustaining” over “disruptive” innovation).

The Perils of Steering

Why not steer? It seems perverse to not intelligently twist and turn through some sort of good map, and it is — at every level below and above (board and market) the CEO. Agile artistry is a good thing almost everywhere except for at the CEO locus.

For the CEO, steering is a perilous thing, almost certain to do more harm than good. If you’re lucky, only your bold “change management” initiative will fail and you can fire a few people, cancel the program, write off some investments, and get back on track.

If you’re unlucky, attempting to do clever steering will crash the company.

A hint as to why not-steering is rational can be found by examining the risk/uncertainty structure of a “full” steering decision. A full steering decision around some conceptual obstacle or landscape feature has a form something like:

IF X then {LEFT and after T1, RIGHT}, ELSIF Y THEN {RIGHT, and after T2, LEFT}

There are at least 6 variables you can get wrong: 4 turning decisions where you can under or oversteer, and 2 timing decisions where you can be too early or too late. There are also 2 testing conditions that can yield false positives or negatives.

Take something like a Gartner hype cycle. You can think of it as a challenge driving course. It has 4 timed “turns” to steer around, and there are at least 4 ways to get each of them wrong (too soon/too late, understeer/oversteer).

Even if, very optimistically, with fairly generous error bands, every qualification or quantification only weakens a steering hypothesis only by 50% (ie there is a 50-50 chance of getting that decision right), the possibility of error snowballs. You have only a 1 in 16 chance of navigating a hype cycle right. 15 out of 16 times, you’ll fail in a way that threatens to kill the company’s core wealth-engine dynamic.

Navigating it wrong doesn’t matter if it a sufficiently low-level steering challenge. A company can survive an underwhelming marketing campaign, a failed skunkworks project, and even a failed product launch. But at the very top, a failed steering episode is usually a CEO-killer, and often a company-killer as well.

And it only grows more complex as you add more complex topography to the map.

By contrast a simple and robust proposition like The World is Flat offers a very robust kind of decision filter that helps stay on a straight line. Because it is a single true/false proposition, you’ll be either be right more often than you’ll be wrong, or you’ll crash-and-burn quickly. If a boolean proposition is more true than not, then all instances of its use add “rightness” in parallel rather than compounding cascade-failure scenarios in series.

The risk structure of a steering decision that is not a fundamental reorientation is complex and fragile, but is also much more amenable to being mitigated through focused, specialized attention, data, and intensive sense-making and mapping efforts.

In other words, it’s a job for somebody else.

The CEO’s job is to stay vectored until they run into a situation that they can neither punch through, nor turn into somebody else’s job. Then it’s game over unless the board grants you another life because it can’t find anyone better.

A Very Small CEO Shell Script

Almost everything about the job of not-steering has to do with managing energy and emotions, but let’s get the decision-making part out of the way.

The CEO algorithm is fairly simple in the intelligence it demands. If running this algorithm were all the job entailed, you could replace most CEOs with very small shell scripts. The better and more powerful the CEO, the easier the replacement.

There is a reason why Alan Turing framed his eponymous famous test in terms of a computer mimicking a “mediocre intelligence, like the President of AT&T.” More intelligence than is necessary for running the straight-line trim trajectory algorithm (which many CEOs, unfortunately, do have) is a liability that turns the leadership environment into a distraction-rich one that derails sparkling genius minds more readily than it does merely mediocre ones.

Forget being “data-driven” or even first-order-predicate-calculus-driven. Good CEOs run on propositional logic. True/false, go/no-go, push-harder/ease-off.

For thinking at their level, they do not like to use the sort of quantified or qualified logic with if-then thinking and statistical analysis that steering requires.

They have to be capable of it at a rudimentary level, so they can constantly see the unfolding tapestry of hundreds of paths-not-taken, and monitor/inspect the work of reports who need to run on more complex kinds of logic (such as first order or higher).

But almost all the time, they have to not-steer.

So the CEO algorithm not-steering algorithm basically looks like this:

  • Filter everything as either relevant or not relevant (this is where books like The World is Flat or The End of Power come in handy). Black and white. No fuzzy logic or Bayesian bullshit.
  • Almost everything is not relevant, and almost all of it can be judged to be not relevant by an somebody else. Only the edge cases need to bubble up. Peripheral vision is critical for steering jobs, but over-rated for the staying-vectored job.
  • For things that are relevant (and actionable), it is either a steering decision or a not-steering decision. If any kind of map exists, it is a steering decision and belongs to whoever owns the relevant map.
  • If it is a steering decision, it is likely one that belongs in a functional pigeonhole. Delegate it to somebody lower down with the right maps, sense-making skills, and specialized situation awareness.
  • If that person or functional competence does not exist it’s a hire-and-learn decision.
  • In either case, understand and inspect what functional point-people do well enough to assess whether or not their steering activity is a threat to the don’t-steer vector, but otherwise leave it alone.
  • If it is not a steering decision, it is usually an acceleration or braking decision (advancing and delaying things are degenerate stop/go versions of acceleration/braking decisions, as are “punch through resistance” decisions about obstacles and “harness this as a tailwind” decisions about new resources, external developments, and novel phenomena).
  • Less frequently, it is a drop/add decision: adding or losing momentum through inorganic, surgical means.

The rarest kind of decision is a steering decision that does not belong in a functional pigeonhole. Some broad-based change in the environment that offers no obvious “make it somebody’s job” approach to getting a handle on it. Something that requires reimagining the business.

Such a decision is a reorientation steering decision. An existential threat for the CEO, and possibly the entire company. The reorientation will most likely happen via the CEO being replaced or a disruptor taking over the market.

Like most people, I resisted this conclusion for a long time because it suggests the null hypothesis that most of what people do to support or influence CEOs in their jobs is bullshit.

Most of the time, in the face of an existential thread, there will also be some sort of foolish attempt at a “change” theater. Because everybody will want to see some steering being done so they can stop panicking. And as most business leaders know thanks to John Kotter’s decades of evangelism, most change management efforts fail.

What few will admit is that this has nothing to do with poor change management capabilities.  Most efforts at change fail because they are either unnecessary, or because they are impossible.

I’ve spent a depressing fraction of my career aiding and abetting change theaters around reorientations that ended up ultimately happening through CEO changes or disruptions. Sometimes because I gullibly bought in, sometimes because it was a way to make a cynical buck when I needed to.

Increasingly, I just say no. If it’s really an existential threat, change the CEO or go work for/invest in the disruptor instead. If it isn’t, it’s probably best to do nothing. Just resist the temptation to steer, or give the CEO steering ideas.

It’s a weird belief for a consultant. Most consultants wishfully hope there was a lot more steering called for, and going on, at higher levels. A popular idea is that ready, fire, aim leadership ought to be replaced with more enlightened ready, fire, steer leadership (a Paul Saffo line I believe).

I’ve come to believe this is wrong. Steering is for markets and employees and robots. CEOs should stick to being orientation locks.

Being True North personified sounds like a dull and exhausting job to me, but clearly somebody’s got to do it.

A Very Big CEO Hammer

If the decision-making algorithm that is 10% of the CEO job, the part requiring intelligence, is simple and reducible to being a very small shell script, the other 90% is reducible to being a very big hammer. CEOs must speak robotically and carry a big hammer. Or rather, be a big, robotic hammer.

The biggest part of the job of a CEO is to stay on course without steering, by managing tempo: energy, emotions, and rhythms. Ideally, the energy is snowballing, the emotions are well-regulated at the right point on the Yerkes-Dodson curve, and rhythms are steadily accelerating to break through the last remaining resistance on the path to total dominance of the market and Bond Villainhood, complete with underground lair.

When CEOs are doing this job well, they look like out-of-control assholes.

They derp. They preach. They steamroller over people. They are painfully blunt and black-and-white in their responses to things people say. They make no effort to soften the blows they routinely deliver. They reduce people to tears. They appear to listen, but then they just do their own thing. When they exercise power, they make sure you know it.

When James Bond asks, do you expect me to talk? They say, No Mr. Bond, I expect you to die.

They hammer parts of their own simple bot-algorithms into others in the form of do-as-I-would aphorisms every chance they get. They say no so much it takes a strong will to be around them without getting conditioned into permanent negativity. There is a bot-like predictability to how they respond to anything, a predictability that induces an entire manage-and-contain culture around them.

Which is exactly the point.

The converse is not true. Most people who display all these traits are merely garden-variety assholes headed off some personal cliff or the other. This is the well-known Steve Jobs fallacy: Steve Jobs was an asshole, therefore if I am an asshole I can be Steve Jobs.

How do CEOs get to this condition? By eating right.

The Care and Feeding of AlphaGo, CEO

Though CEOs are only mediocre at using information to steer, they are excellent at using it to pattern-match and classify to fuel propositional-logic one-step decisions of the Yes/No or In/Out variety.

The combination of 10% running a simple decision-making algorithm with 90% managing rhythms, emotions, and energy looks suspiciously like one of those inscrutable new-fangled deep learning algorithms that just digest piles and piles of data and use it in relatively simple ways to make decisions.

Information processed the right way, is momentum. This is actually obvious when you consider the WorldIsFlat class of inputs that CEOs devour.

I call it a Big Data pattern: simple algorithm, more data over complex algorithm, less data. And the big data (relative to a human-brain scale) is really piled on by CEO-fodder books and essays. Pages and pages of anecdotes, statistical nuggets, microtrends, sketched (and sketchy) arguments, and evocative vignettes. Just when you think Tom Friedman cannot possibly interview one more cab-driver to show that the world is getting flatterererererer, he does.

The pile of data isn’t meant to refine the big claim being installed, or add nuance to it. It certainly isn’t meant to attempt to falsify it. The crushing weight of evidence is designed to confirm and reinforce the One Big Truth and install a pattern-match filter to navigate by it.

The thought hit me as I was reading Moises Naim’s End of Power (2013). I was reading it not because I found the thesis intriguing or novel (it’s basically a combination of The World is Flat (2005) and The Sovereign Individual (1997) viewed from the other side, 10-20 years later), but because I figured it might shed some light on how Mark Zuckerberg is navigating his political troubles, given that he recommended it a few years back (short answer: it does).

I could never write these sorts of books or essays. It requires far too much data-collection addiction. The closest I came was unpacking Marc Andreessen’s software is eating the world at some length. And I did that in my usual maps-and-concepts-heavy way rather than the pile-on-the-data way. It is fodder for the steering classes, not CEOs. All CEOs need is the original Marc essay.

When a small-shell-script-big-hammer CEO has been properly fed for enough time, you get a CEO who can project a reality distortion field and lock orientations.

Vectors and Realities

In case it wasn’t obvious, this post has been something of a parody of the business writing form I’m talking about. Not a full-on parody because I lack the energy to talk to a thousand cab-drivers to pile on the necessary amount of overwhelming weight of inconclusive circumstantial evidence.

And also because I basically believe it. CEOs don’t steer. Seriously.

I learned this the hard way early in my career. When you are trying to persuade middle-managers about something, you need relatively sophisticated, data-rich maps, and competence in some sort of domain-specific sense-making and steering. A Gartner hype cycle is a typical middle-management artifact. Fun for, say, a program manager to steer around and nerd out over. Navigating it wrong doesn’t matter if it a sufficiently low-level steering challenge, but at the top, it is basically a CEO-killer.

When you talk to non-CEO senior executives, any map more complex than a 2×2 is too complex. You can roll more detailed, complex maps up to that level (for example, the famous OG business 2×2, the BCG growth-share matrix, is a higher-level abstraction of experience curves, which are more data-rich middle-management fodder).

And when you talk to a CEO, any map more complex than a 2×1 is too complex.

Either the world is flat, or it is not.

Either software is eating the world, or it is not.

Either IT doesn’t matter, or it does.

Either power is ending, or it isn’t.

Either digital transformation is bullshit, or it’s not.

The mental world a CEO inhabits is a handful of such boolean, propositional logic beliefs that create a very strong reality filter wrapped around a whole lot of momentum. The job of the filter is to protect that momentum.

That momentum takes the form of a massive set of filtered beliefs about the world, creating a reality that is rich enough for the entire organization to inhabit. You’ve heard A. G. Lafley’s famous line that the job of the CEO is to interpret external reality for the company. Now you know what it means: interpret the entire world through a bunch of WorldIsFlat class filters.

It is the momentum of true belief, and there is enough of it to steady (not steer) the course of the entire organization. Sometimes with tens of thousands of intelligent and empowered creative information workers, all steering on their own maps and potentially funneling a tsunami of momentum-destroying distraction towards the top.

This combination of a simple 10% CEO algorithm with a 90% hammering process turns the entire company into an extension of the CEO. So that at least a portion of the potential tsunami is handled at the source.

One reason  (though not the main one) the CEO shell script needs to be simple is to allow at least a partial copy to be installed in everybody’s head through frequent reinforcement.

What would the CEO do? is a dangerous decision-making heuristic to install company-wide because it extends the CEO’s blindspots all the way through the company. But done to a judicious extent, it creates the right kind of tunnel vision to make not-steering at the top a manageable job. Shared blinkers are as important as shared truths. What nobody sees is as important as what everybody sees. This is what it means to project a reality-distortion field.

If everybody were to participate in the business of Lafley’s “interpret external reality for the company” you’d get constant chaos.

This runs counter to everything many people sincerely believe about the value of employee empowerment and autonomy, and turning the entire company into some sort of innovation hotbed running on “20% time” (a bullshit idea I believed in for far too long). The truth is, if you actually tried doing things like that, you’d turn the noise and distraction levels up to “impossible to manage.” If you think management by committee is bad, imagine what happens when the entire company turns into one giant committee. If somebody actually has an idea worth a “20% time” project, it’s probably best if they leave and pursue it as a startup idea that might potentially grow up to be a disruptive challenger.

A Lemon Wrapped Around a Large Gold Brick

What is it like to be at the center of this system of focused energy and blinkered vision non-steering? I hope to never find out personally. CEO frankly sounds like a completely awful job. But I do know what it feels like to run into it up-close, especially if you’re in the way.

Douglas Adams described the feeling of drinking a Pan-Galactic Gargle Blaster as “being hit in the head with a slice of lemon wrapped around a large gold brick.”

That’s what it feels like to tangle with a CEO who knows that the main job is not-steering and is doing it well. A reality distortion field is far more potent than mere culty kool-aid. That merely kills. A reality distortion field is a slice of lemon wrapped around a large gold brick that turns you into an effective zombie, capable of amplifying momentum in the desired direction.

I’ve wondered for a while whether or not the external explanation — replacement-or-disruption — is enough to account for why CEOs don’t steer, and I’ve concluded it isn’t. There’s an internal reason CEOs don’t steer:

Steering kills reality distortion illusions.

See, CEOs represent the ultimate kind of scarce human resource: meaning. Unlike most of us, when CEOs ask why something is worth doing, they don’t go down a path of infinite regress, questioning all notions of meaning and significance, until they end up at nihilistic existential dread.

No, they stop at the first simple answer they find, lock on, and focus on building momentum around it. They cut away everything that might prevent it from compounding, increasing the free energy of the whole. They install little copies of themselves in every brain so the whole thing starts to cohere like a laser.

To be a CEO, especially of a large, profitable company, is to embody the awesome power of a fairly trivial set of big, unqualified beliefs wrapped around an extraordinarily non-trivial amount of raw momentum and generativity. If the trivial beliefs are mostly right, most of the time, the result is powerful: a reality distortion field that contains a source of shared, non-null meaning, a mission worth believing in.

But this only lasts so long as the CEO does not steer. Not-steering is the central behavior needed to maintain a reality capable of generating belief, commitment, and meaning. Steering is a sign of being human and capable of error, less-than-omniscient understanding of the world, and less-than-omnipotent agency over it. To distort reality, you have to play god, and gods only need to aim their little universes once. Steering ruins the illusion, undermines the foundation of belief, and loses the crowd.

Because CEOs find meaning to be a simpler thing than most, they don’t overthink it. So the rest can simply replace their own why questioning with a because the CEO says so. They stop at the first simple answer that justifies their chosen direction of steamrolling, and then start spinning up momentum. More, and more, and more. Every possible tailwind harnessed. Every barrier smashed through and turned into a moat. Every new piece of data cast into a belief structure that strengths the reality distortion field.

What matters about a source of meaning is not the semantics of what it “says” but how much generative energy is coiled up and continually exploding behind it. This is why the most successful companies necessarily have a cult-like quality, induced by the reality distortion field of the non-steering, omniscient, omnipotent leader. You may know it’s an illusion, but it works nevertheless.

Once in a thousand CEOs, the result is a Jobsian dent in the universe.

The other 999 times, they crash and burn, and instead of a dent in the universe, they leave an Ozymandias head in some barren desert somewhere.


CEOs Don’t Steer.

And under certain assumptions, this will likely continue being true for approximately least another decade, and is probably, but not always a good thing.

It’s the endgame of neoliberalism. Whether or not it is true that a new CEO or the appearance of a disruptor is a better alternative for society as a whole than a CEO-steering, for the forseeable future, that’s the way things are.

So CEOs don’t steer.

Except when they do. Good CEOs steer less than bad ones. Big company CEOs steer less than startup CEOs. American CEOs steer less than non-American ones. But yeah, all CEOs steer a little.

But CEO steering is rare, and successful CEO steering is even rarer. Most of the time, the right decision is “stay the course” not “steer”.

When true CEO-steering does happen, it usually involves intense death-and-resurrection level trauma for the CEO in question.

Often involving divorces, PTSD for people standing too close, and mild-to-severe trauma for the corporate body politic. Trauma that sears the experience into organizational memory as a hell-and-back journey.

So if you see a CEO start to steer, you should probably run.

Get Ribbonfarm in your inbox

Get new post updates by email

New post updates are sent out once a week

About Venkatesh Rao

Venkat is the founder and editor-in-chief of ribbonfarm. Follow him on Twitter


  1. I’ve always thought of this as a question of leverage. A CEO is trying to control the efforts of perhaps a thousand people. Most of those people have a strong bias toward stability. The people who really matter to the performance of the company are separated from the CEO from by 3-5 layers of management, each of which takes the input from the layer above it with a grain of salt.

    Under these circumstances, a CEO is pushing on the short end of a lever. It takes a lot of force, consistently applied, to have any effect at all. There’s no need to worry that lack of nuance will cause the company to overreact to his ideas. He’s worried about being insufficiently impactful, not the reverse.

    • Good point, yeah. Relatedly, _not_ applying leverage is often not an option, since some actions are only available in a leveraged form. Like open communications. A public CEO statement or company-wide one is not a way to influence 1 specific person since it will impact everybody. By contrast a random individual can often disguise an action targeting a single person in a public statement.

  2. While generally true for large companies, there are also many exceptions where CEOs have successfully steered the company into new waters. (Jeff Immelt with GE and Digital, Reed Hastings with Netflix and Streaming, etc.).

    I think it also really depends on the individual personality of the CEO, and the cash situation of the company. When these factors are extraordinary, outcomes are unpredictable and more likely to favor the incumbent pulling through.

    • I don’t think there are many exceptions to the rule Venkat presented here. The two examples you gave are still two among thousands upon thousands of CEOs. On top of that, I’m not sure the Netflix example contradicts the point of the article.

      With Netflix, the important question to ask is what was Hastings’ original orientation? Yes the company pivoted from mailed DVDs to streaming, but that is only an orientation change if Hastings’ original orientation was around an axiom like “Make mailing DVDs as efficient as possible.” If, instead, he was oriented around a more general idea like “Increase ease of access to desired media,” then moving from mailing to streaming would be a natural evolution of the business, an obvious adoption of evolving technology to push the company’s core thesis(es). Similar to Bezos expanding into different product categories to satisfy Amazon’s “Give customers more options,” thesis.

      The GE example does seem like a real reorientation, but it comes after the company went through a near death experience during the financial crisis. The company almost died, then spent some time trying to sell off the assets that almost killed it, spent some more time selling off other assets to try to refocus the company, and then started a reorientation towards digital. Venkat’s model would imply a CEO change at that point, so maybe the board really does like Immelt. Maybe he would be one of the rare exceptions to the rule.

  3. “See, CEOs represent the ultimate kind of scarce human resource: meaning. Unlike most of us, when CEOs ask why something is worth doing, they don’t go down a path of infinite regress, questioning all notions of meaning and significance, until they end up at nihilistic existential dread.

    No, they stop at the first simple answer they find, lock on, and focus on building momentum around it. They cut away everything that might prevent it from compounding, increasing the free energy of the whole. They install little copies of themselves in every brain so the whole thing starts to cohere like a laser.”

    Does this invalidate the Gervais Principle idea that corporations are run by sociopaths (who believe in an “Absent God”)? Or are the sociopaths all of the other members of the C-Suite who surround, coddle, and support the CEO while he concentrates himself into a “meaning laser” and carves out a dent in the universe?

    • Aptenodytes says

      This may not contradict the Gervais Principle because Venkat explains in the follow-up essay that sociopaths have their own personal ethics etc., which implies that they just may turn the Absent God they facr into personal meaning.

      • This is sort of orthogonal to the way I explored the same idea in GP. The one difference here is that I did not make a distinction before between CEOs who are the most extreme consumers of their own kool-aid vs ones who produce it primarily to craft illusions for others.

        • Aptenodytes says

          I stumbled across my conclusion by accident. How can I learn to synthesize information more deliberately?

  4. Aptenodytes says

    Coase’s Theorem and Businesses
    As businesses become smaller due to automation and the overhead costs decreasing, I expect more steering, not less, partially because having a large number of workers helps to create the synchronization needed for orientation-locking. Am I (not even) wrong?

  5. Ravi Daithankar says

    An analogy that came to mind is a music concert. The crowd is there to lap up whatever experience or thrills the performer promises, obviously implicitly. I would have said rock-star but a pop-star is probably more accurate for this analogy. Because pop artists and pop audiences represent everything that you would associate with small-script CEOs and a gullible workforce eager to rally around whatever mission/vision/strategy is projected up on screen. There have to be optics of course: a big production, lot of noise, and not much substance.

    Very interestingly, in a pop concert, there is lip-syncing, which is hilariously similar to small-script messaging: just a confident front end with zero body or substance behind it, mindlessly mimed/regurgitated with an infusion of energy only for naive audiences to buy into, hopefully believe, and ideally celebrate. The fact that a pop artist can actually sing for real what they lip-sync on stage hints at how small script CEOs actually DO believe in their oversimplified bullshit worldview; they might be peddling a lot of singular non-truths and binary analyses everyday, but one thing you can’t accuse a good, non-steering CEO of is being inauthentic to themselves. Based on my admittedly limited experience, they actually do believe that that’s how the world works. Contrast this with a lot of successful middle managers and even senior leadership, who happily lie to themselves and go along with something they don’t actually believe in, because that is quite simply the pragmatic thing to do. That in itself, sounds like steering, in a personal capacity.

    And the success or failure of a pop concert is really only measured by one evident parameter: the noise and hysteria the pop-star is able to generate. That is vastly more important than any intrinsic musical ability or skill they demonstrate.

  6. Have you just rediscovered Stafford Beer’s viable-systems model? The top level of identity-maintaining decisions sounds a lot like what you’re driving at.

    • Not familiar with that though I’ve heard it referenced. I assume by default that most things I figure out for myself have probably been figured out by others as well :)

  7. Once the ego gets attached to a concept and its identity gets associated with it, it is against the grain of human nature to easily let it go. This is why change is so incredibly hard unless you are incredibly aware of the workings of the mind and its fault lines.

    Now, what if the CEO and her ego is not identified with a specific direction, but to the underlying “job to be done”? Can she be in a constant search fir truth to get that job done better, thus morphing with market dynamics to steer to the best possible direction once every few years?

    It is hard because that is not the way most human egos work.

    • Khuyen Gia Bui says

      Doing so will require ego death once in a while, which I think it’s practically really hard.

  8. I’d be curious on your take re: how this post relates to this article, as I see several parallels:

  9. Peter Vukovic says

    Oky, so a good CEO is someone with zero interest in exploring other directions than the one he personally sets, and companies are simply betting on the probability that their CEOs are right. If it turns out they aren’t, they get fired, because any change in direction is better handled by another CEO who already has different convictions to begin with. Would this be an accurate summary?

  10. This surfaced a question I’ve had for while “should a startup have a CEO?”. You seem to be reinforcing the case for no.

    • Well, that’s taking it too far. A founder CEO of a startup pre PMF is simply a different role that has some features in common (ultimate leadership responsibility/accountability to investors etc) and other features that basically make it a very different role.

      There’s still significant overlap in behaviors that it’s justified calling that role a “CEO”… it’s just a different subspecies.