Rules for large players
Since I worked at Google several years ago, I had a vague sense that they didn’t see themselves as the globally powerful organization that they in fact were. (This is probably tied to how Google was founded — Larry and Sergey were ordinary Stanford grad students, and so while they had great technical ability, they had no experience in handling big institutions, nor did they tend to think in those terms.) Lately, I’ve tried to cache out in more detail what exactly that means. What would Google act like, if it had an accurate model of itself? What rules govern powerful institutions in the year 2018? Below are some guesses: The amount you are attacked by others will be roughly proportional to your size or visibility; it will have little to do with the actions you take.
As an individual person or small group, most likely nobody will bother to attack you, unless you do something very unusual; this creates an obvious incentive towards risk aversion. For example, suppose I think I hear a burglar in my house. Should I take out a gun? If I shoot someone by accident, the police will come after me, the press will come after me, everyone I know will hear about it, I might well be sued or arrested; it would be a disaster. Hence, leaving morality aside for a moment, there is a huge self-interest incentive for me to not escalate the situation.
In contrast, when an organization is large and influential enough, it is always doing something terrible, by virtue of its sheer size. Suppose that one had fantastic hiring and vetting procedures, and managed to reduce the number of bastard employees who hate joy to just one in two thousand, a tiny 0.05%. Additionally, suppose that each bastard employee can only do one horrible thing per week. Even under those conditions, with four hundred thousand employees, there will still be twenty-nine horrible things happening every single day. Therefore, you will get sued no matter what. You will get attacked in the press no matter what. Some people will hate you, even justifiably hate you, no matter what, and so on. If one is a deontologist, any choice between big institutions is necessarily lesser-of-two-evils, as all big institutions will inevitably break any set of rules on an hourly basis.
Travis Kalanick, for all his faults, realized this very early on — Uber would inevitably get attacked, and so it might as well do all the terrible things it in fact did. If Uber had been better behaved, newspapers and lawyers and local governments would instead (I strongly suspect) simply choose something else to pick on, like all the people whom I’m sure (by simple law of averages) have been robbed or assaulted or racially discriminated against in Uber vehicles. As dark as it is, Uber’s actions are, narrowly, a rational response to their incentives. I’m sure everyone can think of politicians who have done similar things. (I say “narrowly” here because having no ethics does, I think, make an organization much less productive and influential in the longer term. More on that later.)
Strategy should be set in terms of incentives and responses, not rules and systems.
Suppose that I have a bank account with Chase. From Chase’s perspective, I am, essentially, an interchangeable part. Chase is not going to change the way it runs its banks because of anything I might do. Chase might respond to some of my actions, by (for example) giving me credit card bonuses, or charging me fees. However, all of these responses are more-or-less scripted; given X input, I usually get Y output. Therefore, my best strategy is to figure out what the rules are — whether they are written down somewhere, or just de facto “how it works here” — and change my behavior to get the best payoff. Indeed, in the case of banks, there is a whole community of people who fly first-class for cheap by exploiting the rules around credit card bonuses and airline rewards.
Google, Facebook, Apple, etc. excel at this kind of thing, as seen in their use of eg. the Double Irish With A Dutch Sandwich maneuver to reduce their taxes. However, once you get large enough, this stops working well. At a certain size, other players, whether it’s banks, or governments, or utility companies, will start modeling your behavior and deciding how to respond based on their own interests. Using some very loose analogies, it becomes more like a tabletop game than a puzzle game. Whether it’s an opposing organization, or simply an organization that could choose to either help you or not, the key question is how to make it so that it’s in their best interest to do what you want. Amazon recently tried to pull this off, with the bidding between cities for their second headquarters, using what those cities wanted (jobs and economic development) to drive the best deal. Unfortunately, they then screwed it up by using their leverage to bargain for things that don’t really matter (a few billion in tax incentives? really? You have $200B in revenue with 30% annual growth for crying out loud, guys.) Google likewise screwed up their Fiber infrastructure program through internal mismanagement. Phooey.
Vertically integrate, buy or route around monopolies, and remove single points of failure.
Suppose that I discover a wonderful new machine to turn widgets, which cost $100, into sprockets, which cost $200, with expenses of $50 per unit. Of course, I make a $50 profit. However, suppose that there is effectively a single supplier of widgets in the world, with a monopoly position. Initially, they ignore me, and I continue making money as usual. But eventually I am so successful that, after a few years, I start buying up a majority of the widgets. What happens next is that the widget manufacturer raises their prices to $149. I’m screwed, but there is nothing I can do. Economically, all profits from a supply chain will accrue to the stages of that chain that hold monopolistic bottlenecks. This applies not just to suppliers, but to every good, service, or even intangible benefit that an organization might depend on.
For example, when I started work at Google, I had the bright idea to also found a Bitcoin company (protip: never do this). I happily walked down to the Wells Fargo in Mountain View, close to the Google offices, to set up my corporate account. While filling out the paperwork, the bank manager bragged that Google used Wells Fargo for their own banking, which I hadn’t expected. What? Google doesn’t have their own bank? Why not? This surprised me, not because Google has any special kind of insight into banking, but because relying on some other organization for a critical service obviously creates an extra point of failure. In software, Google realizes this, which is why the Google codebase is self-contained down to the level of compilers and operating systems — it’s dangerous, from both a security and a business standpoint, to depend on stuff that you don’t manage yourself. However, they then leave themselves vulnerable from a number of other directions.
Your productivity will be determined by the incentives and management systems you create within your organization.
As a small or even mid-sized company, you typically base a competitive advantage around hiring a good team, and designing a superior product. However, a large enough organization will expand into so many different areas that no one idea or technology provides a critical advantage. Hence, rather than directly assigning people tasks, the job of the CEO becomes causing those under him to, fairly independently, choose the right projects and execute them well.
On a corporate level, Google has recognized this with their creation of Alphabet, an umbrella for all their independent projects. However, it’s clear from how they operate internally that they still don’t really “get it”. Being computer geniuses, Larry and Sergey designed their programmer vetting process themselves, and it shows: many of their early hires were stars, and even running at a huge scale, their hiring pipeline still picks fairly competent people most of the time. You can still walk up to a random Google employee, and be reasonably confident that they know their code. But it shows equally much that Google’s promotion, compensation, and management processes were not given any attention. They do not look like a tool designed to achieve Google’s purposes; they look like something designed in blind imitation of what IBM or SAP did, doubtless by designers who had previously worked there. (As one trivial example, Google tried to build a “technical track” so that great programmers could get paid big bucks without becoming managers, then wrote the promotion criteria for the “technical track” to disqualify “individual contributors” aka non-managers. Great job guys.) I think this is the biggest reason why Google’s original product, the search engine, has seen slower innovation over the last five years or so. I’m now quite confident that Google Search will be replaced over the next decade or two, the only question is who will get to pick that plum.
Though the above may seem kind of cynical, I think it ultimately provides a greater check on institutional bad behavior than any kind of external “law” or “punishment”. Look at Scientology, for example. Thirty years ago, the IRS sought to revoke their tax-exempt status, on the entirely reasonable grounds that Scientology was acting to enrich its leaders, rather than help its members or the public. Scientology didn’t change their own behavior; they responded by going to war with the IRS. They sued them, they harassed them, they set up hotlines to collect stories of IRS abuses, they put up a huge fight for years, to make collecting their taxes not worth the IRS’s trouble. And they won. The IRS simply gave up, and they got a permanent, organization-wide tax exemption. But what happened next? Their incentives and management practices encouraged so much infighting, back-stabbing, and negative-sum games that their leadership self-destructed, and half their upper management quit. They’ve been on a slow downward spiral, and seem doomed to die as their existing membership fades away.
It’s unfortunately hard to find writers who explore this without heavy distortion through some political slant, but many of the basic underlying concepts are covered in more detail in Negotiate with Chad , a blog I recommend.