The Constraint Legibility Hierarchy

While most people can imagine how constraints affect the ability to create outcomes when they are fairly simple, most plans proposed ignore critical constraints entirely. Almost everyone realizes that "the cost of the land" is a significant constraint in terms of determining "the cheapest possible home one could build". This is a straightforward relationship.

People can also often imagine informational constraints of some sorts. "If I know they have a short password, I can guess it easier than if I know they have a long random password". Very comprehensible for the difficulty of guessing a password. Other informational constraints are less legible but still legible: "we need to research fusion before we can build a fusion power plant that is net positive energy". But things then start moving out of the realm of the average human's legibility rapidly.

Even something like "knowing whether something is possible" is a constraint and is on the threshold of legibility. People know "I can't promise it yet if I don't know if it is possible" but adjusting from there to "someone else demonstrated it, so that reduces my difficulty of making it happen" is still quite hard. These are questions really of domain knowledge but they start to bleed into the concept of knowing what other people know and what their own incentives and objectives are. Moving from there into concepts like "Our ability to demonstrate common alignment" as a constraint to "we can work together" is past the point of median human comprehension, I think.

Constraints such as "the minimum group of people that need to be convinced" as a constraint to "getting this outcome" is far past the point of most people. Probably there are far higher tranches of constraints that are beyond my current ability to consider and consequently I am unable to mention them here.

The degree to which constraints become illegible seems to correspond with the number of independent agents participating in an interchange that one needs to have Theory of mind for, scaled by their interest in the interchange. In some sense, the more-legible space is dominated by a domain knowledge constraint where learning can fill in the gap effectively. The less-legible space enters the domain of anticipating other agents' constraints.

An Example From Enterprise SaaS Sales

A simple model of enterprise sales is "I have product that provides $x$ utility to company A. If A finds $x$ sufficient for their business, they will spend $y;y<x$ on a contract with me so that they can capture $x-y$".

One slightly more complex model involves the ability for the agent you're interacting with inside the other company to be able to convince the other agents in the company that they can actually extract the $x-y$ utility. These other agents may be their bosses, or their finance team, or so on.

Another more complex model involves the signal that the deal sends and the precedent it sets for future deals. For instance, a friend of mine had a contract with a client for $100k/year to provide a certain service. As the company grew and usage grew, he wanted the contract to go up to $150k/year. The procurement organization on the other side would not accept 50% growth in spend on a contract since that was unprecedented for any of their vendors. Even though the party inside the client's organization still believed that they would extract more value (and more than before), the procurement organization could not make an exception.

On the other side, my friend's company's economics don't make sense if he cannot grow revenue fast enough with usage. It wouldn't make sense for him to accept 20% growth in the contract size because that would mean that his business's top line revenue could only grow, in the limit, to a certain number. It would be better for his business to fail and for him to go work on something else than to accept slower growing contracts and retaining the revenue.

This last bit wasn't that obvious to me. I'd previously thought that he should accept the lower growth, but clearly this is not the case since if his growth is entirely all contracts like this, it means his business is not viable. It entirely relies on the product being so valuable that he can tag on to his clients' enterprises' growth.

Now, anyone in enterprise sales will recognize this as the shallow end of the pool, but hopefully most people are not in it.

An Example From Politics

Most people like to blame politicans for not doing things that everyone wants. Frequently, however, it is not the case that everyone wants something or that the everyone is evenly contributing to the politicians primary incentive: the ability to continue to set policy.

In the shallows where everything is visible we have the most despised members of a US political party. Frequently, these are congressmen and senators who have best demonstrated their ability to coalition-form in highly-contested states. Often, they have high Wins above replacement, though if you were to do a straight comparison against a more partisan member one will find that they perform poorly. However, these are very valuable members precisely because All Combat Happens On The Frontier. The participants closest aligned to you on the edge are critical to your objectives.

A few such examples come to mind from San Francisco politics. One is that of Scott Wiener's bill from the last year that banned junk fees, but exempted restaurants. Opponents of Wiener reasonably can claim that he made an exception for restaurateurs. However, in a status quo of junk fees with no actions taken by anyone, Scott Wiener's rule bans junk fees for all but restaurants. These bills require a great deal of coordination, and to get sufficient votes requires convincing representatives of various constituencies. What is likely the case is that the minimum coordination action required the exemption. In the days since the passage, no one has proposed a rule to remove the exemption, which somewhat supports that theory.

Another example that comes from SF politics is that of Alan Wong, the supervisor for District 4 who was appointed to that role to replace a short-lived replacement for Joel Engardio who was recalled for his support for the Sunset Dunes park that replaced the Upper Great Highway. Alan Wong has a somewhat-ineffective but full-throated support for returning the park to being a road. The constraint here is in getting someone who does not succeed at that goal while simultaneously being able to coordinate a sufficiently large coalition to vote for him on the whole. The Mayor, Daniel Lurie, who appointed him therefore gets to form a sufficient coalition in this way.