Everything Should Be Made as Simple as Possible, but No Simpler
If you prefer to listen rather than read, this blog is available as a podcast here. Or if you want to listen to just this post:
I.
My father spent many years working for himself as a management consultant. He wasn’t one of these people that advised CEOs on vision, instead his specialty was shiftwork. Companies that operated around the clock, 24 hours a day seven days a week. Quite frequently he ended up having to work with unions which was its own special brand of crazy, particularly if layoffs were involved.
During the winter of 1990, after the holiday rush was over, it was my turn to get laid off. Beyond all of the normal annoyances which accompany getting laid off, I was also annoyed because I felt that I had been laid off in favor of people who were worse than me at the job, but had greater seniority. So I asked my dad why companies did it that way. He explained: because it was a system which was easy to understand for all of the parties. Competence is fuzzy, and it can be hard to judge even if you’re not the person being judged, and no one has an accurate view of their own competence, but seniority is a bright line. Even if it has to come down to the difference of a few days, it’s clear who’s been working there longer. It’s clear to management, it’s clear to the person being laid off, and it’s clear to that person’s wife or husband. That last bit may be the most important of all, your significant other isn’t going to get angry about your lack of seniority, but they may get mad if they feel you were slacking off or alternatively if there was some favoritism involved. And, as we’ll get into, managing anger is a pretty important part of any process.
II.
Last week I was reminded of this story by an article Matthew Yglesias posted to the subscribers of his new newsletter, Slow Boring. The article was titled Making policy for a low-trust world. (Fortunately this was one of his public posts so you can easily read the whole thing if you want.) His subject is pretty clear from the title, and it touches on something real and pressing (moreso after the events of the 6th) how do you carry out policy when people don’t trust those in power?
Yglesias offers up two options:
Layer on more rules: “If people are worried about the discretionary use of power, you need to make sure the decision-makers go through an elaborate compliance checklist.”
Fewer and far simpler rules or what Yglesias calls “it does exactly what it says on the tin” approach.
Yglesias favors that latter and offers up three steps for doing that:
It’s easy for everyone, whether they agree with you or disagree with you, to understand what it is you say you are doing.
It’s easy for everyone to see whether or not you are, in fact, doing what you said you would do.
It’s easy for you and your team to meet the goal of doing the thing that you said you would do.
The shorthand for these steps might be accessibility, accountability and achievability. (Yeah, I got cute and chose three words that began with “a”.) And Yglesias goes on to show what this looks like when applied to vaccine prioritization (he’s been a big proponent of simply prioritizing by age), the fiscal stimulus/PPP program, quantitative easing, and finally local infrastructure. It’s good stuff, (Tyler Cowen called it the best short essay of the year so far) and as I said it’s not paywalled so you should just go read it.
All that said, I want to take things in a somewhat different, and broader direction. First I should mention that I was saying something very similar in a post from 2017. (Truly I was ahead of my time.) Without getting too deep into the weeds (for that read the original post, I think it holds up really well) I was comparing the book Rationality: AI to Zombies (RAZ) something of a bible for rationalists and bayesians with the actual Bible. And basically arguing that RAZ and rationality in general were examples of Yglesias’ first option for dealing with the world. While they aren’t exactly making a compliance checklist (though I think some of that is in RAZ) they are trying to craft a decision framework for every eventuality. Contrariwise the Bible is an example of the second option. Obviously a totalizing religion is going to have a hard time always complying with all three of Yglesias’ steps, but it is pretty rare for someone to say they don’t understand Christianity (step 1-accessibility). And most people (especially non-Christians) feel perfectly comfortable identifying if someone is being Christian (step 2-accountability). Most of the trouble comes in the execution (step 3-achievability) which does create some unfortunate hypocrisy, but hypocrisy is not actually as bad as people want to claim.
All of the steps are important, but as you might have already guessed step 1, understanding the plan, is the most important not only because the remaining steps build on top of it, but also it’s the chief thing differentiating the two options. And it’s not even all of step one, within that step there is one word that’s more important than all the rest… “everyone”. In my aforementioned post, I pointed out that this was a key difference between rationality and Christianity. As an example of what I mean by this the story of someone in jail converting to Christianity or some other religion (see Malcolm X) is so common as to be a cliche. The story of someone reading the 2300 pages of RAZ and converting to bayesianism is so counterintuitive that I’m sure they could make a TV show out of it. Something similar to My Name is Earl (which was cancelled too soon by the way). In other words it’s not enough that your system is understood by bureaucrats, or people who’ve read the right hundred posts on social media (or 4chan) or the right 2300 page book. It has to be something everyone (or at least a percentage in the high 90’s) can understand.
III.
What’s interesting about Yglesias’ essay is that, despite the timing, he didn’t apply this framework to the election, which, for me, is the obvious place to do so. And you can see that this was basically what I was getting at in my post Voting as a Proxy For Power. I offered up three potential systems for deciding who had won. Which, if we restate them in Yglesias’ framework might look like this:
System 1: Elections as they are supposed to work
Accessibility: We’re going to count up all the votes in the individual states, assign the electoral votes from that state to the one who got the most individual votes, and then whoever got the most electoral votes is president.
Accountability: Each party gets to have observers at critical locations to confirm whether we did the above. (I understand that there are disputes about how well this worked, and in general step 2 in this system is weaker than I would like. But in theory counting votes should be something that can be transparent.)
Achievability: Counting votes is a relatively straightforward exercise, and while it’s not unheard of for people to have questions (see hanging chads) nearly everyone feels confident about their ability to do it, and in fact the people who pushed back most vigorously on accusations that the election was stolen were frequently the election officials.
System 2: Voting as a proxy for power
Accessibility: We’re going to have a smooth, non-violent transition of power, as opposed to what happened historically.
Accountability: We’re going to use voting and democracy to grant legitimacy to the person taking, or keeping that power. In a way that’s convincing (particularly to the elites in the media and government who are custodians of the power) even if it’s not perfect.
Achievability: Everyone has done a good job if power is peacefully and smoothly transferred.
Once again the most difficulty comes on step two, but as you can see, this system is arguably actually even simpler and more straightforward than the first. Now let’s look at what Trump and his supporters actually tried:
System 3: Overturn the election by any means necessary
Accessibility: We are going to get to the true winner of the election by uncovering proof, filing lawsuits, creating spreadsheets, tweeting out accusations, spreading innuendo, and crafting conspiracies. As a result of one or all of these plans the election will be given to Trump by the courts, or the state legislatures, or the Insurrection Act, or the military, or Mike Pence, or occupying the capital, or Trump himself in some bold stroke we didn’t even see.
Accountability: Everyone can tell that it’s still working as long as any of the foregoing still has the slightest chance of working, and if all of them have been eliminated, then Trump supporters will provide you with six other possibilities you’ve never even heard of which are the real way to tell that it’s working, and unless every one of these possibilities has been made physically impossible by the laws of nature the plan is still working.
Achievability: People working in this system should: Stop the count (except for a few days in AZ, in which case you should keep counting); release the Kraken; wait for the courts; wait for the state legislatures; watch Mike Pence; disregard everything that happened before January 6th (it’s all happening after that); gather in DC; storm the Capitol; wait for Trump’s instructions on Twitter; realize the video of Trump conceding on Twitter is a fake; and finally pay attention to the Emergency Broadcast System.
As you can see despite cramming this into Yglesias' framework this is the first option he talked about, the idea of layering on more rules, though in this case they’re layering on every conceivable option so that no avenue for victory is left unexplored. And the point is, it’s so easy to convince yourself that this system has to work. That surely if you just account for every eventuality, mistakes won’t be made. Or if you pursue every possible avenue for victory one of them has to work out. But this is one of those times when no plan survives contact with the enemy. Your rules, checklists, and plans don’t exist in isolation, at some point they have to be understood and implemented. When the rubber actually hits the road, the additional complexity is a liability not an asset.
As we have seen in the days since the election, you can be the biggest Trump supporter there is, firmly believing in both his genius and in the fact that the election was stolen, and it still should be obvious at this point that the third system was never going to work because it entirely ignored the all important task of being something everyone could understand. And not merely does it need to be something your supporters can understand, it needs to be straightforward to understand and implement for all of the organizations you need to have on your side to be President when the smoke clears (regardless of whether it’s an election or a revolution/coup). The military can easily understand systems one and two, but even if you assume that they’re mostly on Trump’s side, how are they going to enact system three? Are you sure they’re not going to be confused by Christopher Miller, the acting Secretary of Defence, the guy Trump put in after the election (according to his supporters as part of the whole secret plan) saying:
I strongly condemn these acts of violence against our democracy. I, and the people I lead in the Department of Defense, continue to perform our duties in accordance with our oath of office, and will execute the time-honored peaceful transition of power to President-elect Biden on January 20.
How is anyone trying to execute on system three not going to be confused by that? Trump and his followers have weaponized complexity, but they haven’t figured out how to target anything with it yet.
Okay, as you might be able to tell I’m a little annoyed. And to be fair complexity has been weaponized for a long time, it might in fact be a serviceable definition of postmodernism. But we’ve certainly reached some kind of landmark.
Before I move on, a few notes about stability and history. First off I think we’ve had stability for so long that most people don’t realize how bad a non-peaceful transfer of power is. So let me be clear, I have strong misgivings about Biden, and Democrats, and progressives, and wokeism, and policies like student loan forgiveness, and reparations, etc. etc. But I would take Biden with a filibuster proof Senate majority composed entirely of Andrea Ocasio Cortez clones over full on civil war which ends up being as bad or worse than the last one. And I’d certainly take what we ended up with (President Biden and Democratic control of the Senate) over a repeat of the violence of the late 60’s/early 70s. For example 1972 when there were 1900 domestic bombings. Now unfortunately we may get both but I don’t think storming the Capitol made either Biden’s presidency or domestic terror less likely.
On the other side of the coin people forget how difficult it is to actually pull off a coup or a revolution. I think people imagine that the French Revolution, for example, looked similar to last Wednesday’s march on the Capitol. That some people spontaneously rose up, and the next thing you know the whole government had changed. One day there was the monarchy and the next there wasn’t. But in reality the revolution was largely a very gradual process whereby the Estates General was replaced by the National Assembly which was replaced by the National Constituent Assembly which was replaced by the Legislative Assembly, and so forth and so on until eventually ten years later you get Napoleon, and for the first three years of that period the King was still around.
Mostly I point all of this out to add another angle on how dumb Trump’s plan really was. Not only was it very unlikely to work, it would have been horrible if it had.
IV.
Perhaps, despite its appropriateness, you’ve noticed that I’ve avoided using the word “legible”, as in “Yglesias is contending that policies need to be legible”, which I’ve expanded to the idea that “the transfer of power should be legible”. Even though it’s basically the perfect word to describe what he and I are talking about. I’ve avoided using that word because this post unfortunately fell immediately after my review of Seeing Like a State by James C. Scott. which is critical of the idea of trying to impose legibility on a natural system. And thus that word, right at this moment, has some baggage, and I wanted to make sure I’d laid the foundation of my thinking before I introduced it. But I do think we should consider Scott and the claims made in Seeing Like a State when discussing Yglesias’ framework, because it’s important to identify when “legibility” is a problem and when it’s an asset.
Perhaps the biggest thing to keep in mind is that there’s a great deal of difference between efforts to make the citizenry legible to the state as opposed to making the state legible to the citizenry. In the former case the benefits accrue to the state, and in the latter they accrue to the citizenry and I’m almost exclusively talking about the latter.
Additionally, legibility is one of those things where you should apply as much as is needed but no more. In a sense it’s closely related to the idea of subsidiarity, that programs should be implemented as close to the problem and the people affected as possible. Legibility should be as close as possible to the way nature already works.
It might help to think of there being three possible levels:
Natural
Legible
Controlled
As it says in the Federalist papers, “If men were angels, no government would be necessary. If angels were to govern men, neither external nor internal controls on government would be necessary.” Which get’s at the first and third levels. If men could be trusted to behave without any government that would be the best solution, and this is the state of nature as described by Scott, and the philosophy of anarchists and libertarians (though to different degrees). If on the other hand angels were to govern men, then we could give them control of everything knowing that we would never need to second guess them, and it wouldn’t matter how complicated those controls became. But since there are no angels in sight, the middle ends up being the goldilocks spot described by Yglesias where there are rules and policies, but they’re easy to understand. They’re legible but not complicated.
As I was working through this post it occurred to me that Yglesias’ framework can be applied to the recent reckoning on race, though I’m sure he’d probably rather not go there, and even I am only mentioning it as an observation rather than any kind of recommendation.
What I’ve noticed is that as things have progressed since the death of George Floyd, the complexities of race have become very apparent. A few examples: There’s been a tendency to separate people as being either white or people of color (POC) and yet Asians who would be considered POC have much higher median household incomes than white americans. Affirmative action largely benefits people who are already in the upper middle class rather than minorities that are truly disadvantaged. When it comes to reparations there are all sorts of complexities. Does Oprah get reparations? Do people who recently immigrated from Africa, and have no enslaved ancestors get reparations? And what about the Native Americans?
I’m not saying these problems are insuperable, I’m just pointing out that they lead to exactly the sort of rules layering that Yglesias pointed to as being bad. On the other hand, the old standard of being completely colorblind is legible, straightforward and a perfect example of Yglesias’ criteria. But as I said I’m merely observing, not recommending.
V.
After taking the Yglesias framework up a level, and using it to consider the recent unpleasantness (i.e. from policies to the choosing of people to enact those policies). I think we can take it even one step higher, to the level of values.
As I was working my way through all of this I was reminded of my post on the justice/mercy dichotomy. As usual when I wander this far afield everything I say is pretty speculative, but I once again see a situation where there’s too much focus on justice and not nearly enough focus on mercy. To begin with, while I understand it’s hard for some people to understand, the riot that happened last week, insofar as it had a motive other than “riot tourism” (I forget where I saw that phrase but it seems apt) was motivated by justice. All or nearly all of those people are convinced, deep in their bones, that the election was stolen. That Trump actually won, but the Deep State contrived to make it appear as if he had lost. That if they had been able to sway enough of the senators to change the outcome of the electoral vote counting and give Trump the win, that this would have been just and proper. Now you can go back and read the previous post if you want an explanation for all the reasons why the modern world has made this path particularly easy to follow, and not just for Trump Supporters. So to an extent everyone is obsessed with justice. The problem is that justice and mercy are opposed. You can’t have both. And what we needed last week, and really since the election is more mercy.
Of course calls for the left/Biden Administration/institutions to be merciful to Trump supporters are legion. And while I think that’s an area where we should err on the side of mercy, in this space I’m going to argue that actually it’s Trump and his supporters who need to be more merciful. I understand that some people don’t think that’s possible. They think mercy is something that can only be granted by the people in power to the people who aren’t in power. But in reality mercy can operate even if you’re the weaker party. As long as you have some power you can decide to forgo using it and exercise mercy. Even if you have less power than your opponent, as long as you have any power you can use it to cause harm. Deciding to not to is an act of mercy. As such, conceding is an act of mercy, directed both at the other side (even though they won) and at the nation as a whole. And it’s actually more important if you think justice has not been served. Anyone can be merciful if they think they’re in the wrong, it’s being merciful when you think justice is on your side that poses all of the difficulties.
So what does all of this have to do with legibility vs. complexity? I would argue that mercy is legible. Forgiveness is easy to understand. On the other hand justice, true justice, is enormously complicated. And I’m not arguing that we should abandon our quest for justice. I’m just pointing out that when Yglesias was calling for a framework that could easily be understood that he was also calling for mercy.
As I’ve said this is all on the highly speculative end of things. And I can completely understand that in calling for mercy, particularly from the weaker party, I am in a sense calling for people to accept some injustice, and of the worst kind too: that committed by the strong against the weak. But perhaps, by flipping the framing such that Trump supporters are the ones who are being asked to meekly submit to injustices (whether perceived or real) and to do so for the good of the country, those most inclined to object to my conclusion might be induced to see that it contains a sliver of wisdom.
Perhaps the appeals I make at the end of every post also suffer from the weakness of being too complicated, so let me try Yglesias’ framework:
I’m asking for money so I can prove to my wife that I’m not wasting my time.
You’ll know it’s working by my periodic mentions of having a wife in the present tense.
You can execute on this plan by going to https://patreon.com/jeremiah820 and clicking on one of the “Join” buttons.