A Pattern Language For Rationality

There’s a lens to looking at the rationality project that I’ve been finding enlivening recently, and I think it’s reached the point where more eyes and hands might be useful, while not being anywhere near complete yet. First, some background.

Christopher Alexander was a designer and architect; his thinking, and focus on patterns in particular, were influential in programming; wikis were first invented to facilitate the collaborative creation and modification of ‘patterns’ in the style he described. He wrote lots of books, but I’ll focus on this trio:

The three books, while each on a different subject or layer of ‘design’, all work together and depend on each other. The first identifies the target, why it would even be good to pursue, and how you know whether or not you’ve found it. The second is a detailed description of the patterns they’ve found useful in approaching the target. The third is what it looks like to organize systems to deliberately and durably organize themselves according to this design.

Just like this approach was profitably translated to programming (and other areas of design), I think it’s worth looking at the ‘rationality’ project as a way to design decisions, habits, and thinking, and attempting to deliberately incorporate Alexander’s approach and strategy. This post is an attempt to get started[1] in each of the three directions, rather than fully lay them out; depending on how things go, I might turn this post into a sequence /​ flesh out individual sections.

The Timeless Way of Living

The Tao that can be told is not the eternal Tao; The name that can be named is not the eternal name.

[Consider also ‘thinking’, ‘deciding’, ‘doing’, and ‘being’ in place of ‘living’, for both this section and the next.]

Alexander talks about the “Quality Without A Name”,[2] so-called because no existing English word was a good fit, altho he could point towards it pretty effectively in a few pages; you should read the start of The Timeless Way of Building (Amazon, pdf) to get his sense of it. I think “equilibrium” comes somewhat close–a design has the Quality Without A Name if all of the forces present, both internal and external, are balanced. This isn’t the true name because many equilibria we see don’t have the Quality Without A Name, because for them only some of the forces are active in determining the level. A house might be at ‘equilibrium’ according to the windows and the thermostat, but not according to the human inside who’s not happy with the situation and wants to do something about it.

When talking about buildings, he talks about whether they’re ‘alive’ or ‘dead’. His overall sense is that many design features are ‘obvious’ or ‘natural’; while there might be lots of detail to the model of how to make things that are alive, most of the ability to detect whether or not a building is alive or dead is already ‘baked in’ to being a human.[3]

I think it’s relatively easy to point to good buildings and bad buildings, and somewhat harder to point to good lives and bad lives, mostly because we can walk around the inside of building made by others but not their minds. Nevertheless, it seems possible to collect pictures of what it is that makes life worth living, what properties good decisions have, what virtues we might like to embody. This is, historically, a place I think the rationality project has done pretty well.

You may try to name the highest principle with names such as “the map that reflects the territory” or “experience of success and failure” or “Bayesian decision theory.” But perhaps you describe incorrectly the nameless virtue. How will you discover your mistake? Not by comparing your description to itself, but by comparing it to that which you did not name.

One interesting thing about practical Bayesianism (rather than superhuman Bayesianism) is that among the normal, explicitly considered hypotheses lurks a monster: “all other explanations.” What probability should it have? How should one update on it when observing a new piece of evidence? For the other explanations, one can follow the math as written; that one is precisely a placeholder for that where one cannot do the math, and must guess. Having a token in your language for “that which is not expressed in the language” helps keep you grounded in the reality larger than your mind, rather than trapped in your imagination.

A Pattern Language for Living

One fascinating thing about A Pattern Language (for buildings) is that it is highly opinionated, and only contains what you should do. But it’s often just as useful to have a list of anti-patterns, that is, common models or designs that are recommended against. Then, as you’re designing, you can notice that you’re about to make a mistake, and replace the anti-pattern with an appropriate pattern.[4]

According to me, the rationality project that grew out of Overcoming Bias and LessWrong has a tilted focus in favor of anti-patterns, because of the heavy early influence of the heuristics and biases literature. As well as 37 ways words can be wrong, what are the ways words can be right? LW does have a lot of patterns, several pointers to deep generators of patterns, and some focus on the Quality Without A Name, but it seems to me like there’s significant room for improvement here, and I’m interested in working to map out the space and collect patterns.

In A Timeless Way of Building, Alexander describes the patterns of a person’s life:

A building or a town is given its character, essentially, by those events which keep on happening there most often.

A field of grass is given its character, essentially, by those events which happen over and over again—millions upon millions of times. The germination of the grass seed, the blowing wind, the flowering of the grass, the movement of the worms, the hatching of the insects...

A car is given its character by the events which keep on happening there—the rolling of the wheels, the movement of the pistons in the cylinders, the limited to and fro of the steering wheel and axle, as the car changes direction.

A family is given its character by the particular events which keep on happening there—the small affections, kisses, breakfast, the particular kinds of arguments which keep on happening, the way these arguments resolve themselves, the idiosyncrasies of people, both together and alone, which make us love them...

And just the same is true in any person’s individual life.

If I consider my life honestly, I see that it is governed by a certain very small number of patterns of events which I take part in over and over again.

Being in bed, having a shower, having breakfast in the kitchen, sitting in my study writing, walking in the garden, cooking and eating our common lunch at my office with my friends, going to the movies, taking my family to eat at a restaurant, having a drink at a friend’s house, driving on the freeway, going to bed again. There are a few more.

There are surprisingly few of these patterns of events in any one person’s way of life, perhaps no more than a dozen. Look at your own life and you will find the same. It is shocking at first, to see that there are so few patterns of events open to me.

Not that I want more of them. But when I see how very few of them there are, I begin to understand what huge effect these few patterns have on my life, on my capacity to live. If these few patterns are good for me, I can live well. If they are bad for me, I can’t.

Of course, the standard patterns of events vary very much from person to person, and from culture to culture.

For a teenage boy, at high school in Los Angeles, his situations include hanging out in the corridor with other boys; watching television; sitting in a car with his girlfriend at a drive-in restaurant eating coke and hamburgers. For an old woman, in a European mountain village, her situations include scrubbing her front doorstep, lighting a candle in the local church, stopping at the market to buy fresh vegetables, walking five miles across the mountains to visit her grandson.

These are the patterns on the scale of ‘events’, but one of the most charming features of A Pattern Language is the breadth of scales that it considers. Ordered from largest to smallest, the first entry in the book is Alexander’s sense of how the whole Earth should be organized (a global government made of independent regions, each containing 2-10 million people, so they can be small enough to be tolerable for the people inside them) and the final three (to give some sense of variety) are what sorts of chairs to put in a space, how to light them, and how to decorate one’s walls. The patterns nest within each other, as each larger pattern depends on smaller patterns beneath it (until the simplest, smallest patterns at the end).

Similarly, rationality patterns extend from the great Neo-Enlightenment project all the way down to habits that fire on the 5-Second Level. Also, in a manner which makes them more difficult to organize, the patterns scale both in time and number of people (whereas with buildings, it is practical to simply organize the patterns by volume).

Historically, I think LessWrong has done great at collecting antipatterns, and only well at collecting patterns, and somewhat poorly at organizing them all into a common reference work. (Read The Sequences, we used to say, but much of what has happened since then is scattered instead of curated and carefully arranged. My guess is a solid approach to take here is something like “go thru the Sequences, the CFAR material, the library of Scott Alexandria, and so on, pulling out the patterns and trying to arrange them into a pattern language, identifying the connections between them and filling the resulting gaps. If you’re interested in helping with this, comment below or send me a PM.)

I also have some sense that this sort of ‘life-design’ is a key component of applying rationality consistently and well; some large part of the rationality project’s benefits for individuals have been, I think, from being exercising this sort of intentionality and deliberate thought about parts of their life where people often just operate on autopilot. Having a systematic way to look at this, instead of just noticing things as they come up, seems quite valuable (so long as one makes changes in an organic, sustainable way which is pointed towards the Quality Without A Name; it’s generally a mistake to give edit access to yourself to your less grounded parts).

[Incidentally, when I did this I got 25 patterns for myself, with ‘standing at my computer’ split into a further ten patterns, tho many of those patterns were things that happen on a ~weekly cadence instead of a daily one, which increases the number substantially.]

The Oregon Experiment For Time

The University of Oregon wanted a ‘master plan’ to guide its growth for the next few decades, and reached out to Christopher Alexander; he responded with a bit of a rant on why master plans were terrible, and what they should do instead.[5]

Their goal, as Alexander saw it, was the same timeless way of building that leads to the organic growth of living towns and cities. The challenge was how to do it when there was a monopoly funder and decider. How do you get the users to change their location to fit their needs, while being harmonious with the whole, while everything needs to be signed off on by the center?

This is especially relevant for me because I’m something of a free agent, at the moment. I don’t see any plans that obviously have good impact on x-risk, I’m not interested in pretending any plans are more attractive than they actually seem, and yet I still find it fun to help people out and think about these problems (and think the ‘option value’ of staying involved is quite high). I also am financially secure enough that I don’t need a job, which makes a freelance style more attractive, whereas if there were jobs I thought were obviously worth taking then the benefits of specialization would push hard against being freelance. But this feels like the same sort of situation–a monopoly funder/​decider which is nevertheless trying to do things that are good for its users and generate the Quality Without A Name.

I think this is probably still relevant even for people in traditional jobs or educational situations; after all, it’s still the case that you’re the monopoly funder and decider for what to do with your time and energy, and it’s worth looking into ways to incorporate these principles (or deliberately find competing principles, rather than aping the behavior of those around you or following whatever instructions you receive[6]).

So let’s take a look at Alexander’s principles and see if they can be adapted from “university space use plan” to “individual time use plan”.

  1. Organic Order: use a gradual process to construct the campus, rather than visualizing it all at once.

  2. Participation: users make decisions about what and how to build.

  3. Piecemeal Growth: weight budget overwhelmingly towards small projects (in terms of number; be roughly equal in terms of dollars spent.)

  4. Patterns: control and harmony should be exerted mostly by deciding the ‘language’ of the campus, i.e. what patterns are appropriate; all users should build things according to the relevant patterns.

  5. Diagnosis: once a year, the design committee goes thru the campus and determines which parts are currently ‘alive’ and ‘dead’.

  6. Coordination. Funding process handles a stream of proposals that are put forth by users.

It seems to me like Organic Order, Participation, Piecemeal Growth, and Coordination translate straightforwardly. Organic Order, for example, suggests that I should only be setting vague themes with long-term planning, rather than trying to schedule out all of my hours weeks, months, or years ahead of time.

Patterns and Diagnosis both feel like they would translate straightforwardly, if the other components were present; Patterns, for example, seems like it could just obviously work if I had the pattern language to work off of, and Diagnosis seems like it relies on a sense of ‘alive’ and ‘dead’ which I don’t have a good handle on yet (but could probably wing it–The Timeless Way of Building is pretty confident that people ‘already have’ the ability to tell alive and dead spaces apart).

Some seem like they need rescaling. Piecemeal Growth has budget categories that make sense for buildings /​ renovation projects in a university, but what’s the corresponding categories that make sense for spending time?[7] Note that if you have a sense that projects should be distributed such that each category has equal spending, where you draw the category boundaries now has decision-relevant effects.

Organic Order

As I go thru life, my local information, wants, and capacities will vary substantially; the planning process should take that into account, and not make plans that it expects to become stale.


Lots of what I do is for other people. They know things I don’t, and want things I don’t yet imagine; the planning process should take that into account, and try to bring me into contact with my impact and what it could be.

For buildings, the principle is somewhat obvious; people have lots of implicit knowledge of how their space should be arranged, and should tell the design to the architect, rather than the other way around, or trying to verbalize their requirements. For time, it seems more about establishing clear requirements, and trying to put those requirements in the hands of the beneficiaries.

For everything that I spend serious effort on, “knowing where the score is” seems like a basic thing to keep in mind. If I’m doing something “for my husband”, it really should be clear to me the score is in terms of “how much he valued it” and the score is in him; and if instead I’m doing something “so that I feel like I’m a good husband”, then the score is in me. If I’m doing something to help out people in my house, knowing what they actually want means it’s more likely that I’ll do something useful to them.

Speaking about the EA /​ longtermism space more broadly, I really wish we were better at using prices to convey information. If you’re working a job in the selfish economy, the price your employers are willing to pay you is a pretty good measure of “how much good you’re doing for them”, whereas if you work at a nonprofit, you just have a sense of ‘doing good’, and there’s not much in the way of joint calculation of how much good you’re doing. EA is, in large part, an attempt to fix that–but I think it’s still missing in lots of places locally.[8]

Piecemeal Growth

Part of contact with reality is learning which things succeed and which things fail, which things are effortless and which are effortful.

Alexander argues that often, people consider equal numbers of projects in various size categories, but this means the result is heavily skewed towards the larger projects. Instead, have a process that allocates equal budgets to the various size categories, leading to most projects being small.

I think I’ve often made a mistake where I turn small ideas into big ideas, and thus big obligations. I started writing a fanfiction whose core idea was a six chapter cycle; it turned into an epic that I never finished. A friend who is a successful artist would start her day by sketching out lots of quick things, and then would pick from the pile which one to turn into a detailed painting. It seems likely that her paintings were unusually good because she had that filtering step, where ideas could be tried in a low-stakes environment and then only the best ones got significant effort.

As well, there’s a common pattern in industrial engineering of “WIP as waste”; work-in-progress is not providing value to anyone yet and is causing costs while it exists, and so a system that manages the same output with less WIP is better. If rather than working on five posts in parallel and publishing them all at once at the end, I work on the posts serially, then the first post released can be released much earlier (and be providing benefits while the others are being written, and they can change in reaction to its feedback, and so on).

If I want to spend ten hours on a goal, quite probably the right way to do that is to do five small 1-hour projects, then spend another five hours on the one that seems the most promising. That way, if it turns out that there are dead ends along half of the approaches I thought of, I spend most of my time on an approach that turned out to be promising. Otherwise, the method will end up heavily associated with the goal (consider the difference between when I sit down to spend 10 hours “writing a blog post about X” and when I spend 10 hours “on X”), and I might just give up or waste time for half of my goals.

A less-obvious benefit of this approach is that, from the point of view of the university campus, the ‘already built’ areas of the campus need love too (in order to stay ‘alive’, in good repair, in touch with their user’s interests, and so on), but what they need will primarily be small projects. A committee which only allocates money to large fancy new buildings will see all of its old buildings deteriorate, whereas one which reserves a large fraction of its funds for small projects will see many of the old buildings well-taken care of.

One argument against piecemeal growth for individuals (instead of organizations) is that a university can spend its budget on many projects in parallel, whereas an individual has to do things one after another. Not being able to spend 20% of my focus on five small projects at the same time, whereas I could spend 20% of my budget on five small projects at the same time, means that smaller projects are more costly because of switching costs and delays to other work.


When choosing to do things, have a sense of ‘what they are’, what subcomponents they need, and what supercomponents they support. Have a sense of both what to do and what not to do.

It also seems like having a shared language for variation in buildings aids in specialization—you can easily talk about the differences between a park, a house, and a workshop, and find the building you’re looking for. Similarly, being able to advertise what projects you’re good at, enjoy doing, and hook into your larger goals makes it easier to coordinate.

I’m also growing to appreciate ‘seasonality’ more, in a way that patterns can identify and support. One might naively think that if doing X is better than doing Y, I should just schedule myself to only ever do X and never do Y, but it’s rare for there to be true dominance. A pattern like “Focus in the mornings, connection in the afternoons, freedom in the evenings” ensures that no nutrient goes uneaten for too long (tho what the actual ratio should be is a fact about your situation, instead of about what happens when you divide one by three).


See what is actually true on a pace that allows it to influence what you do; don’t let corners fester unconsidered.

For the university, this looks like an annual checkup, where the planning committee physically tours the campus, simply noting ‘alive’ or ‘dead’ for each space. While you can do a quick tour thru a physical space, doing tours thru time is less easy. My guess is that this category should be ‘retrospectives’ in general, dealing with properties on the relevant timeframe. A daily reflection asks about the aliveness of hours and tasks; a weekly reflection asks about the aliveness of days and projects; a monthly reflection asks about the aliveness of weeks and themes; an annual reflection asks about the aliveness of months, life goals, and principles.


For a campus, it’s obvious who the users are and why they have information that the central committee lacks. If the music school is poorly designed for the students of the school, how will the campus architects know, unless they have a way for the students to tell them, and then participate in the design to fix it?

For me, I think people approach me infrequently with things that I could do that would help them. There have been times when I’ve been busy and had to say no a lot, but my typical response these days is ‘yes’, suggesting that I’m getting too few requests.

The other side of this is a general approach to planning; all projects, regardless of size, have to talk about their goals and what patterns they’re a part of, with larger projects having more preparatory work than smaller ones, and considered in larger budget meetings (as opposed to handled by a more local budget process). Coordination is about the process by which the pieces fit together and change behavior.

Other Principles?

At the moment, my plan is to start off with something that’s a near-copy of The Oregon Experiment, and then adjust it in contact with reality. But perhaps you can see now some piece that I’m missing, or some piece that I’m holding on to despite its irrelevance; I’d love comments if so.

  1. ^

    Well, for me to get started; I think a lot of this stuff is already ‘in the water’ in some important sense, but like scholarship being ‘in the water’ is very different from carefully reading books and writing reviews of them, I think having a loose sense that ‘patterns are neat’ and a big book which links all of the patterns together in a sensible way are very different, and I want to get the latter.

  2. ^

    I saw someone writing about Christopher Alexander abbreviate it as “QWAN”, which 1) makes sense and 2) is easy to pronounce but 3) feels weird or ironic in some way. I went back and forth on whether to use it in the post, and decided not to (while naming the dilemma).

  3. ^

    His design sense is very human-centric, which I weirdly want to describe as being ‘alien to me’? Like, his sense of what makes a building good is very tied to the human perceptual system instead of simpler metrics that are more understandable to an engineer. “A human should have an easy time making sense of this”, he says, even tho that’s much harder to measure than things like the materials cost. I think he’s also about ten percent too certain that all humans share the same sensemaking system, instead of many of these things actually being matters of taste.

  4. ^

    Often, I wish that there were ‘neutral’ categories with ‘good’ and ‘bad’ subcategories, as it seems nonsensical to claim that anti-patterns aren’t patterns in the generic, neutral sense; however, I’ll use the standard language here, where patterns are definitionally ‘recommended’ and anti-patterns are definitionally ‘anti-recommended’.

  5. ^

    As far as I can tell, they implemented Alexander’s suggestions, but I couldn’t easily find much of a retrospective, which makes this a bit disappointing as an ‘experiment’ for us as observers. They also have a set of 12 principles for plans which are quite different from the 6 principles underlying the creation of the process, which seems reasonable.

  6. ^

    For example, see Half-assing it with everything you’ve got as an example of ‘looking behind the curtain’ and figuring out what principle is the right call.

  7. ^

    I’m also thinking here primarily about “high-energy” time; it makes sense to view my morning pomodoros as ‘improvement dollars’ and less sense to view my evening wind-down activities in the same way. But it feels like Piecemeal Growth has similar things to say within each category, even if it can’t translate currencies across categories.

  8. ^

    If you could work for either GiveWell or 80k or MIRI, do you have a number that you can optimize that represents how much good you’re doing for them? If there’s not, how could there be shared epistemic state on that?