Previous post: How Escape From Immoral Mazes
Sequence begins here: Moloch Hasn’t Won
The previous posts mostly took mazes as given.
As an individual, one’s ability to fight any large system is limited.
That does not mean our individual decisions do not matter. They do matter. They add up.
Mostly our choice is a basic one. Lend our strength to that which we wish to be free from. Or not do so.
Even that is difficult. The methods of doing so are unclear. Mazes are ubiquitous. Not lending our strength to mazes, together with the goal of keeping one’s metaphorical soul intact and still putting food on the table, is already an ambitious set of goals for an individual in a world of mazes.
We now shift perspective from the individual to the system as a whole. We stop taking mazes as given.
It is time to ask why and how all of this happens, and what if anything we can do, individually or collectively, about it.
In particular, this post presents an explicit model of the first question, which is:
How did mazes come to be, both individually and overall?
This is partly a summary of the model developed so far. It is partly making the model more explicit, and partly the fleshing out of that model with more gears.
If there are points here for which you believe the previous posts failed to lay the groundwork they should have laid, please note that in the comments by number, so I can consider fixing that. Keep this distinct from any disagreements or other notes on these claims, which are also welcome.
- Every organization has an organizational culture. That culture can and does change.
- Those who focus on their own advancement at the expense of other considerations will, by default, advance further, faster and more often. Those who do not do this will not advance. Increasing amounts of focus make this effect increasingly large.
- Focus on one’s own advancement inside hierarchies causes individuals to self-modify in order to be the type of person who automatically engages in maze-creating and maze-supporting behaviors. They will also see such behavior as natural and virtuous.
- Middle management performance is inherently difficult to assess. Maze behaviors systematically compound this problem. They strip away points of differentiation beyond loyalty to the maze and willingness to sacrifice one’s self on its behalf, plus politics. Information and records are destroyed. Belief in the possibility of differentiation in skill level, or of object-level value creation, is destroyed.
- The more one is already within a maze, the more one is rewarded for maze-creating and maze-supporting behaviors, and for self-modifications towards such behaviors. This creates a vicious cycle.
- Focus on one’s own advancement causes one to wish to ally with others who do the same thing. That means allying with those who are engaging in maze-creating behaviors, and who are likely to do so in the future. Those people are likely to have future power. They are aligned in support of your new values and likely actions.
- Changing the organizational culture towards a maze, which we will call raising the maze level, benefits those who wish to engage in maze-like behavior at the expense of those who do not. Those wishing to raise maize levels implicitly, and sometimes explicitly, coordinate together to reward maze behaviors, culture and values, and punish other behaviors, cultures and values.
- Those who wish to have strong allies notice that strong potential allies want to ally with those who support mazes, and with those who ally with those who support mazes, and so on. This creates a strong incentive to strongly signal, in a way that others who support mazes can recognize, our support for maze behaviors and rising maze levels. One does this by supporting maze behaviors and maze allies whenever possible over all other considerations, without any need for explicit coordination or reciprocity, and by other costly signals of maze virtue.
- The larger an organization and the more of a maze it becomes, the closer competition among its middle managers resembles super-perfect competition plus political considerations. Slack is destroyed. Those who refuse to get with the program, where the essence of the program is support of mazes over non-mazes, stop getting promoted or are pushed out entirely.
- Such behaviors choose how to react to people largely by observing those people’s culture and values. Those who wish to get ahead in such worlds must self-modify to instinctively support such actions, whether or not doing so is locally in their self interest. Being too aware of one’s local self-interest is therefore not in one’s broader self-interest. Humans are much better at doing all this, and at detecting it, than they are at faking it. The way one gets such behavior is through cultivating habits, including habits of thought, and choosing one’s virtues. This self-modification creates even stronger implicit coordination.
- There are contravening forces that can potentially outweigh all these effects, and result in maze behaviors being net punished. But they require those opposed to maze behaviors, culture and values to devote substantial resources to the cause, and to bear substantial costs. The more of a maze a place has already become, the harder it will be to turn things around or even stop things from getting worse. If such efforts are to succeed, this needs to be a high priority.
- Even if maze behaviors are net punished for now, those who have embraced the maze nature will be skeptical of this. Even if they observe such behaviors being net punished now, they will not expect this to continue. Given the state of our world and culture, this is a highly reasonable prior. They also have knowledge of their own maze nature and presence within the organization, which is likely to raise maze levels over time and is evidence that the fight against mazedom is failing. And they stand to gain a huge competitive edge by raising the local maze level. Thus, the fight never ends.
- Damage done is very difficult to reverse. Once particular maze behaviors become tolerated and levels rise, it takes a lot of effort to undo that.
- Once people who support mazes are in places of authority in a given area, that area will rapidly become a maze. This is true of organizations, and of sub-organizations within an organization.
- If the head of an organization believes in mazes, and has the time to choose and reward the people of their choice, it’s all over. Probably permanently.
- Mazes reward individuals who engage in maze behaviors and exhibit maze culture and values, and punish those who do not so exhibit, even outside the maze or organization in question. This includes customers, producers, business partners, investors and venture capitalists, board members, analysts, media, government officials, academics and anyone else who supports or opposes such patterns.
- All strengthening of mazes anywhere creates additional force supporting mazes elsewhere. Mazes instinctively support other mazes. As society falls increasingly under the sway of mazes, it implicitly cooperates to push everyone and everything into supporting the behaviors, culture and values of mazes.
- The end result inside any given organization is that maze behaviors grow stronger and more common over time. This is balanced by maze behaviors making the organization less effective, and thus more likely to fail.
- Occasionally an organization can successfully lower its maze level and change its culture, but this is expensive and rare heroic behavior. Usually this requires a bold leader and getting rid of a lot of people, and the old organization is effectively replaced with a new one, even if the name does not change. A similar house cleaning happens more naturally in the other direction when and as maze levels rise.
- Maze behaviors grow stronger and more common over time in any given organization barring rare heroic efforts. As organizations get bigger and last longer, maze levels increase.
- When interacting with a world of low maze levels, or especially when interacting with individuals who have not embraced the maze nature, mazes are at a large competitive disadvantage versus non-mazes. Organizations with too-high maze levels become more likely to fail.
- As organizations fail and are replaced by smaller upstarts via creative destruction, revolution or other replacement, maze levels decrease.
- Replacement of old organizations by new ones is the primary way maze levels decline.
- As the overall maze level rises, mazes gain a competitive advantage over non-mazes.
- If society sufficiently rewards mazes and punishes non-mazes, non-mazes can stop failing less often than mazes. Existing organizations become increasingly propped up by corruption. New organizations will start off increasingly maze-like, signal their intent to become mazes, and raise their maze levels more rapidly. They will still usually start out at much lower maze levels than old organizations.
- New organizations and smaller organizations also have more benefit in survival and growth from non-maze behaviors versus maze behaviors, as they have a greater need to do things mazes cannot do, or that they cannot do without huge additional overhead. Even in the scenario where large organizations benefit from maze coordination more than they are hurt by maze inefficiency, it can still benefit smaller organizations to minimize maze levels.
- Mazes have reason to and do obscure that they are mazes, and to obscure the nature of mazes and maze behaviors. This allows them to avoid being attacked or shunned by those who retain enough conventional not-reversed values that they would recoil in horror from such behaviors if they understood them, and potentially fight back against mazes or to lower maze levels. The maze embracing individuals also take advantage of those who do not know of the maze nature. It is easy to see why the organizations described in Moral Mazes would prefer people not read the book Moral Mazes.
- Simultaneously with pretending to the outside not to be mazes, those within them will claim if challenged that everybody knows they are mazes and how mazes work.
- As maze levels rise, mazes take control of more and more of an economy and people’s lives.
- Under sufficiently strong pressure the maze behaviors, value and culture filter out into the broader society. Maze behaviors, values and culture are seen increasingly as legitimate and comfortable and praiseworthy. This happens even outside of any organization. Non-maze behaviors are increasingly seen as illegitimate, uncomfortable and blameworthy.
- The result of these effects is that people in societies with high maze levels, especially those with power and wealth, increasingly and increasingly openly oppose and vilify the creation of clarity, engaging in any productive object-level action, and participation in or even belief in the existence of positive sum games of any kind. Simulacrum levels continue to rise.
- Given sufficiently high societal maze levels, talk in support of maze behaviors would eventually becoming more and more open, and dominate discourse and how people are educated about the world, as people explicitly and publicly endorse and teach anti-virtues over virtues.
- We would see increasing societal inability to create clarity, engage in actions or do anything other than repeat existing patterns. Costs to all still possible actions would rise. Existing patterns would expropriate increasing portions of remaining resources to keep themselves afloat, and increasingly ban any activity outside those patterns.
- The default outcome on the scale of individual organizations is the rise and fall of those organizations over time.
- The default short term outcome on the scale of a nation, when maze levels and simulacrum levels increase, is declining growth, dynamism, slack, discourse, hope, happiness, virtue and wealth. People increasingly lose the things that matter in life.
- The default long term outcome on the scale of a nation is the rise and fall of civilizations.
- We do in fact see all of this. Here and now.
The next few posts will flesh this out more and provide, as best I can, answers to the other questions.
Next post: How Doomed are Large Organizations?
I think the big question here is #4. Most of the rest kind of just follows, I think, once you understand that this sort of thing exists. The question is, how does it get started in the face of people in charge who have skin in the game that should require them to stop it? Not “how does it get started” by itself; that basically just requires, people are vulnerable to politicking; the rest kind of follows. But when people have reason to stop it, it becomes a question.
Of course, “middle management performance is inherently difficult to assess” is itself a partial answer. But the step to belief in lack of differentiation — so, hey, you shouldn’t even try to weed out the problem — still seems unexplained to me.
I think part of the answer also has to be that people aren’t aware of the problem; but I think part of the explanation for that has to be that mazes don’t actually fail fast enough. Otherwise, every CEO would be on guard (which wouldn’t necessarily be worth enough by itself, keeping out politicking is hard, but it would help). (Of course, this assumes they do in fact have skin in the game… you said before once you’re on top you can’t outrun the problems you cause, but this may not be true if you can go somewhere else.)
So part of the answer has to be, mazes are still capable of doing things to a surprising extent, and then the question becomes why is that true. But also it’s just like… they should still get outcompeted, right? So mazes have to be sufficiently common for them not to. I guess I’m confused what the equilibrium here…
The idea of a moral mazes is related, in my view, to the idea of corporate bullshit: buzzword-laden justifications and explanations, crystallized in sleek powerpoint presentations that consultants were paid tens of thousands of dollars to make, that are passed through the organization with an marked indifference to their truth.
Charitably, their purpose is to produce alignment and avoid paralysis by indecision, which becomes riskier as corporations get larger. They make the decision legible to people who are’t close to the details of the process, which is important for executing a decision properly. As organizations get larger, it’s harder to avoid trade-offs between decision quality and and the ability to execute it. Getting decisions endorsed by highly polished, composed professionals who ooze prestige is one way to make it easier to execute, since “doing what the prestigious person recommends” is an easy heuristic for people in large organizations to follow.
To your request for comment, here are a couple pieces of groundwork I thought were missing (might be b/c I just missed them):
1. A 1-paragraph explanation somewhere in the sequence of what a maze *is* and how it relates to Moloch (I think I understand the answer but it would help to have you state it)
2. A description of what it means for society (as opposed to a single organization) to be maze-like
Substantive comment: I’m not sure “self-modify” is a helpful description of why organizations end up dominated by optimizers and their mazes (point #3). At the scale of one person’s career, behavior is inelastic; culture and character matter much more than incentives. Darwinian selection over the population of all employees, not Lamarckian self-modification, is the main reason why optimizers end up in power. Such optimizers always existed in the population, without having to self-modify– otherwise how would the vicious cycle of mazedom get started?
I don’t have any real grounds for asserting this– it’s just an impression from my (limited) experience. Moral Mazes might have evidence to support or refute it.
Pingback: How Escape From Immoral Mazes | Don't Worry About the Vase
I think point 3 is inadequately justified. Clearly creating mazelike conditions can be helpful to advancement within a hierarchy in some circumstances. I definitely accept without further justification that in a mazelike org, a focus on personal career advancement will inevitably promote habitual maze-supporting and, less inevitably, promote habitual maze-deepening. But it doesn’t seem to follow that this is true in not-yet-mazelike orgs, and hence in the general case. I’m also not clear how general “They will also see such behavior as natural and virtuous.” is intended to be; particularly, does this apply only to the people who have self-modified to be habitual maze-team members, or also to those who have not yet done so but are considering it? The former seems obvious, the latter plausible but nonobvious.
Also, typo in point 10, I believe; chose -> choose.
I think this comment is getting at something critical, which I also question.
If Alice is a maze rat, wouldn’t she be seeking to have a bunch of non-maze people as her peers, to better outrace them?
Typo fixed, thanks.
Other stuff is trickier but yes I think the point strongly stands.
If you are in a non-maze and lack the maze nature, whether or not you can usefully become more maze-like depends on people’s level of reaction to that. If they tolerate it, you definitely want to do it, which will move things towards a maze. If they don’t tolerate it, then that’s all the more reason to modify conditions and move in allies who will allow you to gain the advantage in these ways. Maze tactics by default help you, so it makes sense given your goal to make things more maze-friendly if you are more willing than others to use maze tactics. And those who are more willing, get ahead, and by default care more, etc.
As for the virtue question, this goes hand in hand with the self-modification, with both causing the other. Mazes and the things they support teach us to do this on both ends, reversing our moral intuitions. VCs, for example, are clearly trying to do this to founders.
Consider this scenario:
1. I have 3 peers (eg plant managers of different plants)
2. I know the “next up” position will be opening up in 6-12 months, and one of the 4 of us will get it. Prior probability of promotion (pop) = .25 (25%)
3. One of my peers leaves unexpectedly, and I am the deciding vote on replacing them with a Rat or non-Rat
4a. If I replace them with a non-Rat, our pops are Alice = Bob = Carol < .25 < Me < 1
4b. If I replace them with a Rat rat.pop goes up by some amount, because there is more tolerance/valuing of Maze behavior, but there is now a second Rat, so the pops are Alice = Bob < .25 < Rat = Me < .5
So, adding another Rat would only be better than adding a non-rat if:
1. Being the only Rat has less than 50% chance of promotion
2. There are not so many Rats that the non-rat pop is already 0
That second factor looks like a countervailing force, where the existing Rats want to maintain just enough other Rats around to maintain their advantage, but keep at least one (and probably at least 50% of the relevant population) as non-Rats to improve their own odds?
Adding rats as subordinates or superiors is strictly good for you, because they will help you. Also, adding rats is how you instincts say to operate and helps you do the things you want to do. It’s only for direct rivals that things run in both directions, where yes you want some mix of the two and it’s not clear what it is. But that misunderstands the maze nature – the maze nature isn’t maximizing expected utility, it doesn’t think doing math is a thing, and treats people who think numbers are a thing as losers.
> If you are in a non-maze and lack the maze nature, whether or not you can usefully become more maze-like depends on people’s level of reaction to that. If they tolerate it, you definitely want to do it
I think this is the specific bit which seems like an unjustified leap to me. The maze nature does not seem to be helpful to acquire/self-instill unless a meaningful fraction of your upward reporting chain (not necessarily your boss, but your boss’s boss, their boss, etc.) has the maze-nature.
Pause for teminology: I’m going to define “to be bemazed” as “to have the maze-nature”, because it’s less awkward to talk about this way. And because “rat”, a noun rather than an adjective, obscures the fact that you can move between categories. (c.f. https://pastebin.com/WrDKYkjK from Cryptonomicon)
If everyone who reports to you is bemazed, but you aren’t and no one you directly or indirectly report to is bemazed, I don’t see any benefit to bemazing yourself. It seems to me that bemazement has to flow downhill, because those who are unbemazed, who are judging based on other, less inherently destructive standards, will value progress toward bemazed goals less than progress toward unbemazed goals.
I think some confusing mentioned in the comments about how mazes develop initially is that the previous posts in the sequence didn’t emphasize that mazes benefit rats willing to run them AND, in the absence of (a) knowledge of the existence of mazes or their extent; and furthermore (b) a consistent, concerted, and active effort to prevent mazes from developing (further); maze-running is locally advantageous.
One example of maze-like-ness is something previously only ascribed to managers: un-differentiation. Working, even ‘on the line’, for a manager that either can’t or for some reason *refuses* to differentiate among their subordinates based on (even weakly) objective criteria can create a mini-maze all of its own. Some managers skin is actively insulated from being in the game, e.g. nepotism.
Typo in #14: you mean sub-organizations withIN an organization. Presumably.
Yes, I will fix. Thank you.
Pingback: Mazes Sequence Summary | Don't Worry About the Vase
Pingback: Unifying the Simulacra Definitions | Don't Worry About the Vase
I’m loving this series so far, but there’s a few points that I’ve been unclear on for a while, and I’d really like to clear it up:
You keep using the term “maze levels”, and even having read all the previous posts in the series, I still feel like I only have a vague intensional definition of what “maze levels” refer to, rather than an extensional understanding of what “maze levels” mean exactly. I.e.
1. What are the characteristics of “low maze levels”? What do agents within those mazes do that agents outside of mazes do not do, and vice versa? I know that they begin to remove all slack (very solid definition of that in your slack reference post, thank you!) from each other’s lives, and that they begin to make political alliances based on pro-maze values. There also seems to be an implicit-but-not-explicit suggestion that other illegal/horrible/evil things start being done, to people, places, & things external to the maze & organization, but a range of real examples of such things seems so far to be left unsaid. I could probably guess, but I could easily guess wrong.
2. What are the characteristics of “higher maze levels”? In what specific ways should experiences with them & within them be expected to differ from experiences with & within “low maze levels” & non-mazes?
The answers to these questions may well be buried in the Quotes from Moral Mazes page, or in the book itself. I haven’t gone too deeply through those yet because they were marked as “not required” back at the beginning.
Also, I read that link on “Simulacrum Levels”, and I’m sorry this will unavoidably sound rude, but in the excerpts, the author of that book seems to talk about things in that very vague ‘high-level’ manner of people who don’t concretely understand what they’re talking about. (I know there’s a post somewhere on LessWrong warning about people who use overly vague language.) My understanding of “Simulacrum Levels” is thus also nebulous, but I suspect that nebulosity may be a problem with the concept itself rather than my understanding of it.
Again, loving the series so far, and thank you for writing it!
Maze level refers to the general amount of maze-style activity and influence in a system or nation/culture – the amount to which one anticipates that rewards and punishments and ability to do things are determined by dynamics inherent in mazes, and to which one can expect to face maze-style interactions, and so on.
On Simulacrum stuff, the book itself is continental philosophy and basically unreadable, I consider my model of it to not depend on the book although it takes its terms and inspiration from it. But also yes there’s a lot of nebulosity (I don’t use that term myself ever but it seems to apply here).
Low maze levels means that the dynamics here mostly don’t apply, and people are not evaluating others largely as potential allies and according to whether or not they will follow the values and principles of mazedom. Morality is considered a feature rather than a bug, being too selfish or scheming a bug rather than a feature, anticipation of how things will look to others in the future are not central, people have the felt experience of having slack and choice, and so on. There is the sense that something might ever happen or get done for reasons, and that someone one encounters will care about whether value is being brought to the table or created, etc. When you come up with or hear an idea you think “will it work? it is a good idea?” rather than “will this help my boss sell an image to his boss? will this make me look like a person who is likely to succeed?” Versus the opposite for high.
Thank you very much, that is exactly what I needed to know!
I’m about 1/2 way through your list. Encourage more small business, looks like the answer.
And I like that solution, except big business is more efficient somehow… well saving in scale.
We could have some small business cooperative, which sounds communistic. If you sign up with a cooperative to buy screws (or something) how do you know someone isn’t ripping you off?
Big business has many advantages, a lot of them are regulatory arbitrage – high fixed costs are imposed and you only have to pay them once, so you win in a fight. One could imagine different rules that didn’t do that. A lot more of them are reputation – people who think working for big business is good so they get to hire better. And some of them are really awful, like small businesses struggling to buy health insurance.
How do you know the cooperative isn’t screwing you with the screw price? You can always check amazon’s price or home depot’s, and continue to check every so often to be sure. You still have to verify quality of course, but prices are easy to check.