"I think I am, to give a straight answer to Ben’s question, committed to the view that in an important sense, there’s actually no such thing as metis."
I like this framing a lot. Perhaps then a practical definition of metis is actionable knowledge that is not yet techne. There are arguments to be made about whether all metis can become techne, and these will mimic some of the arguments about the inevitability of "artificial intelligence," whatever that might be.
Trying to stay practical, one of the challenges for the cyberneticists is an is/ought problem. Cyberneticists assert that bureaucracy ought to be agile and responsive. Anarchist anthropologists and historians like Scott chronicle how all implementations of bureaucracy become inflexible, rigid, rule-based authorities.
Certainly we should not let history constrain the future, but it frames the study. Is it hard to engineer interconnected adaptive bureaucracies because people did it badly in the past or because bureaucratic rigidity is an attractive fixed point of human dynamics?
The thing I struggle with around books like Scott’s (or Graeber, with whom he shared some instincts I think) is that the last 500 years have been a running experiment in running more practical, human centred societies next to ones governed by bureaucracy with an increasing thirst for record keeping and process. And the current state of that is that every single one of the former has been overrun and taken over by one of the latter, and that on the whole societies with more bureaucracy (aka state capacity) , with more techne do better than those with less.
It’s true that those with more sophisticated response systems within the bureaucracy, as you describe, can create somewhat better outcomes for citizens than those without.
But on the other hand being deliberately deaf to its citizens’ wishes gives the state more options to endanger and perhaps overthrow its neighbours, as can be seen from the fact that North Korea still is a risk to South Korea, or Russia endangers the EU, despite in both cases being economic and demographic pipsqueaks compared to their opponents.
I kind of agree with this - Scott spent his whole career writing about agrarian reform and resistance, but when you actually look at SouthEast Asia, you have to say that he must have missed the big picture.
For me it’s the big anarchist blind spot. I think you said in your book that the cyberneticians took it as an important reference point in looking at an organisation that it was *continuing to exist* - this told you that it had found a sustainable equilibrium for now. Essentially all anarchist-approved societies tend to fail this test, some very rapidly. As someone who has emotional sympathy with a lot of their project I find that frustrating but it can’t be ignored.
Graeber was the only one who managed to articulate this explicitly, but if you read anarchist works, a common subtext is that organizations that continue to exist *is* the problem (they'll call it hierarchy but it's not a blind spot for anarchists, they see it quite clearly and just don't like it), because when a new problem arises that said structure can't solve, more often than not it'll preserve itself to the detriment of the people living under it.
It's why in The Dawn of Everything, Graeber keeps wondering why societies tens of thousand of years ago had more varied arrangements, and some even changed structures seasonally, and current societies stopped being able to do that.
Isn't that a bit like me saying "Oleksander Usyk is currently unbeaten as a boxer, that's boring plus I think I'm a nicer person than Usyk - so I'll be fighting him tomorrow"?
I mean, a lot of these guys are interested in praxis, right? What use is it to put forward a system, however nice, that inevitably (and I mean inevitably, I'm not sure there's a more dependable medium- to long-term rule in history than 'bureaucracy conquers anarchy') gets crushed by by the opposition, however nasty?
Another side of *metis* is that it can only be included when you see it. Like a camera that can only tell "a" truth - the one in front of it. It never reveals what's *behind* it.
I find this interesting in a specifically Chinese context. Mainland Chinese society has a lot of deeply embedded feedback mechanisms that are designed to ensure more senior officials receive accurate information about the ground state, for the purpose of decisionmaking. Occasionally this is referred to by chinese academics as a form of democracy, in the sense that if enough people are grumbling about something there is a good chance it will make its way to the ears of someone in a position to do something about it
> “metis”, in my view, really just means “everything that didn’t seem worth the effort to turn into techne”.
I'm not sure I'm convinced by this .... If you go back and look at Scott's examples of practical skill there are probably some that could be formalized into actionable rules but I don't think that's going to go all the way. You could inculcate a football team and its individual players with all the rules you like but in a 90-minute real-time contest, some people are going to have the ability to read the game and act on their reading in real situations and some aren't and are never going to have.
counterpoint! there is actually a bureaucracy which goes around formalizing rules about football - it's just that rather than running a bureaucracy, they use them to produce the FIFA2024 video game. it would be unrealistic to expect a bureaucracy to actually play a game that way, but that's why they employ football managers.
And I think that for the Scott's argument to work, he needs more than the AI-sceptic case that some tasks can't be brought down to algorithms. "Metis" has to be a form of experience that isn't available to a bureaucracy, so it would need an example where "why can't they just hire a specialist manager" doesn't work.
Not sure I understand your reply here. First of all, football managers don't supply all the on-field nous, most of it resides in the players. (I'm reminded of Shankly's remark to a trainee, "The trouble with you, son, is that all your brains are in your head".) But more to the point, Scott can't have believed that metis is unavailable to bureaucracies as such since his examples include many cases where they employ people with it. See for example the list on p.314 "Firefighters, rescue squads, paramedics, mine-disaster teams etc."
This is the thing I'm trying to describe - all the most convincing examples of Metis are physical, embodied activities that resist quantification because they don't involve language. But the ones that Scott needs for the distinction to be important need to be specifically *administrative* capabilities. So he ends up trying to extend it to things like "how to farm in a particular terrain" or "how to fight an oil well fire", where it's much less obvious that things can't be turned into techne. (The agricultural examples never made any sense to me at all, as agronomy has been one of the huge successes of the last century, while the firefighting ones are more interesting because this is an area in which group decision making has been really closely studied; there's a big and respectable school of thought saying that the structured intuition of experienced fire captains can't be improved on, but it's not the only view by any means).
To me, it looks like another example of the "Anarchist Exception", the little admission that actually gives the whole game away, like Nozick's ("of course, you have to allow for things like compulsory vaccination programs") or Graeber's ("I'm not going to consider commercial debts between businesspeople"). Actual textbooks for bureaucrats often start by admiring the East India Company's ability to control a continent with a staff of a couple of hundred, but that sort of organisational structure isn't what he wants to talk about.
(Sidebar - I am not conceding here! Every year, there are fewer and fewer footballers who are better at "reading the game" than the AI of that year's FIFA, this is a matter of cost and computational convenience rather than something that's impossible in principle).
Your argument has a “How do we get a bureaucracy with strong feedback?”-shaped hole in it.
Even private entities often die out instead of changing their ways—and they live in feedback rich environments. For public service the rule is that even clearly incompetent employees can’t get fired.
Singapore did it, but they implemented the vision of one guy not vulnerable to being voted out. And they were lucky that the guy not vulnerable to being voted out wasn’t an idiot.
With one bit of information voted in every four years or so, what’s the chance of getting any sort of feedback rich bureaucracy?
Nitpick-that’s-not-really-a-nitpick: voting is not the only source of bureaucratic information (and even voting is more than one bit of information). I’m not minimising the challenges involved in achieving a ‘feedback rich bureaucracy’, but this is grossly overstating them. (Perhaps you know this, and it was an exaggeration for effect; either way, worth clarifying.)
Yes. There is lobbying, which sometimes works for the best. Some parts of the public (for instance actual experts) either scream at or nudge lawmakers to change the system in certain ways.
But if those screams can’t be potentially leveraged into votes they often fall on deaf ears. And lobbyists with deep pockets do have clear cut ways of leveraging their money into votes.
There could also be the creation of some public service virtue ethics that’s instilled into everyone and especially into politicians and public servants. But enforcing any sort of virtue ethics is not really part of current western political culture.
Either way, the methods of enacting any sort of feedback rich bureaucracy are extremely haphazard. And it’s unlikely that such a system can be a “set it and forget it “ system. So even if you get lucky and you implement it once, you can’t reliably maintain it.
Faced with this, the rational choice is to limit such bureaucracy as much as possible—while trying to improve it as best you can (making it feedback rich for instance).
I think this probably also differs a little between what you could call the policy bureaucracy vs. the service delivery bureaucracy (there is probably a better cybernetic framing of this distinction). The latter tends get a *lot* more feedback than even the slightly expanded mechanisms you’ve described above. I’d say a fair bit of this tends to feed back to the policy bureaucracy, and most of it seems to be seen as potentially vote-influencing by politicians.
I started to write about this in The Unaccountability Machine but didn't really do it justice as it's a whole nother book, but we have to begin from the point of view of recognising that although things haven't been going all that great recently, the USA and the global capitalist system has done really pretty well at adapting to changing conditions over the last century. Maybe they just got lucky, but I think there's a lot more implicit communication channels than we necessarily recognise - (spoiler for the last chapter - in my view, a lot of things we call "crises" because we're living through them and they suck, are actually the system's means of reorganising and stabilising itself).
I think it’s OK to still call this lobbying, just probably a more wholesome version of it. I think the main reason this type of lobbying is wholesome is that the goals are relatively clear (e.g. no one likes long lines) and also easy to audit.
When the Amazon bureaucracy tried to BS Bezos about short phone waiting times, he just picked up the phone and called customer service, proving them wrong.
If you could decompose all bureaucracy goals into unambiguously good things and cheaply auditable metrics then yes, there would be no problem with implementing feedback rich bureaucracy. (At least if we’re patient enough)
In the real world, even if it’s theoretically possible to decompose bureaucratic goals like that you could have:
1. An unreasonable tradeoff between clarity and other goals (e.g. judges need to have some latitude in sentencing)
2. The time it takes to establish such a decomposition scientifically enough is too long: by the time this is established, the problem on the ground changed.
That’s why the metis vs techne debate does not really decide the issue: what’s important is that a problem gets solved by the political sausage making before the problem on the ground changes. Arguing that metis exists for some problems is just an extreme example of this, when the problem can never be decomposed into unambiguous good and cheaply auditable goals.
But the real world problem is much harder than just eliminating metis, when you take political considerations into account.
While there’s a sense in which you can define terms however you like, I don’t think a label like ‘lobbying’, that is so strongly associated with influencing government, is very useful for describing a phenomenon that is if anything more prevalent in relation to non-govt service delivery. It’s apt to just lead to misinterpretation and miscommunication imho
“Everything that didn’t seem worth the effort to turn into techne?” Mr. Davies, I have the wrathful spectre of a Professor Scott holding for you on line two…
"I think I am, to give a straight answer to Ben’s question, committed to the view that in an important sense, there’s actually no such thing as metis."
I like this framing a lot. Perhaps then a practical definition of metis is actionable knowledge that is not yet techne. There are arguments to be made about whether all metis can become techne, and these will mimic some of the arguments about the inevitability of "artificial intelligence," whatever that might be.
Trying to stay practical, one of the challenges for the cyberneticists is an is/ought problem. Cyberneticists assert that bureaucracy ought to be agile and responsive. Anarchist anthropologists and historians like Scott chronicle how all implementations of bureaucracy become inflexible, rigid, rule-based authorities.
Certainly we should not let history constrain the future, but it frames the study. Is it hard to engineer interconnected adaptive bureaucracies because people did it badly in the past or because bureaucratic rigidity is an attractive fixed point of human dynamics?
The thing I struggle with around books like Scott’s (or Graeber, with whom he shared some instincts I think) is that the last 500 years have been a running experiment in running more practical, human centred societies next to ones governed by bureaucracy with an increasing thirst for record keeping and process. And the current state of that is that every single one of the former has been overrun and taken over by one of the latter, and that on the whole societies with more bureaucracy (aka state capacity) , with more techne do better than those with less.
It’s true that those with more sophisticated response systems within the bureaucracy, as you describe, can create somewhat better outcomes for citizens than those without.
But on the other hand being deliberately deaf to its citizens’ wishes gives the state more options to endanger and perhaps overthrow its neighbours, as can be seen from the fact that North Korea still is a risk to South Korea, or Russia endangers the EU, despite in both cases being economic and demographic pipsqueaks compared to their opponents.
I kind of agree with this - Scott spent his whole career writing about agrarian reform and resistance, but when you actually look at SouthEast Asia, you have to say that he must have missed the big picture.
For me it’s the big anarchist blind spot. I think you said in your book that the cyberneticians took it as an important reference point in looking at an organisation that it was *continuing to exist* - this told you that it had found a sustainable equilibrium for now. Essentially all anarchist-approved societies tend to fail this test, some very rapidly. As someone who has emotional sympathy with a lot of their project I find that frustrating but it can’t be ignored.
Graeber was the only one who managed to articulate this explicitly, but if you read anarchist works, a common subtext is that organizations that continue to exist *is* the problem (they'll call it hierarchy but it's not a blind spot for anarchists, they see it quite clearly and just don't like it), because when a new problem arises that said structure can't solve, more often than not it'll preserve itself to the detriment of the people living under it.
It's why in The Dawn of Everything, Graeber keeps wondering why societies tens of thousand of years ago had more varied arrangements, and some even changed structures seasonally, and current societies stopped being able to do that.
Isn't that a bit like me saying "Oleksander Usyk is currently unbeaten as a boxer, that's boring plus I think I'm a nicer person than Usyk - so I'll be fighting him tomorrow"?
I mean, a lot of these guys are interested in praxis, right? What use is it to put forward a system, however nice, that inevitably (and I mean inevitably, I'm not sure there's a more dependable medium- to long-term rule in history than 'bureaucracy conquers anarchy') gets crushed by by the opposition, however nasty?
Another side of *metis* is that it can only be included when you see it. Like a camera that can only tell "a" truth - the one in front of it. It never reveals what's *behind* it.
I find this interesting in a specifically Chinese context. Mainland Chinese society has a lot of deeply embedded feedback mechanisms that are designed to ensure more senior officials receive accurate information about the ground state, for the purpose of decisionmaking. Occasionally this is referred to by chinese academics as a form of democracy, in the sense that if enough people are grumbling about something there is a good chance it will make its way to the ears of someone in a position to do something about it
> “metis”, in my view, really just means “everything that didn’t seem worth the effort to turn into techne”.
I'm not sure I'm convinced by this .... If you go back and look at Scott's examples of practical skill there are probably some that could be formalized into actionable rules but I don't think that's going to go all the way. You could inculcate a football team and its individual players with all the rules you like but in a 90-minute real-time contest, some people are going to have the ability to read the game and act on their reading in real situations and some aren't and are never going to have.
counterpoint! there is actually a bureaucracy which goes around formalizing rules about football - it's just that rather than running a bureaucracy, they use them to produce the FIFA2024 video game. it would be unrealistic to expect a bureaucracy to actually play a game that way, but that's why they employ football managers.
And I think that for the Scott's argument to work, he needs more than the AI-sceptic case that some tasks can't be brought down to algorithms. "Metis" has to be a form of experience that isn't available to a bureaucracy, so it would need an example where "why can't they just hire a specialist manager" doesn't work.
Not sure I understand your reply here. First of all, football managers don't supply all the on-field nous, most of it resides in the players. (I'm reminded of Shankly's remark to a trainee, "The trouble with you, son, is that all your brains are in your head".) But more to the point, Scott can't have believed that metis is unavailable to bureaucracies as such since his examples include many cases where they employ people with it. See for example the list on p.314 "Firefighters, rescue squads, paramedics, mine-disaster teams etc."
This is the thing I'm trying to describe - all the most convincing examples of Metis are physical, embodied activities that resist quantification because they don't involve language. But the ones that Scott needs for the distinction to be important need to be specifically *administrative* capabilities. So he ends up trying to extend it to things like "how to farm in a particular terrain" or "how to fight an oil well fire", where it's much less obvious that things can't be turned into techne. (The agricultural examples never made any sense to me at all, as agronomy has been one of the huge successes of the last century, while the firefighting ones are more interesting because this is an area in which group decision making has been really closely studied; there's a big and respectable school of thought saying that the structured intuition of experienced fire captains can't be improved on, but it's not the only view by any means).
To me, it looks like another example of the "Anarchist Exception", the little admission that actually gives the whole game away, like Nozick's ("of course, you have to allow for things like compulsory vaccination programs") or Graeber's ("I'm not going to consider commercial debts between businesspeople"). Actual textbooks for bureaucrats often start by admiring the East India Company's ability to control a continent with a staff of a couple of hundred, but that sort of organisational structure isn't what he wants to talk about.
(Sidebar - I am not conceding here! Every year, there are fewer and fewer footballers who are better at "reading the game" than the AI of that year's FIFA, this is a matter of cost and computational convenience rather than something that's impossible in principle).
Perhaps, but whatever the FIFA AI is doing, one thing it is not doing is actually playing a game of football.
To be honest I would not mind if you spend the rest of the year elaborating on all of the variations on this...
Your argument has a “How do we get a bureaucracy with strong feedback?”-shaped hole in it.
Even private entities often die out instead of changing their ways—and they live in feedback rich environments. For public service the rule is that even clearly incompetent employees can’t get fired.
Singapore did it, but they implemented the vision of one guy not vulnerable to being voted out. And they were lucky that the guy not vulnerable to being voted out wasn’t an idiot.
With one bit of information voted in every four years or so, what’s the chance of getting any sort of feedback rich bureaucracy?
Nitpick-that’s-not-really-a-nitpick: voting is not the only source of bureaucratic information (and even voting is more than one bit of information). I’m not minimising the challenges involved in achieving a ‘feedback rich bureaucracy’, but this is grossly overstating them. (Perhaps you know this, and it was an exaggeration for effect; either way, worth clarifying.)
Yes. There is lobbying, which sometimes works for the best. Some parts of the public (for instance actual experts) either scream at or nudge lawmakers to change the system in certain ways.
But if those screams can’t be potentially leveraged into votes they often fall on deaf ears. And lobbyists with deep pockets do have clear cut ways of leveraging their money into votes.
There could also be the creation of some public service virtue ethics that’s instilled into everyone and especially into politicians and public servants. But enforcing any sort of virtue ethics is not really part of current western political culture.
Either way, the methods of enacting any sort of feedback rich bureaucracy are extremely haphazard. And it’s unlikely that such a system can be a “set it and forget it “ system. So even if you get lucky and you implement it once, you can’t reliably maintain it.
Faced with this, the rational choice is to limit such bureaucracy as much as possible—while trying to improve it as best you can (making it feedback rich for instance).
I think this probably also differs a little between what you could call the policy bureaucracy vs. the service delivery bureaucracy (there is probably a better cybernetic framing of this distinction). The latter tends get a *lot* more feedback than even the slightly expanded mechanisms you’ve described above. I’d say a fair bit of this tends to feed back to the policy bureaucracy, and most of it seems to be seen as potentially vote-influencing by politicians.
I started to write about this in The Unaccountability Machine but didn't really do it justice as it's a whole nother book, but we have to begin from the point of view of recognising that although things haven't been going all that great recently, the USA and the global capitalist system has done really pretty well at adapting to changing conditions over the last century. Maybe they just got lucky, but I think there's a lot more implicit communication channels than we necessarily recognise - (spoiler for the last chapter - in my view, a lot of things we call "crises" because we're living through them and they suck, are actually the system's means of reorganising and stabilising itself).
I think it’s OK to still call this lobbying, just probably a more wholesome version of it. I think the main reason this type of lobbying is wholesome is that the goals are relatively clear (e.g. no one likes long lines) and also easy to audit.
When the Amazon bureaucracy tried to BS Bezos about short phone waiting times, he just picked up the phone and called customer service, proving them wrong.
If you could decompose all bureaucracy goals into unambiguously good things and cheaply auditable metrics then yes, there would be no problem with implementing feedback rich bureaucracy. (At least if we’re patient enough)
In the real world, even if it’s theoretically possible to decompose bureaucratic goals like that you could have:
1. An unreasonable tradeoff between clarity and other goals (e.g. judges need to have some latitude in sentencing)
2. The time it takes to establish such a decomposition scientifically enough is too long: by the time this is established, the problem on the ground changed.
That’s why the metis vs techne debate does not really decide the issue: what’s important is that a problem gets solved by the political sausage making before the problem on the ground changes. Arguing that metis exists for some problems is just an extreme example of this, when the problem can never be decomposed into unambiguous good and cheaply auditable goals.
But the real world problem is much harder than just eliminating metis, when you take political considerations into account.
While there’s a sense in which you can define terms however you like, I don’t think a label like ‘lobbying’, that is so strongly associated with influencing government, is very useful for describing a phenomenon that is if anything more prevalent in relation to non-govt service delivery. It’s apt to just lead to misinterpretation and miscommunication imho
“Everything that didn’t seem worth the effort to turn into techne?” Mr. Davies, I have the wrathful spectre of a Professor Scott holding for you on line two…