Dominic Cummings lays out how a project to change US politics could look.

New Comment
73 comments, sorted by Click to highlight new comments since: Today at 2:19 AM
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

One nice thing about startups is that they mostly fail if they aren't good.  When MySpace stagnated there wasn't one blessed successor, there were 100 different ones that had to fight it out.  The winner is, modulo the usual capitalist alignment failure, a better company than MySpace was.  Most of its competitors weren't.  From society's perspective this filter is great, maybe the best thing about the whole startup ecosystem.

Cummings doesn't seem to know this.  Replacing the Pentagon with a new organization ABC Inc. is not the hard part (although it is pretty hard).  What's hard is to know that you should pick ABC Inc. and not DEF GmbH.  Cummings thinks what makes startups good is their youth (he wants to sunset them after 15 years, for example), but that's wrong: most young startups aren't good, and fail.  To make it work you need 100 successor Pentagons, and some way of making them compete.

Is that a serious blow to Cummings' thesis, though? If your idea is, say, "we need a ministry of eduction but the current one is irrecoverably broken", you don't need to invent Google or Amazon to replace it. Depending on how bad the status quo is (which is partly an empirical question), any random thing you come up with might be better than what's already there. In which case his use of the term "startup" would be misleading, but the overall thesis would stay relatively intact. That said, I am quite sympathetic to Chesterton's Fence in this argument. In particular, trying to abolish and replace the Pentagon on day 1 of a new administration (as Cummings suggests in his essay) is... optimistic in a world where other nations can hear you say that.
Cummings seems to be making this same argument in the comments: the Pentagon is so unbelievably awful that its replacement doesn't have to be good, you can pick its successor at random and expect to come up with something better. To believe this requires a lack of imagination, I think, an inability to appreciate how much scope for failure there really is. But this is not really a question we can settle empirically -- we can only talk in vague terms about most of what the Pentagon does, and the counterfactuals are even less clear -- so I won't argue the point too much. More seriously, not every young organization is a startup. A new bowling team is not a startup, a new group at Amazon working on a new service is not a startup, and when Camden NJ replaced its whole police force with an entirely different organization that was not a startup either. "Startup" has a lot of specific connotations which mostly don't apply here. And yet, it's the word that Cummings picked. Maybe he doesn't know this stuff, even though it's widely known to many people. Or maybe he does know, and used it anyway. I think this is why people keep coming up with Straussian readings of this essay: they have a sense that he's not sincere about his intended methods and goals. For what it's worth, I don't think Cummings plans to help overthrow USG (and I don't think Yarvin does either): he's getting paid real money to rehash old grievances in front of a friendly audience, that's all. Put him in the same bucket as Paul Krugman.
Any random thing you come up with might be better than what's already there. But it might also be worse, even if what's already there is terribly broken. Maybe there are cases where institutions are so spectacularly screwed up that literally anything you might do is likely to be better, but I wouldn't bet on there being many.
That seems like the crux of the issue. I can absolutely imagine a world where institutions have become sufficiently bad due to misaligned incentive structures that any random thing would be an improvement (though I would still not want to settle for random). For instance, take Gall's Law ['s_law]: "A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over with a working simple system." We might be in a situation where institutions that originally worked have increased in complexity to the point that they no longer work, in which case they might be unfixable and one would have to start over. In any case, whether we actually live in that world seems like an empirical question, but when I compare the world of today to my sky-high expectations (from e.g. transhumanism), there are tons of institutions I would not call particularly functional. Finally, some institutions are considered so bad by some people that they'd rather abolish them with no replacement than leave them as-is, in which case you wouldn't even necessarily need the random thing to replace it. Zvi's FDA Delenda Est stuff comes to mind.
I agree that we might be in a world where the institutions are unfixably bad and the only thing to do is to start over. But if reality and your expectations diverge, I'm not sure it's a good general practice to assume that reality is at fault. Perhaps those expectations were unrealistic.
Again, that seems like a general argument to do nothing. Like, you could have justified not rebelling against slavery when slavery was a thing in the US.
That's not a good example, since the North already functioned without slavery.
You could use that as a general argument against about any change. Funny to find that on a transhumanist forum.
Obviously what you actually have to do is to figure out whether the positives outweigh the negatives. But if all I'm offered is "things are pretty bad so any change is probably good" I'm pretty comfortably replying "new things like this are usually pretty bad so there's a real risk that the change makes things worse". :-)
I think we don't have the same way of imagining what a "random alternative" would be like. For example, I don't imagine that a random alternative would be the kind generated by a monkey randomly typing on a keyboard in a reasonable amount of time. Or even the kind generated by an unexperienced child or teenager. I imagine whoever would have the chance to enact an alternative would be more likely to understand how not to "break society by mistake" than a randomly chosen person in the population. I might be totally off making that analogy, but you seem to me like my aunt who's afraid her computer is broken every time an unexpected window pops up in her browser. She sees her computer as something beyond comprehension, where changing even the tiniest thing could cause iredeemable damage. In reality, an experience computer user makes plenty of changes that would frighten her, but are safe. And her computer is full of useless stuff that were auto-installed by/with other stuff and slow it down.
I endorse Taran's comment that's a sibling of this one. Most startups fail, even though they are generally run by smart hardworking people who have spotted something that could genuinely be better. Let's run with your computer software analogy. Ever worked on the insides of a large "mature" software system? It's common for those to be full of cruft and mess and things no one quite understands and unexpected interactions, such that small changes really can cause severe damage. It's also notorious that trying to do a wholesale rewrite of such a system is usually a bad move. The situation there is similar to the one with startups, and indeed is sometimes literally the actual same situation. Eventually your big old crufty legacy-software system will likely get replaced by something smaller and simpler that does the job well enough and is easier for its developers to work on. (That will probably be made by a startup.) But any particular attempt to replace it, your own included, is likely to fail.
I think there's a difference because the legacy software doesn't develop itself the way a bureaucracy does. It's not made up out of actors that try to get more power for themselves.
I agree with Taran's comment as well. I possibly underestimated how likely to fail an attempt at replacing the current system is. I just think the danger of letting the situation rot is underestimated too. The world is moving on, fast. To keep the software analogy, we're keeping the same legacy software, but demanding it be used on new use cases every year. That's not sustainable. I'm open to third options.
The original startup analogy might be a useful intuition pump here. Most attempts to displace entrenched incumbents fail, even when those incumbents aren't good and ultimately are displaced. The challengers aren't random in the monkeys-using-keyboard sense, but if you sample the space of challengers you will probably pick a loser. This is especially true of the challengers who don't have a concrete, specific thesis of what their competitors are doing wrong and how they'll improve on it -- without that, VCs mostly won't even talk to you. But this isn't a general argument against startups, just an argument against your ability to figure out in advance which ones will work. The standard solution, which I expect will apply to transhumanism as to everything else, is to try lots of different things, compare them, and keep the winners. If you are upstream of that process, deciding which projects to fund, then you are out of luck: you are going to fund a bunch of losers, and you can't do anything about it. If you can't do that, the other common strategy is to generate a detailed model of both the problem space and your proposed improvement, and use those models to iterate in hypothesis space instead of in real life. Sometimes this is relatively straightforward: if you want the slaves to be free, you can issue a proclamation that frees them and have high confidence that they won't be slaves afterward (though note that the real plan was much more detailed than that, and didn't really work out as expected). Other times it looks straightforward but isn't: sparrows are pests, but you can't improve your rice yields by getting rid of them. Here, to me the plan does not even look straightforward: the Pentagon does a lot of different things and some of them are existentially important to keep around. If we draw one sample from the space of possible successors, as Cummings suggests, I don't think we'll get what we want.
Exactly! Hes mistaken survival of the salient examples for some kind of intrinsic quality.

I have extremely mixed feelings about this and similar proposals. On the one hand, the diagnosis seems to be correct to a significant extent, and it's something that very few others are willing to talk about, and it also explains many otherwise hard to explain facts about the lack of recognition of institutional failures after covid (though contrary to what Cummings says there has been some such soul-searching which I've discussed in a few previous comments).

So there's a huge amount of important, non-trivial truth to this proposal.

On the other hand, from the outside, how would you distinguish what he's proposing from an actual authoritarian power-grab?

The media portrays a ‘conservative’ government actually controlling the government as proto-fascist ... the rule of law’ is now often used as a slogan to justify judges deciding political issues

You're telling your target audience that what you are attempting will be very hard to distinguish from an authoritarian, proto-fascist, rule-of-law denying attempt to take power for yourself. I fully believe that this isn't what Cummings wants, but even assuming he's 100% sincere, this still presents a problem.

The problem is that in order to fi... (read more)

Lots of bureaucracies did better than the US bureacracy, so theres a blueprint for fixing bureacracies that doesn't involve disbanding them, or implementing epistocracy. Other countries do it by holding enquiries and firing people. Cummings discusses these problems in a very abstract way, as though they are universal, but things actually function differently in different places. It's noticeable that some places with strongman leaders, like Brasil, did really badly (worse than the US and UK) under COVID... while some technocratic places with bland leaders did really well.
2Sammy Martin1y
I agree - and in fact small doses of what Cummings suggests does just look like holding enquiries and firing people, and maybe firing the leadership of a particular organisation (just not like 50% of all govt departments in one go). In fact in my original question to Brennan [] , I asked and I listed some examples of particular bureaucracies that did well in countries that in general failed (one of which was the vaccine taskforce set up, in part, by Cummings). So clearly it is possible to just get the particular thing right without solving all the systemic issues. My point was that, if you've decided you need wholesale reform of how government makes decisions, doing a complete end-run around most existing institutions to build your 'startup' replacements has a much worse downside than e.g. experimenting with epistocracy, because it concentrates power in a really small number of people, while epistocracy doesn't. But I don't think either is what we should be reaching for to solve a particular imminent problem.
I'm not very familiar with Brennan's work, but I can't imagine how epistocracy could be feasible in the US...its just an invitation to civil war 2.0. Edit So..."we" the technocrats recalculate to get whatever result "we" like. And everyone tolerates having their actual vote erased and replaced with what they should have voted for.....yeah.

In this piece, Dominic Cummings appears to be endorsing both of the following propositions:

  • The current and recent-past elected leaders of the US and the UK have been incredibly useless and incompetent, leading to (e.g.) huge numbers of preventable deaths in the Covid-19 pandemic.
  • It would be a good idea to have "the government actually controlling the government".

Since the second of these would put more power (maybe much more power) in the hands of those elected leaders, it seems like one of three things must be the case:

  • He is confident that the only paths that lead to "the government actually controlling the government" also lead to much better leaders, perhaps because in order to grab that power they would need to be better leaders.
  • He is confident that if we had "the government actually controlling the government", the real power would not be in the hands of the Trumps and Bidens and Johnsons, but of their Cummings-like advisors whom he expects to be able to wield it responsibly and effectively.
  • He is lying or bulshitting or thinking unclearly somehow. (In this category I include options like: what he really wants is an authoritarian state and he doesn't much care about what it doe
... (read more)
There are other possibilities here. For instance, Matt Yglesias has written multiple articles on abolishing the filibuster, and one of his arguments was (to badly paraphrase from my lousy memory) that it's bad in a democracy when the winners of an election can't govern, even if one considers these winners bad or evil or something. For one thing, it means even if the winners are "good", they still can't govern. For another, it means voters get increasingly disillusioned because not enough changes after an election. And finally, if politicians are allowed to implement terrible policies, these terrible policies will actually materialize, in which case voters can learn that they're terrible, and if they don't like them they can vote the bums out who implemented them. (Of course this sets limits to what you should be able to do even with a strong majority, e.g. no-one should be able to stop elections. And separately, it assumes that the electorate has a sufficient ability to learn.)
What's the alternative anyway? If the winners of an election can't govern, then the election is a sham. It's not only a waste of time, it's a deliberate misdirection meant to fraudulently legitimate the actual power.
Your point: if Cummings thinks elected leaders are incompetent, isn't it a problem that his solution is giving them more power? The way I read it: Cummings says elected leaders are incompetent exactly (or at least firstly) because they fail to exert power. Ergo, if they did have more power, de facto they'd be less incompetent.
That would be the possibility of conflating "better" with "stronger". I don't find the prospect of stronger leaders as such an encouraging one, because while Lincoln and FDR were strong leaders who used their strength to do good things there are plenty of examples (I think rather more) of strong leaders who used their strength to do very bad things. Taking what I think is an optimistic view of Cummings's motivations: I think he wants strong leaders who can push through reforms that will make their country better. But if you just optimize for strong leaders, which is pretty much what he seems to be trying to do, I think the default outcome is that you get strong leaders who can push through reforms that will make their country worse.
If a strong leader is a prerequisite for any improvement, what choice do we have? I think that's his point of view, and it makes sense (as in, it's consistent). The way to counter it would be to show a path to lasting improvement that does not require a strong leader.
If you are not in the worst possible dystopia, you have the choice of sticking with what you've got.
Then you face having to make the same choice in 10 years but with worse options.
2Daniel Kokotajlo1y
Depending on AI timelines etc., this may be fine.
True. I just want to point out the irony on hoping for a providential all-powerful machine in order to avoid relying on a providential all-powerful human. What makes you think the AI saviour will be in more virtuous hands?
2Daniel Kokotajlo1y
I don't think that.

As is often the case in political writing, the diagnosis of the problem seemed far more plausible than the proposed solution.

There's just something weird about the humungous proportion of career bureaucrats and governmental employees vs. the few people who get elected. E.g. according to this, in the US there are ~9 million federal employees plus ~16 million people employed by local and state governments. Contrast that with <1000 (?) federally elected politicians, and you get a proportion of >9000 unelected employees per elected politician, which soun... (read more)

English pedant note: it should be either "how ... look" or "what ... look like", but never "how ... look like".

Someone's been reading Yarvin. 

He explicitly cites him at the end!

Just scanned through the post.  FYI, I broadly disagree with this and a few other takes I read from Dominic Cummings when it comes to enforcing ambitious political changes based on strategies he presents as the rational choice viz a viz predicted scenarios. But don't want to get into detail here.

Highlighting this point from Dominic's summary:

The Valley is the natural place to build the best model of the electorate and some weird subculture there is more likely than DC to look at the problem with the fresh eyes it needs. It’s also the natural place to

... (read more)
I don't think Cummings believes what the article calls technocapitalism. It defines as a fundamental aspect: If you read Cummings recruiting call for wierdo's he calls for creating significant technical innovation within government with new technologies like seeing rooms which don't seem to be developed by private for-profit enterprise. The general idea is that the government used to be able to do technological innovation and that a government that controls the government could again get the government to do technological innovation that's currently not done. When Cummings speaks about "startups replacing parts of the US government" he's not talking about privately funded and owned companies replacing parts of the US government but about creating new governmental organizations. Cummings says [] very explicitely:
Ah, I was vaguely under the impression indeed that Dominic’s article suggested for-profit tech start-ups would replace government departments. So thanks for correcting that impression. To be clear though, that’s not core to what I’m disagreeing about here. My disagreement here is not about for-profit entities coming in and running operations more efficiently, it’s about insular monopolistic actors privileging their technical expertise for deciding how to built the systems that will hold sway over the rest of society, while attempting little real dialogue with the diverse persons whose lives they will impact. IMO Taiwan’s g0v collaborative is an impressive example of technical innovation for amongst others resolving disinformation and tensions between the polar sides of technocratic/technocapatalistic top-down enforcement and populist bottom-up destabilisation (tricky not to get stuck in political idealogy when discussing those issues). Just read these articles on g0v and vTaiwan. They’re kinda idealistically promotional, but still highly recommend them: * [] * [] As a case in point, I would try comparing * Cumming’s conception of a seeing room, which from a casual glance seems to be about providing key government decision-makers with a more accurate and more processable overview for making a big decision. * with the way vTaiwan activists compile digestable information from responses to ‘rolling surveys’ amongst stakeholders and relevant experts, before inviting them to deliberate and build a rough consensus on stances online.
Cummings describes how their focus group's found that people in them were talking about wanting an immigration system like Australia and then they added that wish into their platform. Focus groups usually don't give you very detailed policies but part of the plan with them is to do create a platform that actually includes all those things that 70-80% of the population want and that currently gets ignored. Audrey Tang was a hacker from Silicon Valley who wanted to do innovation in politics. That's not possible in the US or UK right now but is in Taiwan, so they went back to Taiwan to do it there where they created what would be a startup in Cummings sense of the word. Audrey Tang is completely the kind of person Cummings wanted to hire in his call for weirdo's. I don't think there's any good basis for suggesting that Cummings wouldn't want systems to listen to diverse experts. When Cummings worked in the Department of Education, that department was so disfunctional as he describes that it couldn't repair the lift in their building. There are a lot of decisions involved in making such a system more functional that are not about building some policy consensus. Making good organizations about organizing large government departments is important. Yes, people like you or Glen Weyl let political idealogy cloud their ability to see clearly and that leads to errors just as thinking that what Cummings means when he says startup is something private. Cummings explicitly tells everyone that they have to watch Brad Victors videos and Dynamicland. Brad Victor setup Dynamicland in a way that's very intentional about not building technology average early tech adopter but for a wide variety of diverse people. The main political fight is whether you want an enviroment where radically new departments (and that includes things like the one that Audrey Tang runs) are possible or not and not about the individual choices of technology.
Started watching this talk by Bret Victor []on representing code for humans. Interesting, thanks for the share
I would be careful here about ascribing some singular definitive personal motivation to why I’m sharing these opinions (in the vain of ‘he and that other guy he talked with are clouded by ideology and that’s why he jumped to this factually incorrect conclusion’ or ‘he didn’t announce the names of the authors of a draft so based on that one can conclude this person doesn’t prioritise transparency’). Particularly when you quickly spot something about my take to disagree with, and might only grasp a small portion of where I’m coming from. Better to first have an actual face-to-face conversation and listen, paraphrase, and check in on each other’s views along with the context needed to interpret them. I’m deliberately not characterising you here based on your comments. I try somewhat awkwardly to stay open to what I’m missing. That having said, you’ve clearly read a lot more about Dominic Cummings' work than I have. I appreciate the detailed remarks. They help me break up and reassemble the broad impressions I personally got from reading a few blogposts. On getting input from focus groups from 70-80% of the population – is the focus here on soliciting and addressing commonly held or majority views that are ignored, or also on aggregating distinct minority views? This sounds cool. Let me watch a video of his. On setting up organisations that can manage themselves, I agree that this seems a major problem in the US government for instance (and also in e.g. the Dutch Ministry of Justice and Security, and Ministry of Foreign Affairs from the stories I’ve heard of gross misspending and budgets reallocated to political ends from insiders involved there). This EconTalk podcast [] suggested actually recruiting an experienced chief operating officer in the US government, or at least to oversee an effective central auditing department. I don’t have strong opinions about that one – only that the incent

New to LessWrong?