Epistemic status: Opinion and speculation based on community organizing I did between 2015 and 2018.

In this essay I’m going to suggest that the Keynesian/Hayekian dispute in economics is an interesting model of community organizing. This is a toy community-building model built on a toy economics model. I am not an economist. But I have listened to many hundreds of hours of econtalk!

Okay, cool, you don’t know what you’re talking about, but please go ahead and mansplain this complex eighty-year-old dispute to me

Sure, no problemo. Have a seat, Robin Hanson.

The Model

JM Keynes made his reputation in the 1930s, partly by thinking about the causes and solutions to the great depression. He argued that we should use government spending to stimulate the economy during a downturn[1]. This is a top-down approach that is optimistic about the possibility of moderating economic cycles of boom and bust.

FA Hayek was a contemporary of Keynes. Hayek had a different idea: it's government policy that distorts markets which leads to price distortions. Price distortions lead to misallocation of assets. Misallocation leads to unsustainable bubbles. Bubbles lead to busts. So, government spending to deal with a bubble bursting just plants the seeds for the next bust. Hayek is making an economic argument here, but he’s also making a cool argument about epistemology. Hayek was concerned that civil society contains rich, grounded knowledge of local (market) conditions. A centralized top-down bureaucracy will never be able to fully  comprehend that contextual knowledge. Hayek pithy compressed this critique into the following bon mot:

The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design [2]

So, Hayek’s analysis is more bottom-up and pessimistic (“epistemologically modest”) about governments influencing the economy[3]. I’ve found this tension between top-down and bottom-up approaches in economics to be a productive toy model for thinking about similar tensions in community organizing.

To some extent, I’m using Keynes and Hayek as metonymies for top-down and bottom-up organizing. Much of this essay could have been written without mentioning either figure. However, the analogy is productive because it will let me import some of their ideas in a couple of other places [4].


I started thinking about this model when a co-organizer in the EA Group I was running was enthusiastic about running surveys of our community members. And I was… not. I had several concerns, including some obvious ones:

  • Our community was fairly small. I expected we would just be sampling noise.
  • Might this drive some people away? People like to do Fun Stuff in their free time. For a certain type of perverted mind, all this EA bullet-biting, metacognition, longtermism and [meta]ethics is fun. These perverts may or may not also find filling out surveys fun, but I didn’t want to chance it [5].
  • I worried about opportunity costs for my co-organizer and community members. Even if our members enjoyed surveys, and there was a signal we could detect, was it worth the time and attention cost?
  • What would we do with the data? My conception of community organizing was very laid back (one could say laissez-faire): I maintained a regular, predictable monthly schedule of meetups. The same quiet room, in the same dive bar, on the same night of the month. I bought nachos and tried to be hands-off, welcoming and value-aligned. What, if any, metrics we should be trying to optimize for? I was trying to nurture a vibe and a mindset.

I’m actually less confident about my minimalist, bottom-up approach now, given the current state of EA than I was at the time. A top-down approach may be better suited if you’re on a more project-focused path, and it seems there are a lot more EA projects. Also, maybe my co-organizer had ideas I just didn’t get.


  • There was also an aesthetic preference. Aesthetically, I like "stewardship" as a template for community organizers. I think the aesthetic of stewardship is that stewards should not draw attention to themselves, as organizers, unless there is a significant benefit to the community in doing so.

On this stewardship issue, I’ll build a wispy bridge back to Hayek.

Just as there’s pressure on governments to print cheques with the names of the current leaders on them [6], there’s a tendency for organizers to do things so that they can have been seen doing a thing, to reap some prestige.

A Hayekian framing here might say that organizers are in a position to impose taxes of time, or attention on a community, and to print the currency of prestige for themselves. This is the closest (Tortured? Irritated?) connection I can make to Hayek’s concerns about top-down structures distorting prices and markets.

Of course, a Keynesian would respond: yep, and public goods are a thing, and we want more public goods, and organizers who step up and help provision them should be paid in prestige. As well as in actual specie.

Additional implications

Group Dynamics

One area where top-down structures might work particularly well is in transmitting EA values. We don’t want people becoming confused about what EA is, e.g. student groups that evolve towards doing trash pickup days while believing they’re doing impactful EA work.

A more ambiguous area is the desire to make EA events more welcoming. Eg, EAG has a Code of Conduct. Intuitively, it seems reasonable that a Code of Conduct might help make events more welcoming[7]. But should we take a next step and assign volunteers at EAG to actively nudge spaces and events towards being more welcoming and accessible? I don’t know! Nudging meetups towards being more welcome was one of my goals when I was doing community work. I think I sometimes succeeded at inclusive nudging. I’d like to hope it was net positive (although I now realize that I have no real idea what, if any effect my nudges might have had).

Having organizers nudge events towards being more welcoming seems very Keynesian. Nudging was also one of the explicit goals for volunteer Room Managers at EAG London. For example, it was suggested that if we felt someone was monopolizing a conversation, or talking over folks a lot, that we might encourage more equitable speaking time. I think I would have entirely endorsed equity nudges as a goal a priori. In practice: as a room manager, I felt as useful as snake suspenders! There wasn't much to do! I got aggravated during my second shift! This was super precious time! The opportunity cost was insane! I couldn’t sense any need to manage or nudge the discussions! Nothing was going wrong, or needed my attention! The rooms were too large and I could only follow a tiny fraction of what was happening! I was missing sessions! God damn you John Maynard Keynes you sexy English Bastard [8]!

I ended up making an executive decision during my second shift to abandon my post for a time and pop into another session. To be fair: the EAG organizers were extremely vocal that people should speak up if they were losing value due to shift scheduling. I also hadn’t expected to get much out of attending talks. However, in practice, the talk I popped into was great. And, I don’t know, I’m pretty agreeable, and not very inclined to rock boats. I wanted to be a team player etc. All of which made me reluctant to complain about the volunteer schedule at the time.

As I write this, I have this deep urge to add that I still think trying to manage conversations to increase equity might be a good idea. Maybe it sometimes is! But it definitely might sometimes be a distortionary misallocation of resources.

A Keynesian/Hayekian pipeline model

As part of my conclusion, I’ll assay this model I came up with for EA community building. Our pipeline is: filtering → educating → empowering.

Filtering: I think that EAs are mostly filtered then refined, not "converted". Some cluster of psychological traits make people more likely to be receptive to EA[9]. Filtering is about initially finding people who resonate with the EA pitch.

Educating: Once you have discovered someone who resonates sympathetically with the EA pitch, you want to expose them to good EA content, and ensure that they get the key points with high fidelity.  

Empowering: Doing whatever we can to help EAs gain influence over the future.

This model suggests to me that educating is the most naturally top-down stage. Filtering does not strike me as either inherently Keynesian or Hayekian. Empowering could involve either mode, but Hayek’s insight that outsiders will always have trouble understanding what’s going on seems important[10]. Indeed, we often can't even get introspective access to what our own true bottlenecks are. Sometimes we just need to flail around a bit and try stuff and hill climb up an agency gradient. In those situations, it might be hard to suggest any top-down strategy to help people, other than the most classically Keynesian one: giving people money.


I presented a sketchy-ass cartoon model of a postwar economics dispute that might be a productive metaphor for a common tension in community building.

I’m confident that there are true tensions and trade offs between top-down and bottom-up approaches to doing community organizing work, and that reflecting on them will be productive. As I’ve pointed at in several places, I’m less confident in my analogy to Keynes and Heyek’s dispute. However, I don’t think that analogy is very weight-bearing in this argument; I’ve found it a useful shorthand, and I think there are interesting connections to both men’s work, but it’s impressionistic at best.


I bashed this out quickly for the Community Builder Writing contest. Amber Dawn Ace provided really helpful comments and proofreading to an earlier draft. Akash provided helpful early encouragement and suggestions.


[1] The wikipedia for Keynes sums this up with “Keynes advocated the use of fiscal and monetary policies to mitigate the adverse effects of economic recessions and depressions”

[2] If you play the EconTalk drinking game you may now take a shot.

[3] I don’t know what Keynes thought about this argument. Please leave a comment if you do! (Please do not leave a comment to tell everyone if I’ve completely botched the economics here.)

[4] And it’s fun.

[5] I don’t know if FA Hayek was Too Austrian to write about Fun Stuff, but commerce, which he was pretty into, is often Fun. 

[6] Do not take this to mean I am against redistribution. BRING ON THE UBI

[7] Has this been studied at all? This is the first time I’ve actually thought about this question, and I find that I expect they probably don’t have any detectable effect. I’d give at least 35% Codes of Conduct have no detectable effect, and maybe 15% that they have a detectable negative effect.

[8] Look at him

J M Keynes smoldering

[9] The intuition here is similar to Hume’s notion behind the idea that “reason is ever a slave to passions”. Ie, there are some deep upstream priors, and we often confirmation bias our way up gradients that confirm them. I think this intuition partly explains phenomena like  mainstream ML researchers who maintain there’s no xrisk, can’t possibly be any xrisk from advanced AI. This might be what Ada-Maaria Hyvärinen was running into when she talked to ai safety interested ML researchers. Ada-Maaria asked some researchers what got them into AI safety. They all pointed at Superintelligence. My intuition is that these researchers had priors that Superintelligence resonated with, whereas Ada-Maaria found Superintelligence unconvincing and perhaps confusing, possibly because it clashed with some of her deep priors.

[10] E.g. see Scott Alexander’s Reverse all Advice and Agnes Callard’s Against Advice


New Comment