Socialism in large organizations
I’m a programmer who’s into startups. For my first startup, a site that provided student super in depth student reviews of colleges, I remember asking what people thought. I’d get all of these really encouraging responses. “Oh, that’s so cool! I wish that existed when I was applying! That’s gonna be so helpful to prospective students!”
Then for my second startup, I had similar experiences. I built an app that helps people study poker and received lots of great feedback. But for both startups, when it actually came time to sign up: crickets. When it actually came time to fork over some money: crickets.
The lesson? Talk is cheap. Actions speak louder than words. It’s all about the Benjamins. That sort of stuff.
Now I work as a programmer in a large organization. Things are very different.
My team builds a tool that is used by many other teams in the organization. We don’t, in my opinion, do a very good job of, before building a feature, checking to see if it truly addresses a pain point, and if so, how large that pain point is.
As an entrepreneur, if you do a poor job of this, you fail. You don’t make money. You don’t eat. Survival of the fittest. But in a large organization? It doesn’t really matter. You still get your paycheck at the end of the day. There’s no feedback mechanism. Well, I guess there is some feedback mechanism, but it’s not nearly as strong as the feedback mechanism of having users voluntarily open up their own wallets and handing you money for the thing you’re providing to them.
It reminds me of socialism. In socialism, from what I understand, there is a centrally planned economy. Some pointy-haired bosses decide that this group of people will produce this widget, and that group of people will produce that widget. If you do a clearly horrible job of producing the widget, you’ll get fired. If you do a clearly incredible job, and respect the chain of command, you’ll get promoted. But almost always, you’ll just walk away with your paycheck. It doesn’t matter too much how good of a job you do.
Well, sometimes it does. Sometimes there’s more elasticity with who gets fired and who gets promoted. But even in those scenarios, it’s highly based on KPIs that are legible.
I’ve been watching The Wire recently, and the show revolves around this a lot. Teachers having their students memorize things for the test. Police officers prioritizing quantity over quality in their arrests. Lawyers not taking cases that’d hurt their win percentage. Politicians wanting to do big and flashy things like building stadiums that look good to voters. But, of course, things that are legible are often not a very good proxy for things that matter.
I’m just babbling here. These aren’t refined thoughts. I’m not someone who knows much about management, organizations, politics or socialism. But I wonder: is this a useful frame? Is it one that people already look at things through? How common is it? Does it point towards any solutions? What if large organizations had some sort of token economy where teams had a limited budget of tokens and used them to get access to various internal tools? What if they just used real money?
This problem has been studied extensively by economists within the field of organizational economics, and is called the principal-agent problem (Jensen and Meckling, 1976). In a principal-agent problem a principal (e.g. firm) hires an agent to perform some task. Both the principal and the agent are assumed to be rational expected utility maximisers, but the utility function of the agent and that of the principal are not necessarily aligned, and there is an asymmetry in the information available to each party. This situation can lead the agent into taking actions that are not in the principal’s interests.
As the OP suggests, incentive schemes, such as performance-conditional payments, can be used to bring the interests of the principal and the agent into alignment, as can reducing information asymmetry by introducing regulations enforcing transparency.
The principal-agent problem has also been discussed in the context of AI alignment. I have recently written a working-paper on principal-agent problems that occur with AI agents instantiated using large-language models; see arXiv:2307.11137.
References
Jensen and W. H. Meckling, “Theory of the firm: Managerial behavior, agency costs and ownership
structure,” Journal of Financial Economics 3 no. 4, (1976) 305–360
S. Phelps and R. Ranson, “Of Models and Tin Men—a behavioural economics study of principal-agent problems in AI alignment using large-language models”, arXiv:2307.11137, 2023.
That makes sense. I was trying to point at something more specific than the principal-agent problem though. Ie. I agree that it fits under the umbrella of principal-agent problem, but it feels to me like it’s a specific type of principal-agent problem. I’m not sure how to describe it well though. Something about large organizations with an emphasis on large businesses.
Perhaps ‘The Theory of the Firm’? The very existence of large companies is a puzzle if you are a naive believer in the power of free markets, because if the market is efficient then individuals can simply contract with other individuals through the market to achieve their desired inputs and outputs and there is no economic advantage to amassing individuals into higher-level entities called companies. The reason this doesn’t this work is because of transaction costs. An example transaction cost could be the time invested in finding partners to make contracts with for all your inputs and then forming and enforcing contracts. In the ‘the theory of firm’ the insight is that individuals can lower their transaction costs by forming a higher-level economic agent- ie by becoming employees at a larger company. Transaction costs explain why large firms exist, though they don’t eliminate principal-agent problems and may exacerbate them.
As an aside, similar considerations of transaction costs may explain the evolution of multi-cellular life. In evolutionary biology lower level competing units of selection cooperate together to form higher-level entities- genes/genomes, cells/organisms, organisms/groups, groups/societies, resulting in major transitions in evolution. The endosymbiosis between bacteria that led to the evolution of Eukaryotic Cells can be thought of as analogous to forming a company in order to reduce transaction costs. This is discussed further in S. Phelps and Y. I. Russell. Economic drivers of biological complexity. Adaptive Behavior, 23(5):315-326, 2015. [PDF]
Hm. Interesting. I’m confused though. If there were no companies and it was just individuals contracting with one another, I feel like the big issue would be that one person can only do so much. You can’t build cars or apartment buildings with one person. Transaction costs seem trivial in comparison to that.
The point is that all the people making cars in the company could, in principle, do the same job as self employed freelance contractors rather than as employees. Instead of a company you have lots of contracts between eg assembly line workers, salespeople and end customers, without any companies. The same number of cars could in theory be built by the same number of people in each case. The physical scenario would be identical in each case. The machinery would be identical in each case. But in the freelancer case you still have lots of people building cars but there is no invisible company to coordinate this activity instead you are relying the market.
Oh yeah, I see. Very interesting. Do any other transaction costs come to mind?
The main problems are the number of contracts and the relationship management problem. Once upon a time drawing up and enforcing the required number of contracts would have been prohibitevly expensive in terms of fees for lawyers. In the modern era Web 3.0 promised smart contracts to solve this kind of problem. But smart contracts don’t solve the problem of incomplete contracts https://en.m.wikipedia.org/wiki/Incomplete_contracts, and this in itself can be seen as a transaction cost in the form of a risk premium. and so we are stuck with companies. In the theory of the firm companies are indeed a bit like socialist enclaves. Individuals give up some of their autonomy and agree not to compete with fellow employees in order to reduce their transaction costs. As you point out the flip side of this is that it can create new principal agent problems, but these are rarely insurmountable. The principal agent problems that we should really worry about are the ones that occur between companies, particularly in financial services. It was a principal agent conflict between rating agencies and investment banks that led to the great financial crisis, as dramatised in the film the Big Short.
Sears tried creating an explicit internal economy. It did not end well. https://www.versobooks.com/blogs/news/4385-failing-to-plan-how-ayn-rand-destroyed-sears
You aren’t the first person to notice this, no.
See “The People’s Republic of Walmart: How the World’s Biggest Corporations are Laying the Foundation for Socialism” by Leigh Phillips and Michal Rozworski.
While the small business sector is an important source of economic dynamism, the success of the large corporation so constructed probably bears some thought.
You could look into what happened with Sears when the CEO, inspired by some of the thinking you’ve laid out here, tried a decentralized model: https://tinyurl.com/sears-collapse-harvard
Oh, very interesting. I made a note to look into these more in the future but in the meantime, do you have any comments on it?
Good point. I agree. Although I do think it is quite plausible that they are successful despite the poor structure rather than then because of it.
In particular, I’ve always suspected that asymmetric information explains a lot of their success: as a consumer, there’s a ton of options, it’s way too time consuming and frustrating to try to research stuff, so you usually just go with the safe big name option that you see other people using.
A few random points:
Large organizations necessarily have slack, or else it turns into an immoral maze hellscape. This correlates only weakly with the political system (I happened to work in both).
Incentive misalignment is common but generally non-fatal. Proxies becoming optimization targets are fine as long as the optimization pressure is low or moderate.
People generally want to do a good job unless heavily incentivized against, punished or discouraged.
A tight and accurate feedback loop is definitely important, as well as the ability to A/B...Z test. In the case of a startup that may include taking deposits before releasing a product, so that you don’t get blindsided by “crickets”.
I think that’s an incomplete lesson about startups. Yes, talk is cheap, but also you probably misdiagnosed the hard part, and your description to enthusiastic users didn’t match your delivered product. Both reviewing colleges and studying poker are relatively simple coding projects and quite difficult content problems. How well an app solves those problems is very little about UI or algorithms, it’s about generation, curation, and transformation-for-specific-user-needs of information. I can’t tell from your description how much of the real problems you were able to address before becoming discouraged at lack of use.
I work as a very senior engineer (not so much a programmer anymore), formerly at very large companies, currently at a medium-sized company (~130 engineers and managers), and I will say that the binary feedback (success/failure) is far less immediate, but the actual usable feedback (who is doing what, what features are being asked for, and overall whether to invest more or less in the team) is there, as long as anyone cares to look (and either I’ve been very lucky, or it’s rather common for many to care).
It’s unclear what version of socialism you’re talking about, so I don’t have a lot to say on that topic. I think there’s a lot of modeling value in exploring the nuances of legible vs il- or semi-legible, and who gets to pick the KPIs and weighting among them. My base opinion is that the universe is it’s own best model, too finely-detailed to summarize perfectly inside a small segment of itself (a brain or computer cluster). So “legible” is ALWAYS about abstractions and models, which means the hard part is realizing that it’s going to be wrong in many cases.
This is the thesis of Charles Koch’s book on scientific management: if markets are so great why not allow them inside the walls if the company? Imo, what he does, buy low margin businesses and improve them, is much more impressive than what Buffett does, buying good margin businesses and leaving them be.
Very interesting. I created a todo item in my personal backlog to read about it more, but in the meantime, do you have any thoughts or comments on what that book says?
He got it from his father who saw the strong contrast between USSR and American ways of doing things.
Wikipedia: “extended litigation blocked Winkler-Koch from selling the technology in the U.S. for several years.[12] In the words of Jane Mayer, “Unable to succeed at home, Koch found work in the Soviet Union.”[15] Between 1929 and 1932 Winkler-Koch supported the Kremlin and “trained Bolshevik engineers to help Stalin’s regime set up fifteen modern oil refineries”[attribution needed] in the Soviet Union during its first five-year plan.[16][17][18] According to Mayer, “Over time … Stalin brutally purged several of Koch’s Soviet colleagues. Koch was deeply affected by the experience, and regretted his collaboration.”
For the rest of his life he would push against centralizing influences in decision processes. This often took the form of encouraging competition, skin in the game, and profit sharing for those who did well. Charles continued developing this playbook.
For within-firm “token economy”, my comment here seems maybe relevant.