This may sound like a question that is very big and impossible to properly answer in a LessWrong comment. And I think that's true. But still it seems like an important question to learn about, for many different purposes (I was thinking about this in the context of AI, but I assume it is also useful for other contexts, e.g. politics or entrepreneurship).

Abstractly the answer is something like "provide goods and services for people". But what concretely does this involve?

For instance one type of good that is provided is food, and a core piece of the manufactoring chain for food is farming. This is in principle like in children's stories. With farming, you have to put seeds (and fertilizer etc.) into soil, wait for a while (sometimes watering it, especially if you live in a dry region) for it to grow into edible plants, and then take those plants out of the soil again.

But of course people don't manually do this anymore, but instead use machines. And I've been told that they don't even manually drive the machines, but that they are instead automatically run. So I guess farming work really involves setting up and maintaining the machines? Which presumably involves all sorts of tasks related to the problems that can arise in the farming machines, which I don't know much about. So for example one valid answer to this question would be a description of some mechanical problems that can happen with farming machines and how those mechanical problems get fixed. Other valid answers for this part of the economy would include things about how the machines get produced, and how farmers sell the food they produce.

Speaking of selling, a huge part of the economy is coordination work. Sales, lawyers, managers, etc.. But even "coordination work" is fairly abstract. What exact tasks might it involve?

So yeah, I would be curious how much knowledge of what the economy is doing that we can collect.

New Answer
New Comment

2 Answers sorted by



But even "coordination work" is fairly abstract.

A slightly less abstract summary is "assigning scare resources".

Imagine that you are a world dictator, and to keep things simple, you want to build an agrarian utopia. However, instead of Pol Pot, you are actually not an evil guy, and you genuinely want people to live happily and have enough food.

Okay, so you have a certain budget each year. Or, to remove the complications of using money, you have a certain amount of resources (including man-days) that you can spend during that year. How much will you (make your subjects) spend, assuming that they are 100% obedient, so no need to actually spend most money on your personal security...

  • literally digging in the garden (manually, or using a machine)
  • drilling and refining oil for the machine
  • mining the iron for the machine
  • researching a better type of machine
  • transporting food from place A to place B
  • oil / iron / research for the transporting machines
  • or maybe you should instead resettle people from place B to place A
  • building the new houses gets complicated quickly. And the more you think about, the more complicated it gets. For example, you still need some form of education, even if it is "narrowly" focused on agricultural work, iron machine design, construction, etc.

One part of the "economy" is that these things often just happen, by random people finding their personal motivation (often profit) to research them, do them, and maybe suffer a personal loss if their guess was wrong.

IDK if it's historically true, but I've read that the origin of coinage was to make it possible for governments to automate the provisioning of their armies. Instead of directly controlling it all, you mint coins, give them to soldiers, tell the soldiers what they're responsible for, and make a law that every citizen has to pay a certain amount of those coins in taxes each year. Directly or indirectly, everyone has to do something to benefit the soldiers in order to get those coins.

Indeed. And this is actually another angle through which I encountered the question in the OP. Sometimes the government tries to control the economy by changing the properties of money, e.g. changing the central bank rates in order to cause inflation. But I would like to get a better mechanistic understanding of which decisions shift in response to the interest rates. I think I kind of get the dynamics within finance, but finance is very abstracted from the object level.
Ok, then if we set finance aside, we're left with a set of resources that can be used for all sorts of purposes. Minerals, land, plants, animals, equipment, people, skills, knowledge, etc. Then some entity creates some form of money, and we all agree to use it as the unity of account and medium of exchange and store of value. That can happen in different ways, like the one I mentioned in the previous comment. At this point, people want money because money is the tool that lets them get other things that have value to them.  People start producing things of their choice and selling them for a price of their choice. By this process they learn, individually and collectively, what prices other people are willing to pay. Scale it up and we get market prices and efficient markets for goods and services. Everything has a price relative to everything else. That price changes over time for all sorts of reasons. The government can manipulate the overall level of prices, aka inflation and deflation, by altering the amount of money in circulation: unless we increase our productive capacity or decrease consumption, more money chasing the same goods and services means prices go up. It can also change relative prices through selective taxes and subsidies. Or by purchasing stuff, which affects prices the way private actors do, by raising demand for it. If things I want to buy or normally buy get more expensive I either change what I buy, decrease total consumption, draw down savings, go into debt, or find a way to increase my income (aka start producing more or better value for sale in the labor market). Everyone else doing the same gradually shifts the patterns of consumption and production. Hopefully in ways that increase long-term productive capacity. Repeat forever. The whole edifice of finance and business and the relevant branches of law are built to facilitate this. Interest rates are kind of a weird tool to use to affect the money supply, because it’s hard to know at a
That's the part of the story that I feel I get well enough. However this story abstracts over what things you would buy, what you'd use those things for, etc.. It's essentially still a story of finance, but with a different focus. What I'm wondering about is the dual side of the picture. Like for instance you mention the builders deciding to build fewer houses as a result of the interest rates. Is that because construction workers are sitting around looking at interests rates changes/housing price changes and deciding based on that? I don't think so, because I don't think the construction workers have the capital to build houses. I think there's probably a bunch of channels in the economy where the information trickles through, e.g. maybe someone in a company that needs stuff built so they can rent it out asks the bank for loans and then finds that the loans have become more expensive. However I don't feel I have a conprehensive picture of what the exact interactions would look like.
I'm not sure a single human brain can hold that comprehensive of a picture at once? If it could we wouldn't need all those channels. The whole point of markets is that they make that process work mostly invisibly to each of us. For the house example, if I'm a buyer my budget usually isn't something like "$500,000." It's more like "$100,000 plus $2500/month." Depending on loan terms that could mean a $600k house or a $300k house. The seller determines the listing price, but the buyer decides whether to bid. So a general change in loan terms across the population can mean I bid on the same house at up to $300k or up to $600k. Which, in turn, means that as interest rates go up, sellers drop the price to get a sale at all, or get a sale in a reasonable timeframe. Even if I'm a builder and nothing affects the builders costs at all and the production cost stays fixed at $200k, I'm going to be a lot more aggressive about hiring more crews and buying supplies and equipment if I can net $400k per house rather than $100k per house. It's less risky: If the successful projects have a 200% ROI instead of 50% ROI, I can afford to borrow to finance up-front expenses for more projects at once (relative to my existing assets) without worrying about paying them back even when some of my projects fall through. In practice, it's a bigger effect than that, because lower borrowing costs for homeowners tend to also correlate with lower borrowing cost for builders. Also because the same effects ripple through from the earlier parts of the value chain, affecting borrowing, hiring, and scale-up costs in mining, forestry, metallurgy, factories making machines and tools, factories making parts and semi-finished materials, all of it. Alternatively, or concurrently, if people can afford to pay more because of loan terms, a builder may decide to build bigger and more luxurious houses if that gets them better returns. Plus, existing homeowners are more likely to renovate and hire contractors when
Probably not, but if I try anyway I might still get a better understanding of some things.

(This is a reasonable comment and an appropriate answer to the question in the OP. I happen to already know this, so it's not so interesting to me, but it might be interesting to other people. I am more interested in, what are the concrete types of tasks, decisions, constraints, environments, interactions, etc. that happen in the economy. Which can involve coordination tasks, but not just coordination tasks. That said, I don't mind this big picture/theoretical comment. Just giving hints for what I am personally looking for.)



I'm going to give one of the easy answers that probably a substantial amount of the respondents here can give: programming. Partly the point is to illustrate what sort of answer I am looking for.

One of the main tasks a programmer handles is fixing bugs. Often bugs are discovered because the user has encountered a situation where the program does something undesirable (for instance crashing or giving the wrong output). Usually this just annoys the user and nothing more happens, but sometimes the user reports the error and what they did before it happened to customer support, and then customer support enters it into a list of tasks that the programmers can look at.

The programmers may estimate the priority of various bugs that are reported. This priority can be estimated by various things, such as how commonly users report the bug (or statistically how common it is in logs emitted by the program), how severe it is described in affecting the usability of the software, and how much programmer time it is expected to take to solve it.

The bugs (and other tasks such as features) etc. are then assigned to programmers in the team based on things like familiarity with the parts of the software (often programmers mainly work with only one part of the programs they make), and the programmers work to fix them.

Typically the first step in fixing a bug is to reproduce it, so we know how to tell that it has been correctly fixed, and so we can inspect the program while it runs to figure out why the bug happens. The programmer may try to follow the steps that the user described in order to reproduce the bug, or they may use their knowledge of how the program works to infer different ways of reproducing the bug. If they cannot reproduce the bug, they may decide not to fix it, or send questions back to the customer to get more info on how to reproduce it.

In order to understand how the bug happens, they may read the code to see if they can deduce the rules that caused the bug. They may also run the program in a step-by-step manner, seeing how each variable affects each other. At times this can be difficult, for <complex reasons that I'd ideally explain but I'm on my phone right now so cutting it out for brevity>.

Once the bug has been understood, the programmer has to come up with a fix. Sometimes fixes a simple, e.g. changing the code to eliminate a silly programmer mistake. Other times the bug originates inevitably from other logic, and one has to break that other logic to fix it. The bug can also occur due to missing code, e.g. maybe only a special case was handled in the past, but now a more general case should be handled.

The fix must then be written into code, which involves a bunch of complex problem solving that I'm not even sure I know how to describe in non-technical terms. It's pretty important though so I should probably return to it after I'm off my phone. (Or if you feel like describing it, dear reader, I would encourage you to do so in a comment.)

Often programmers will also write a piece of code that can test the fix. This means that over time, the project can end up accumulating enormous numbers of automatic tests. And that's the next step: typically after a change is made, all the tests are run to ensure that nothing breaks.

Before the code is made a permanent part of the project, typically other programmers on the team will look at it and give comments on how to make it easier for them to understand, how to make it faster, and so on, such that the final code is reasonably good.

While doing all of these things, programmers typically have to manage a lot of technical things. For example, if the programmer is working on a web app, then the web app is typically run by multiple programs that work together. In order to run the app to e.g. reproduce a bug, the programmer may therefore need to configure the different pieces so they know how to find each other. Often configuration is partly handled by automatic tools, but these tools are often buggy and the programmer has to know how to navigate around those bugs.

This isn't a complete description of what programmers do, rather it is a description of one slice of programmer work (solving bugs).

I'm not sure I understand how this relates to your original question. Everyone's job exists at some level because they have a comparative advantage doing it, and specializing in it, compared to other people doing it for themselves. Even doctors have doctors and lawyers have lawyers. 

I'm a consultant with a background in physics and materials science. Companies set themselves goals: growth, emissions reduction, and improving sustainability are the ones I focus on most. They work with my employer so a neutral third party can explain to them the landscap... (read more)

Programmers are part of the economy and this was an explanation of some things programmers do, so by transitivity it is a partial explanation of what the economy does? My original question was not "why do economic transactions exist, rather than everyone just doing what they personally need?". It was "what does the economy do?", i.e. which activities are executed within the economy?
Then in that case, the focus on job activities is way too narrow. The answer to "which activities are executed within the economy?" is roughly "all of them, except those things which literally can't be bought and sold, even on the black market." And that is a very small and shrinking set. Every choice to do something for myself is also a choice not to hire someone else to do it, there's no neutral opt-out decision, even if measures like GDP don't accurately capture household production. Technically with modern medical technology I could even use money to outsource my breathing, eating, blood circulation, and blood filtration. Economics is a totalizing approach to explaining how we assign and distribute scarce resources across every choice we make. So in that sense I'd say that "what the economy does" is provide options and incentivize people to take those options, in full generality.
Yes but I would like to have a conprehensive list of all activities people engage in. Rather than just abstracting it into "all of them".
I'm starting to think I'm talking to an AI and not a human.
Sorry if that came off rude, it just seemed like a kind of question I would never expect anyone to ask. It's not like there's one ready-made somewhere, and it's not like anyone could type one out. No matter how many examples anyone gives you, you'll still be missing almost all of them and need the abstractions anyway. And if you have internet access and have lived a reasonable number of years as a human, you already have a sufficient basis to know that, and also to build on without such a list.
1Thomas Sepulchre
Wait, wouldn't such a list be extremely long?
In theory yes, but I don't expect anyone to be able to create a full list, so I expect to only end up with a few items on it.
2 comments, sorted by Click to highlight new comments since:

But of course people don't manually do this anymore, but instead use machines. And I've been told that they don't even manually drive the machines, but that they are instead automatically run.


I don't think either half of this sentence is true.  A lot of crops are gathered by hand, either:

  • Because there is no technological alternative (e.g. strawberries, safran), or some quality standards require hand-picking (e.g. grapes)
  • Because, despite there being some mechanical alternative, not every country in the world uses this level of automation as a standard

As for "machines drive themselves nowaday", even if by "people" you mean "US farmers", I don't think this is true. It is true that some such machines exist, especially for the farming of wheat and corn, but they aren't as widely used (yet) as you suggest.

As a side note, putting "of course" at the beginning of a statement doesn't make it true or stronger, although I don't believe in any ill intent on your part.

Because there is no technological alternative (e.g. strawberries, safran), or some quality standards require hand-picking (e.g. grapes)

Oops I specifically had wheat in mind when writing the stuff about farming. You are right that there's lots of stuff this doesn't hold form.

As for "machines drive themselves nowaday", even if by "people" you mean "US farmers", I don't think this is true. It is true that some such machines exist, especially for the farming of wheat and corn, but they aren't as widely used (yet) as you suggest.

I see, thanks.