1. Mankind started trading in barter. Trade was simple so people could use word of mouth, their brain, or pen and paper, to find barter trade deals. However, as trading networks grew in size, human communication and brain power was too slow to scale.

  2. This lead to people adopting a common rare-enough item as an object of value (e.g. gold, fiat, etc) to simplify value calculations. This solved the scalability problem.

    In return it brought other problems: using the value of a common item for all trades will cause values of savings to inflate/deflate regardless of the work that was done to bring it. E.g. if I produced leather, and the worth of leather was increasing, but the worth of the common medium was dropping, then my savings' worth will drop even though they were by producing value-gaining leather.

    Simplifying trade was worth that unfairness. Because at least trade could happen in large trade networks.

  3. However, now we have computers and Internet. We can communicate semi-instantly at an international level, and we have enough computational power and advanced algorithms to solve the barter problems fast enough (e.g. approximate knapsack solvers can be fast enough).

    The question is: What if we announce our trades in a global transactions log that is somehow guaranteed to be accurate/correct? Will this resurrect the barter system as a superior alternative to gold/fiat? If "no", why not?

1. Background

1.1. Fiat notes

When I do some work for someone, and he/she pays me money, such as a 5 USD note, then the note is simply an indication of debt. I.e. an indication that I can get something worth 5 USD any time later. E.g. I may later on spend it to get a drink, food, etc. Before I spend that 5 USD note, it is just a debt (not of use) until I use it.

The reason this works is because people agree that a 5 USD note has certain value. So, while others do not know exactly what I did to the person which persuaded him/her to give me the 5 USD note, they know that the person perceived the work as worthy of 5 USD, and this is sufficient to people.

1.2. Transactions log

Imagine a transactions database, that specifies (in enough detail) that works done to different people.

For example, instead of stating that I received a 5 USD note from someone, it rather states the work that I did to that person, such as offering him two cups of coffee.

Then, instead of that person giving me a 5 USD note back, the person rather says something like "I acknowledge that these 2 cups of coffee are worth the broken window that I repaired for Alice".

This goes on recursively. E.g. Alice would have a transaction saying that "I acknowledge that the broken window that Bob repaired for me, is worth my 1 week's work for Charlie".

As we follow such recursions, we will eventually some base cases, or recursion bottoms, where we find things like "I farmed 1kg wheat and gave it to Charlie", and Charlie's confirmation "That is true; John did indeed give me 1kg of wheat".

1.2.1. Assumptions

Suppose that the transactions log is:

  1. Computationally scalable both in space and time. I.e. it is very convenient to search the database and to resolve all recursions in instant time, even with small mobile devices.
  2. The transactions log is reliable and that its entries are actually true. For example, if you find "I gave 1kg of wheat to Charlie", then this actually did happen.
  3. There is no double spending. For example, Alice cannot use the same 1 week of her work to get goods, or services, twice. If she uses a 1-week worth of her work to get her broken window fixed, then she cannot use that 1-week's worth of work again.

2. Questions

  1. Is this transactions log just a more detailed, or verbose, version of the fiat notes?
  2. If such transactions log exists, while honouring the assumption that it is fast/scalable and true, will it replace the fiat notes (i.e. become money)?

What I've done so far

A. Initial thoughts

The fiat notes, or even gold coins, simply abstract the work that was done by exchanging some common commodity (or notes).

The entries in the transactions log are just a more detailed version of the same: the actual works are specified, without being abstracted by their worth in some common unit (e.g. grams of gold, USDs, etc).

Technically, if one has such the transactions log, one can recursively navigate it to eventually link entries back to goods and services, including common ones (e.g. wheat, gold), which eventually leads to allowing trading parties to reach a mutual understanding (or agreement) about the worth a given transaction entry.

For example, as these transaction logs recursively link people to each other, we will find out things, such as: the coffee that I was making, was benefiting a doctor that benefited a carpenter that fixed the broken window of some engineers at AMD (a CPU maker), which benefits some fabric makers in China. This way, me making coffee in, say, the United Kingdom, can possibly persuade a fabrics supplier in China (that wants some AMD CPUs) to give me some of, say, 1 roll of his fabric, in return to me giving him ownership of 50 cups of coffee that I made.

With today's Internet and faster computers, it might be possible to solve such graph problems efficiently even with cheap mobile devices. Individual people may tune their local applications to value different works differently, based on their individual requirements, which is then used for trading parties to reach agreements, automatically as the application solves the graph problem.

I think, the reason humanity went to abstract trades by using gold or fiat is because they lacked the computational capacity that we have today. The Internet, and fast computers, is very recent innovations to our lives today. Not too long ago, e.g. 1980s, computers and internet were perhaps too slow for this.

However, now that we have much faster computers and Internet, I think there is a massive possibility to directly solve the graph problem, with all its details, as long as we have the assumptions in section 1.2.1 honoured.

B. After 1muflon1's comment

... Money in economics is rigorously defined as something that fulfills roles of: 1 medium of exchange, 2. unit of account, 3. store of value ...

I think, the transactions logs will have those properties if:

  • Ownership of transaction entries can be transferred from one to another. This can also be partially. E.g. I transfer 1% of the entry concerning a transaction to someone else, then only the receiver can spend that %1 (i.e. no double spending).

  • Entries have a limited supply. This has to be true by assumption (2) in section 1.2.1, which states that entries are true (i.e. entries represent actual work that was done).

  • Entries have a perceived value. This depends on people. E.g. the value of gold is not for its utility (e.g. a conductor of electricity) but for its perceived value by people (which is mostly arbitrary simply cause it shines and rare enough, which made people compete for it).

    Applying this to transactions log, I see no reason why entries must not have perceived values. I think these are the ways by which entries will gain a perceived value:

    • Some entries may have historical values. E.g. entries representing works done of historical value, such as things similar to demolishing the Berlin wall, or some coffee cup that was made to a figure that eventually become a historical figure.

      The popularity of such events may make them more demanded, for the same reason the original painting of, say, Mona Lisa, values a lot more than duplicate one even if the duplicate has better colours.

    • People who engage in trades long enough, they may discover pathways of transactions that eventually link people in different countries. E.g. Someone giving a coffee cup to another in UK, for free, could eventually result in lowering the expenses of some carpenter in China within a few months.

      People who will realise this delayed reward, will realise a practical value of transactions.

I explained above how I think the points above show that the transactions log (which satisfies assumptions in section 1.2.1) is usable as "money". But there is a practicality issue that calculating the value is hard. E.g. how many coffee making transactions are worth one iPhone?

I think this difficulty is why people, in the past, have chosen to limit themselves to a common resource, such as gold, fiat, bitcoin, etc.

However, nowadays, thanks to us having fast computers in our pockets (smartphones) and fast enough internet, I think it is very doable to let people negotiate the worth of, say, iPhone, in the unit of coffees and repaired windows, automatically within less than 1 second of time.

Different people can adopt different preferences in how they set up their algorithms for automatic negotiations in ways that their individual gain is maximised. E.g. if I have a surplus of coffee cup selling transaction entries, I may prefer to sell in them.

A side effect of using this transactions log, which I find useful, is that I find it offers a more fair response against cases of inflation or deflation. E.g. if the demand for coffee decreases, then, while this will lower the value of my past work (selling coffee cups), it will not harm the value of others' works who were doing other things, such as selling doors.

This side effect is more fair than using a common resource as money (such as gold, fiat, bitcoin, etc). Because, using this common resource will eventually harm (or benefit) every saving equally. Why should a coffee maker's past work gain (or lose) value equal to a door maker's past work? I think it is unfair to assume that they are equal. E.g. in luxury times, coffee demand falls (like any luxury), while demand for engineering increases (like arms manufacturing).

C. After Nobody's comments

C.1. No single-number unit of account

My thought: No proof is supplied for this claim so far. Should unit of account be a single-number? Why shouldn't the worth of a thing be relative to the trading parties? Personally, I think the worth of a thing is relative to trading parties.

C.2. If multi-dimensional number is used to measure money, then it is the knapsack problem, which is too expensive to solve computationally

My thought: Trading parties don't have to solve the knapsack problem. It suffices for them to approximate the solution in a computationally efficient manner. Surely an approximation will have an error (e.g. conducting trade by exchanging a sub-optimal set of transaction entries).

I propose this definition for trading error:

Definition. A trading error is when someone gets less with the same amount of balance when he could've otherwise gotten more.

E.g. Say I have balance of making 5 coffee cups and repairing 5 doors, with these payment options:

  • Option 1: I could get 1 kg of cheese and an iPhone if I give the coffee cups' entries to the cheese maker, and the door repairs' entries to the iPhone seller.
  • Option 2: I could get 2 kg of cheese and an iPhone if I give the door repairs' entries to the cheese maker, and the coffee cups' entries to the iPhone seller.

If I choose option 1, then my trading error is 1kg of cheese.

In other words: when a trading party $p_0$ used his balance $\mathcal{B}$ in order to buy goods $\mathcal{G}$ from other parties $p_1, p_2, \ldots$, when he could have gotten more goods $\mathcal{G}^+$ with $\mathcal{B}$, where $\mathcal{G} \subseteq \mathcal{G}^+$, then his trading error is the difference $\mathcal{E} = \mathcal{G} \cap \mathcal{G}^+$.

However, using a single-number (as done with today's common money) is also an estimation of some kind, albeit, so aggressively that only a single dimensional number is used to represent the worth of any trade. This also has an approximation error, and a question is, which error is lower? The error by approximating the knapsack problem above? Or the error by the aggressive approximation by today's money which gives a single number to describe the worth of everything?

If you're a carpenter, and people realise that the worth of your carpentry services are increasing, then you will be able to do more with the transaction entries where you sold your services. But, if you were paid in today's money (which is a single number representing the worth of everything), then the worth of your past services will not increase to reflect the fact that the worth of carpentry is increasing.

Ironically, it may even go down, simply because a currency is generally inflating for reasons unrelated to your carpentry work. This can be extremely unfair, and was justified only in a time when we lacked CPUs to approximate the knapsack problem.

So, a fundamental question is: which error is lower?

  1. The error by using an approximate knspsack solver against the transactions log in order to obtain a local representation of worth relatively to the trading parties?
  2. The error by using the aggressive uni-dimensional number as a global single-number to absolutely represent the worth of all works for all trading parties (the approximation used by today's money)?

Another fundamental question is: which trading error is more fair?

  1. An error against a trading party, that is due to decisions made by the same trading party?
  2. An error against a trading party, that is outside of the control of the trading party?

I personally think that the 1st (when using transaction logs) trading error is better to have than the 2nd (when using common money), because at least only the 1st you have the choice to reduce your error, while the 2nd does not give you such choice. This superior partitioning of responsibilities eventually will put pressure at the right place to reduce errors (e.g. individuals will seek better knapsack approximators).

  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$
    – 1muflon1
    Jul 20, 2022 at 9:41
  • $\begingroup$ Suppose I have a milk cow (or a ledger claim to a milk cow) to exchange, and I want a cup of coffee. Probably both the cafe and I would agree that the milk cow is worth considerably more than the cup of coffee, indeed it's worth many, many more cups of coffee than I want to have on ledger. But milk cows aren't easily divisible (you could cut them into parts; but then they aren't good for milk cows anymore). Most forms of money solve this small-change problem through divisibility, or perfect convertibility into smaller units. Does your ledger system solve it in the same way? Or a different way? $\endgroup$ Jul 21, 2022 at 15:58
  • $\begingroup$ @AlabamaScholiast - Same. Basically, you give the cow to some person $P$ ($P$ could be the cafe owner, or anyone else), and an entry $W$ in the transactions log will appear acknowledging your work that you have made a cow and given it to person $P$. You will be the owner of this work $W$ (i.e. making a cow and giving it to $P$). Then, you can transfer ownership of $W$ to other people, possibly by dividing it. E.g. you could transfer 0.001% of $W$ to the cafe (in return to get a cup of coffee). $\endgroup$
    – caveman
    Jul 21, 2022 at 17:26
  • $\begingroup$ << You give the cow to some person P ... and an entry W in the transactions log will appear acknowledging your work that you have made a cow and given it to person P ... E.g. you could transfer 0.001% of W to the cafe >> What units do you see W, and the partial slivers of W, as denominated in? Are these shares of past work (denominated in e.g. labor-hours), shares of ownership in the milk-cow (going forward), shares in the proceeds of the milk-cow (in milk, or sale price, or...), or in units of some (any) other good with a forward value to P (say, commodity coffee, or ramen noodles, or ...)? $\endgroup$ Jul 21, 2022 at 18:40
  • $\begingroup$ @AlabamaScholiast - Good questions. This is my question here. Basically I am asking if $W$ can become money. E.g. if people keep doing works, and we have a transactions database of works $W_1, W_2, \ldots, W_n$, then will these become money by themselves? I will write my thoughts so far in another comment, but this is basically my question. $\endgroup$
    – caveman
    Jul 22, 2022 at 15:45

3 Answers 3


There are some huge problems with your idea.

There is no surplus value in your scheme. If you trade Alice a couple of cups of coffee for fixing a broken window, then you and Alice are square. But now you want to have a transaction with me. What value am I getting in exchange for the service or goods I render to you? Presumably you've already consumed the coffee, so you have to find something else you can trade me for my service or goods. If so, why would I care that you and Alice had previously worked out a fair trade?

People do not want all of their entirely legal transactions known to the public at large. For example: "Received two hours of family therapy in exchange for 100 cups of coffee" or "Received treatment for STD in exchange for changing a flat tire".

Fiat is fungible, barter trades are not. A large market participates in establishing the exchange rates between fiat and goods and services. The rates might not always be "fair" in the abstract, but they do eliminate a massive amount of negotiation and bargaining. You might think that the two cups of coffee was a fair exchange for fixing the broken window, but what if I disagree? What if you and Alice are in cahoots, and there was no broken window or two cups of coffee, and you and Alice are just going to split the value of whatever you get from me? What if you want to trade me your working for the Nazi Party an hour for a cup of coffee? Do I have to agree?

How are you going to handle large purchases? If I want to buy a house for the equivalent of \$500,000USD do I have to compile a list of several thousand small transactions that I made that add up to the equivalent of $500,000?

  • $\begingroup$ I and Alice are square in that case. But, another option is: I may rather fix Alice's door without getting a coffee from her. Instead, getting a smart phone from some dealer in China. Why? It depends if they find value in it. Two ways they may find value in it: (1) They like it as a noble act, in case is some important figure to them. (2) They happen to contest on it as a rare resource in case it becomes an ancient transaction (for a similar reason people like collecting ancient artefacts despite being useless). In fact, same goes to fiat notes. $\endgroup$
    – caveman
    Jul 17, 2022 at 14:31
  • $\begingroup$ Other points: If people do not want to disclose the service they get or give, then they may choose to not do it. E.g. I may just say "I transfer ownership to the ancient ancient Alice coffee cups to Eve", without specifying a reason. Then Eve will own it and can use it to get, say, some SHA-512 ASIC. $\endgroup$
    – caveman
    Jul 17, 2022 at 14:34
  • $\begingroup$ Finally, about disagreement in price, or negotiations: this happens even with fiat/gold/etc. We still negotiate. Surely negotiating with a more detailed transactions log is more involving, but, at the same time, we can automate it by using algorithms to solve the shortest path to maximum profit for us. I think the reason we didn't have this in the past is because we didn't have computers nor Internet, so we had to use one common item as a transaction medium. $\endgroup$
    – caveman
    Jul 17, 2022 at 14:41
  • $\begingroup$ Forgot to comment on your (C) cahoot and (N) Nazi arguments: (C) This violates assumption (1) in section 1.2.1. (N) No, and this is nice side effect (feature). $\endgroup$
    – caveman
    Jul 17, 2022 at 14:43
  • 1
    $\begingroup$ "There is no surplus value in your scheme..." I don't think this is true. If Alice sells Bob two cups of coffee, and she doesn't have a window that needs fixing, then Bob gives her a voucher for a window fix. Alice could trade this to someone else for whatever goods or services she needs, and that someone could trade it onward, and so on. Eventually someone who needs a window fixed will come into possession of the voucher and redeem it with Bob for the actual work. There are problems with the scheme, but this isn't one of them. $\endgroup$
    – Nobody
    Jul 18, 2022 at 12:54

This is not money because there is no unit of account

Referring to the definition of money that people gave you in the comments, the biggest element lacking in this scheme is that there is no unit of account. In the real world, the contents of your bank account is a single number, while in this scheme it is a long list of credits for random goods and services. Just keeping track of how much you own would be a complicated undertaking. Can I make rent this month with 17 cups of coffee, 11 window fixes, and a haircut in my account? Beats me.

The lack of a unit of a unit of account creates other problems. Making change is one. Even if everybody agrees that 11 cups of coffee is worth 5 haircuts, if all my wealth is denominated in cups of coffee, then how do I pay for two haircuts? Prices also become a problem. When I walk into a store, what, exactly, is printed on the price tag for the items I wish to buy? How do I compare prices between two stores if one is quoting prices in cups of coffee and the other in haircuts? I would have to know the current conversion rate between coffee and haircuts, even if I am bald and don't drink coffee.

There are other problems with the scheme, such as the difficulty of defining exactly what is meant by "a cup of coffee" (fancy espresso drink, or drip?) or "a haircut" (expertly styled, or just run the clippers over it?). However, even if you assume a solution to those kinds of problems, the fact remains that money has to be a yardstick by which the value of things can be compared, and you don't have that in a system where every conceivable good or service purports to be "money".

  • 1
    $\begingroup$ As for defining goods accurately enough, I think we don't have to define it in a way that is meaningful globally (e.g. such as "coffee" as a label). I think it is sufficient to see how the transactions/trades form a graph to link various people together. E.g. coffee producer $p_1$ could make a better coffee than $p_2$, but $p_2$'s worth will be higher to, say, me, because it links me to, say, AMD to get CPUs (because, say, some AMD staff buys coffee from $p_2$). So, I think, it doesn't really matter if it is coffee (so any ID is OK). I just used "coffee" as examples. $\endgroup$
    – caveman
    Jul 19, 2022 at 14:49
  • 1
    $\begingroup$ @caveman The whole idea behind a unit of account is that all that matters is the number. If I owe you \$100, you don't have to care whether I give it to you in bills, coins, or a bank transfer. All that matters is that you get \$100. So, no, you can't really have a unit of account that is relative. A unit of account is one of the definitional characteristics of money, so without that, you don't have money. $\endgroup$
    – Nobody
    Jul 19, 2022 at 15:30
  • 3
    $\begingroup$ BTW, you seem to assume that computers can solve the calculation problem associated with bartering so many different types of goods, but I'm not sure that's true. For example, I think the problem of making change in your scheme turns into an instance of the Knapsack Problem, which is known to be NP-complete, and therefore very hard to solve for large N, even for computers. $\endgroup$
    – Nobody
    Jul 19, 2022 at 15:38
  • 1
    $\begingroup$ @caveman I don't know about a proof as such, but suppose you have a multidimensional unit of account. Are the dimensions comparable (i.e., can be converted to one another) or incomparable? If they're comparable, then you can simplify greatly by picking one and converting to it, and then you're on your way to reinventing regular money. If they're incomparable, then every payment becomes a multiobjective optimization problem, which is hard (even for computers). Basically, there's a reason why people invented money and why it has the characteristics it has. $\endgroup$
    – Nobody
    Jul 19, 2022 at 19:10
  • 1
    $\begingroup$ @caveman Comments aren't really the place to debate these matters, so I'm going to bow out. If you're serious about pursuing this idea, your next step should be to dig into the conventional theory of money and identify some circumstances, even if only hypothetical, in which conventional money has problems, and show how this alternative scheme solves those problems. $\endgroup$
    – Nobody
    Jul 20, 2022 at 12:47

Money has almost nothing to do with accounting software. The greenback existed long before a centralized ledger.

More of an issue though is that there is no single source of exchange values. There’s no huge database that says a repairing a broken window is equivalent to one pound of ribeye steak. That database would have to be essentially infinite. What is broken on the window? The glass? What size is the glass, is it double paned glass etc. What grade is the meat? Choice? Prime? Is there a quantity discount? What if the meat is only good for one more day can I accept less than the exact equivalent of fixing a broken window?

Separately. Being paid \$5 for your work doesn’t create a debt. Being paid \$5 settles the debt to you and gives you \$5 credit in the economy.

There are two entirely different concepts at play, one is price theory the other is accounting software. Accounting software can’t create money (unless you’re selling the software).

  • $\begingroup$ I edited my post. Does it make you change your thought? As for the $5, I think it rather acknowledges a debt (perhaps that debt was actually created when I did some work for the person who paid me $5). I think, the debt is actually settled when I spend the $5 to get some work back (e.g. get some food to eat). I know that this is not a common terminology with fiat currencies. However, my discussion is fundamental about the theory of money. $\endgroup$
    – caveman
    Jul 16, 2022 at 17:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.