Every time I look at accounting, the different kinds of accounts baffle me. I can never keep straight what each kind of account is used for, or which ones have positive credits and which ones have negative credits.

As far as I can tell, the point is to double the amount of work in the hopes of catching certain kinds of errors. Which makes sense when you have humans making the entries and humans doing the arithmetic.

But I grew up in a world where computers do all of the math, and it always looks to me like it's violating the Don't Repeat Yourself principle. If you say the same thing in two different places, one of them is always going to be wrong.

I feel as if, had accounting been designed in the modern era, we wouldn't have done it that way.

I'm not an accountant and my failure to understand does not make the thing wrong. But my bafflement at "credits decrease an asset account" feels emblematic of something being genuinely off base.

> But I grew up in a world where computers do all of the math, and it always looks to me like it's violating the Don't Repeat Yourself principle. If you say the same thing in two different places, one of them is always going to be wrong.

This is wrong on a couple of levels.

In your understanding do RAID disk arrays and backups violate “the Don’t Repeat Yourself principle”? Is one of the copies of the data guaranteed to be wrong? Do data backups duplicate data because of pre-modern thinking?

But on another level it’s irrelevant, because in double-entry bookkeeping, there is no duplication of information. If you buy an apple for a dollar, your journal entry will mark a dollar out of cash — which is true because you now have 1 less dollar — and a dollar against your “Food” expense account — which is true because the thing you just spent a dollar on was food. If you took away either entry, you would be losing information. The fact that both entries have to balance isn’t because of duplication, it’s because the same dollar can't exist in more than one place at a time, which is axiomatically true regardless of whether you use a computer.

The point is not to double the amount of work to reduce errors. The point is to record both where money came from and where it went to. This simplifies analysis, reporting, etc. down the line.

The fundamental unit of a double-entry system is the transaction, which records from where things came and to where they went. In software parlance, it's an event-sourced system rather than the stateful/interactive system of single-entry accounting.

> the point is to double the amount of work in the hopes of catching certain kinds of errors

That's also how I like to think about it, as a kind of checksum mechanism. Double-entry bookkeeping originated in medieval European markets, which were often open-air, noisy, dirty, full of thieves and other dangers. Keeping your records straight in that environment must be a challenge, and having a logic that allows you to catch some mistakes can be a powerful tool in that context.

But there's more to it, eg it allows you to make a distinction between expenses and investments, so it is actually a truly different way to think about your financial situation than eg looking at cash flows only. Spending X on a buying livestock has a different economic meaning than spending X to pay a security guard. There are economic historians who argue that this kind of perspective was an enabler of early capitalism, because it enables people to see that money spent on an investment isn't lost value.

Thanks for that. Lemme probe a bit deeper.

If you spend $X on sheep, you credit $X where? Debit it from where? You put "$X worth of sheep" on what account? When the sheep die, do you credit some account with "$X worth of dead sheep?" (Where presumably they remain as dead-sheep forever.)

(I'm sorry, I know that sounds dumb.)

How is that "$X worth of dead sheep" different from "$X worth of security guarding" that you supposedly received?

> If you spend $X on sheep, you credit $X where? Debit it from where?

You debit your 'sheep' account (maybe called sth like inventory) and you credit your cash account, or a liabilities towards suppliers account. Your equity stays constant in either case (no profit or loss impact), if you paid cash you've swapped X worth of cash for X worth of sheep, otherwise your liabilities went up by X.

When they die, you debit some kind of expense account (extraordinary losses or sth like that), and you credit the sheep account. In that case, you've made a loss of X and your equity (when you next draw up your balance sheet) will be lower by X (assuming that's the only business case). It might even go negative, eg if the sheep were your only asset and you still have the liability towards the guy who sold them to you.

Thank you. I really appreciate that.

It still feels exactly the opposite of what I thought a credit and debit were. Surely when you add sheep, you credit the sheep account and debit the cash.

I get it. It's just a shift.

Yeah, best not to think of any semantics wrt to the words debit and credit. Debit is left-hand side, credit is right-hand side, and just remember the rules how they work :)

Haha that sounds like shut-up-and-calculate to me. The alternative is to update the rules to use our intuition on whether the action we’re tracking is good or bad for us and map that to increasing or decreasing balances that represent what we own, owe, earn and spend.

More deets here: https://news.ycombinator.com/item?id=40021506

> the Don't Repeat Yourself principle

I think this is meant mostly to apply to code. For data, redundancy is a very common strategy for assuring accuracy and durability.

> If you say the same thing in two different places, one of them is always going to be wrong.

Yes! This is the exact point. The mismatch flags that there is an error that needs to be investigated and added to the ledger as a correcting transaction.

The reason isn’t to double the amount the work to catch errors. storing both sides genuinely stores extra metadata about the transaction that is not captured by only storing it once.

In double-entry, you’re tracking two different things: - Your current net worth. These are tracked in State accounts (Asset and Liability). - The reasons your current net worth changed. These are tracked in Change accounts (Income and Expense).

Say $5 enters your bank account, that’s recorded as an increase in an Asset account. Whether your net worth changed or not depends on the reason. There’s lots of reasons that could have happened, and that’s recorded as the second entry: - Took out a loan: increase in Liability (no net worth change) - Got paid back for a loan: decrease in Asset (no net worth change) - Sold something: increase in Income (net worth did change) - Got a refund: decrease in Expense (net worth did change)

I agree that re-engineering accounting is necessary for software engineers to build financial systems. I think it’s time to ditch credits and debits and instead deal in State and Change accounts with negative numbers.

PS I go more into it here: https://news.ycombinator.com/item?id=39994335

  Not really, the meaning of debit and credit depends on the type of account

That's how most accountants think about it. But I think there's something more fundamental: a CR entry is an increase is what the company owes (to creditors or shareholders), and a DR is an increase in what the company owns.

EDIT: see this link for how this relates to the accounting equation https://news.ycombinator.com/item?id=32501707

That's (somewhat) true for accounts that represent stocks (assets, liabilities, not really for equity though), it's not true for accounts that represent flows (income, expenses). Income is recorded as a credit entry in an income account, eg (the corresponding debit entry would typically be on something like a current account or claims on customers).

Consider the positioning of income and expense accounts within the accounting equation. Essentially, they are components of equity. View equity as the company's obligation to its shareholders. A credit entry signifies an increase in what the company owes, whether to creditors or shareholders, while a debit entry reflects an increase in the company's assets.

Isn't that exactly backwards to what most people think of credits and debits? If you credit me something, I now have something. I don't owe anything.

I can kinda squint and see "Oh, you want the universe to balance, so if I have something it is some kind of karmic debt". But it still feels like exactly the opposite of what I grew up thinking of these terms to mean.

It's the opposite because when a counterparty (like a bank or a store) says they're 'crediting your account', they're talking about the impact from their perspective, not your perspective.

Second sentence of that wikipedia article is: "A debit entry in an account represents a transfer of value to that account, and a credit entry represents a transfer from the account."