Most people think, that the whole point of Proof of Work in Bitcoin is to determine payment finality for the receiver of a payment: When more Proof of Work is build upon it, payment finality grows. Zero confirmations are just a promise on payment finality, one confirmation starts the Proof of Work accumulation, and six confirmations are considered final.
It's not that this is wrong. But it is just a part of the whole picture. Overfocusing on it - like most people do - delutes the attention from how things happen for real. For example, some people dismiss unconfirmed transactions entirely as insecure, because "they are not backed by Proof of Work", which is "the only point" of Bitcoin. You should never accept unconfirmed transactions and so on.
In BTC, the developers damaged unconfirmed transactions, because "they are not secure anyway". They added Replace-by-Fee, which allows to resend the same unconfirmed transaction with different properties, which can be an annoyance. Further they added a delay in transaction propagation by nodes, a random time between 1 and 5 seconds. This makes unconfirmed transactions slower. They are no longer instant, which is a significant downgrade of user experience. But hey, they are not secure anyway, and not really Bitcoin. Why not just use PayPal?
Such considerations overfocus. They ignore hat Proof of Work is just a part of determining payment finality for the receiver of a payment. There is a large array of further factors, and in most cases some are more important than Proof of Work. For example, the amount: Economically it makes less sense to double spend a 10 Dollar unconfirmed transaction than to double spend a 500,000 Dollar transactions with two confirmations. If you validate the identity of your counterparty and have access to a legal system for prosecution, it will massively decrease the risk of being double spended (and thus improve finality). Finally, if the service you sell is not finalized, like an account at netflix, the finality of a payment itself is not so important.
When you determine the finality of your payments, Proof of Work is just one of many factors. It is not even a very reliable factor. It can fail: For example, in Bitcoin Gold and Ethereum Classic the miners attacked exchanges with doublespending transactions with a lot of confirmations and Proof of Work. The security of a confirmation always depends on the amount of the payment and the context, and it is hard to determine it absolutely.
People make mistakes and lose money everyday. It can't be the purpose of Bitcoin to protect people from being frauded. Nothing can do this. But if "the whole point" of Proof of Work is not to prevent people being double spent - what is it then?
The answer is simple: Proof of Work is not here to protect people from double-spends - it must protect the ledger of transactions from incorporating double spends. The difference is important.
If people suffer from double spendings, even after many confirmations, the system lives on. Bitcoin Gold and Ethereum Classic did not die after the Proof of Work failed to protect exchanges from double spends. The result was just that exchanges needed to learn, that Proof of Work is just an indicator, and that they interpreted its interaction with the envirenment wrongly. Proof of Work is allowed to failed here.
However, it is not allowed to fail in protecting the ledger to incorporate a double spend. If a single double spend becomes part of the ledger, everything is broken. When the nodes have different histories about the same coin, consensus will be gone and the network will fragment. It will become impossible to agree that a coin - the input of a transaction - is valid. Bitcoin will no longer be usable.
Double spends are a deadly poison. Putting a single one permanently into the ledger will kill the whole system immediately. This is the real service, Proof of Work provides: Protecting the ledger. Helping merchants to determine the finality of a payment is more like a side effect of it.