Smart contracts are a bet on our ability to write code that is free from flaws. History suggests that the odds of that are not good. I'd rather trust a bank and the legal system.

> History suggests that the odds of that are not good. I'd rather trust a bank and the legal system

Which is the silly part about "smart contracts" in the first place. Any smart contract dispute will just "devolve" back into the meatspace legal system. Anything that isn't a dispute because it is just "routine execution" (eg: moving money from account A to account B automatically) can simply be called "business automation" that consist of a bunch of "business rules"--basic terminology that has existed since the dawn of computers.

Smart contracts do nothing but add overhead to any transaction.

The code that runs on your bank and "legal system" was written by people who make mistakes as well.

Writing smart-contracts isn't easy (and I author and audit them for a living...) but it's doable.

Your trust on a new technology grows as you get used to it. Anecdotally, I remember how hard it was to get OS virtualisation accepted in corporate environments just a decade ago and my last enterprise customers where already using all form of virtualisation (net, storage, os, etc)..

> The code that runs on your bank and "legal system" was written by people who make mistakes as well.

And the results the code produces is constantly audited, both internally and externally.

That is true of many cryptocurrency companies, platforms and libraries.

There are live businesses using them today, and I'm not just talking of UNICEF and Axa...

The problem is that there is no way for a smart contract to actually know if goods have been received or not, so it is still subject to abuse.

That's true but it can be made to work with oracles or a third-party arbitration (eg: check out Kleros, Truebit and Plasma) system.

But now you’re no longer “without a third party.”

There are nuances to the type of third parties we are talking about here.

The type I'm discussing, something like the Lightning Channel and Plasma operators, these are roles that have cryptoeconomic properties that don't allow them to take over your wallet or refuse your service.

edit: The base premise of having a smart-contract being itself the escrow is perfectly achievable. This doesn't mean that such a setup is the best solution to every use case and I believe that we'll see plenty of developments in this area in the coming years.

For now the best approaches use game theory and computer science to reduce the amount of "trust" the parties must have in each other and the system itself but they are very much under active research & development.

That isn't "trustless" anymore! You gotta trust a bunch of unknown crypto startups now.

Are these guys really add more value to the economy than chargebacks? I very much doubt it.

That's not correct. You can have a third party that is trustless, ie that cannot perform any action on its own that would favour or harm you.

Take for instance the role of Plasma operator or Lightning Channel operator; The role allows for a third party to have the costs of setting up infrastructure for you to use, but at no point are your funds held by the operator and you are always able to exit a contract if you submit on-chain evidence of fraud.

They could easily collude with the party you are making a contract with, and rule against you however.

That's not the case in many (most?) of the cases I mentioned.

In plasma for instance, the operator and seller could collude against the buyer but any "illegal" operation they do on-chain is enough for anyone to trigger a mass-exit and cause financial and reputational loss to the operator.

A common theme in all the on-chain payment technologies I mentioned is that they do not put any one party in "charge" but the volunteers who run staking channels of any sort (plasma, lightning, etc) can and will suffer financial damages (loss of deposit, etc) if proven to act maliciously.