Transaction fees are there, but don't have to be enormous. Our code can also optimistically unlock the page under the assumption that pending transactions will eventually go through, so block time is also less of a concern.
Right now the tricky thing is that it requires a browser extension if you want to transact directly with crypto (or use one of the few browsers that come with a built-in web3 wallet).
We also have a managed crypto wallet service that allows people to pay with credit cards, but it isn't widespread because it can only be used to buy keys for locks we have approved.
The cost of crypto transactions has always been known, it just used to be paid for by the userbase as a whole when the block reward was more substantial, whereas right now the user pays for their own transactions.
I actually think the Lightning network is a semi-viable solution, it essentially overlays a "credits" system onto the original network that's especially appropriate for micro-transactions.
Besides, "every node" is not true of Bitcoin even today. Neither my desktop nor my mobile Bitcoin client save a complete history. Miners need history, but maybe there can be a different model of mining. I'm not a customer at all the credit card companies out there, only one. Maybe I could also be only be a customer of only one fixed "miner" in some future system? Possibly a new but fixed one I can pick per transaction, rather than today's essentially random assignment of who gets my fees? Yes, this wouldn't be Bitcoin anymore, but as I wrote above, I don't think Bitcoin is what we need anyway.
Iām pretty sure that last time I repaired an ASIC it had just one crappy 256MB sd card (which was the reason it needed a repair). Maybe pool coordinators still need history, idk.
That's for purely on-chain transactions. There's also a partly-offchain solution called "zkrollups" that uses zero-knowledge proofs to reduce the storage for a simple value transfer to 11 bytes, without introducing any unpleasant security assumptions like requiring the user to be online to receive payment or prevent cheating. Combining that with sharding gets to about a quarter million tx/sec.
If that's not sufficient for micropayments, it's fairly straightforward to extend zkrollups to make payments probabilistic: e.g. to pay one penny, send a 1% chance of paying a dollar. There are some potential doublespending issues here but for voluntary tips that's not a problem.
I can only hope the former will notice that offering ancient technology is not a good way to keep customers.
So then it's not free. Instead of charging you a transaction fee, they charge you "time", since time is money, and they can make money off the in-flight money (via interest, ostensibly).
SEPA is bank transfers with delays of entire days - that makes them completely unfit for micropayments.
In Australia, our regular banking system lets you register your mobile phone number (or business/company number or email) as a 'PayID' and then that's all the info you need to send somebody an instant payment... Even our older system usually only takes until the next afternoon or morning on a business day (I believe the older inter-bank payments are settled in two batches per business day).
I think the usual explanation is that this system gives banks a pool of money that they are always free to invest in the time that it's neither in the original owner's nor the recipient's account.
The Wikipedia page says "Area served: Sweden" and points out that even Norway and Denmark use similar systems. So this is quite limited. But sure, would be good to have something like this globally.
I agree it would be great to have something like this globally. Or at least to begin with inside of the SEPA (which recently introduced international instant payments)