We could all switch to proof-of-stake, but then anyone would be able to create a valid-looking blockchain in a few minutes, and the only way to get the right chain would be to connect to nodes you know or trust, at which point a blockchain becomes meaningless in the first place.
The challenge is about coming to agreement. Reaching agreement on information which determines how much money people own is hard. Proof-of-work reaches consensus by having a vote -- on which transaction history is correct -- constitute something that cannot be forged: energy.
Most importantly, though, it's decentralized. It's simple to implement a fast, centralized currency and payment system. But Bitcoin's main attraction is that it values decentralization greater than efficiency and throughput, thus making it, primarily, a store of value, as opposed to a means of payment.
100 GB blocks is politically unrealistic in the foreseeable future but technologically not really, especially because network and storage continue to get better and cheaper even if CPUs are stalling. Nodes don't need to store the full chain, and 10 years of that size of block would only be about 50 PB. Gigabit networks are only becoming more and more common, a 1 Gb connection will let you get 75 GB over 10 minutes (and using the math from the wiki you only need 0.76 Gbps to handle 200k tps). Also not every node needs to see every transaction, the effective bandwidth can be much lower, just like the effective storage.
And as the wiki points out, PayPal has a much lower TPS than Visa or others but is still widely used as a direct consumer-to-merchant or friend-to-friend payment system. In addition, the better straw argument against Bitcoin as a payment system has always been "who is going to wait 10 minutes on average, sometimes much longer, especially if you want the gold-standard 6 confirms, to buy a stupid pack of gum?"
The whole point of SEPA transfers was to move all transfers to be cleared directly via the main ledger, in a matter of seconds, across different countries and currencies.
If Bitcoin is inferior to an existing payment system used by 500 million people, then it’s worthless.
If Bitcoin is strictly inferior to an existing payment system used by 500 million, then it's worthless.
IMO the major value proposition of bitcoin is that no one can exert influence to invalidate tx or blacklist certain users. Payment systems that have this property will always have some utility for organizations that nation state actors are trying to stop, e.g. ransomware, any darknet market provider, political dissidents, freedom fighters/terrorists, people trying to participate in a market in a controlled economy etc.
So the increasing cost has nothing to do with use (aka transactions). It is a consequence of providing security to a growing amount of value on the network. There is simply no better way to do this currently assuming you want a permissionless architecture. It is step 1 towards creating an open finance system. Other steps like proof of stake will eventually solve this problem.
Bitcoin does a few transactions per second on-chain, however those few transactions per second support many more transactions per second on systems built on top of Bitcoin in a similar way to how the tx-per-second Canada's inter-bank transfer system does is quite low, but those settlement transactions support orders of magnitude more transactions per second on systems built on top of that settlement system.
An example of exactly what you proposed is Blockstream's Liquid sidechain technology, which maintains a high-speed ledger secured via multi-party signing between various exchanges to allow high speed, highly efficient, off-chain transactions backed by on-chain Bitcoins.
I’m sorry, but I don’t see the value in Bitcoin if it is worse than SEPA transfers.
With bitcoin you can send money to anywhere in the world for 20cents. Thats cheaper than a 30$ wire transfer.
Also, the censorship resistant part of bitcoin is really important, and is being used by the Venezuelans right now.
And there are technologies being built on top of it that will make it even better.
This is like 1994 internet right now. Even though it works, and has significant benefits, there are still a whole lot more stuff to come that is being worked on.
There's no need to trust neither the processor or the issuer, only the protocol & distribution of hashing power.
then I'd argue it is still centralized - you have this 10 nodes acting as the traditional bank.
Scales to 100k tx/sec, supposedly.
Off the top of my head idea:
What about a system that uses a bit of real-world, legal responsibility to generate consensus. Something a bit like ICANN and the domain name system.
Take 10 or so trusted people or organistions worldwide, get them to each run a node, each of which coordinate via the Raft protocol or whatever. A distributed ledger is agreed on by each of those 10 nodes.
This should use far less power and be able to handle more transactions per sec than bitcoin.