In Satoshi’s whitepaper, the innitial setup is the same: Alice wants to send bitcoin to Bob. Directly, in an instant the BTC amount would be transferred and recieved. That’s the whole point.

Transfer: Alice → Bob

What happens? Alice’s wallet constructs a transaction spending her UTXO. She signs it with her private key, broadcasts it to the Bitcoin P2P network, and a miner includes it in a block. Bob’s node sees the confirmation. Within 25 bytes of locking script the transaction is then complete and the amount is creditet to Bob's wallet.

OP_DUP OP_HASH160 <bob's pubkey hash> OP_EQUALVERIFY OP_CHECKSIG

However, Alice may do some things different: This is more her way of doing things nowadays.

She opens her digital wallet app, which she doesn’t hold the keys to — Steve does. Therefore, she doesn't need to run a node — Steve also does this.

Alice innitialises the transaction to Bob's wallet, Steve then constructs and broadcasts the transaction within the network. On the other end, Bob doesn’t need to validate anything either. Steve's App tells him whenever the transaction arrives. So Steve built a wallet on Alice's and Bob's trust, Alice's transaction and it's arrival at Bobs wallet adress, proof it to work.

The cryptographic proof that makes Bitcoin Bitcoin — the part where you don’t have to trust another party — is missing, a proof of work only transmitted by Steve.

Alice and Bob didn't even have to know Steve, nor eachother.

Alice → Steve's wallet app → Steve's wallet app's node → mining pool → block → Steve's wallet app's node → Steve's wallet app → Bob

The transaction was technically valid. It obeyed every consensus rule. It paid the fee. It will persist in the UTXO set until spent. By every protocol-level measure, it was a real Bitcoin transaction. But by the measure that actually matters in P2P cash system transactions (two parties transacting without a trusted third party) it failed.

Alice doing thing's differently nowadays is quite possible given the median Bitcoin experience in 2026. There are 1.26 million BTC sitting in ETF cold storage. 194 public companies holding bitcoin on balance sheets. Exchanges custodying the vast majority of retail coins.

The protocol is peer-to-peer. The product is client-server.

Even if the tools for real P2P usage are better than ever: Sparrow Wallet builds transactions locally. Lightning settles in two seconds for a fraction of a cent. Running bitcoind on a Raspberry Pi costs less than a month of Spotify.

The exit from custodial dependency is always available. Alice could hold her own keys, run her own node, and send directly to Bob’s address without any corp. in the middle.

Still, she chose trust over work. Maybe as it's easier, maybe because of it's effortlessness.

And, maybe we then need to build on self-custody again.

In Satoshi’s whitepaper, the innitial setup is the same: Alice wants to send bitcoin to Bob. Directly, in an instant the BTC amount would be transferred and recieved. That’s the whole point.

Transfer: Alice → Bob

What happens? Alice’s wallet constructs a transaction spending her UTXO. She signs it with her private key, broadcasts it to the Bitcoin P2P network, and a miner includes it in a block. Bob’s node sees the confirmation. Within 25 bytes of locking script the transaction is then complete and the amount is creditet to Bob's wallet.

OP_DUP OP_HASH160 <bob's pubkey hash> OP_EQUALVERIFY OP_CHECKSIG

However, Alice may do some things different: This is more her way of doing things nowadays.

She opens her digital wallet app, which she doesn’t hold the keys to — Steve does. Therefore, she doesn't need to run a node — Steve also does this.

Alice innitialises the transaction to Bob's wallet, Steve then constructs and broadcasts the transaction within the network. On the other end, Bob doesn’t need to validate anything either. Steve's App tells him whenever the transaction arrives. So Steve built a wallet on Alice's and Bob's trust, Alice's transaction and it's arrival at Bobs wallet adress, proof it to work.

The cryptographic proof that makes Bitcoin Bitcoin — the part where you don’t have to trust another party — is missing, a proof of work only transmitted by Steve.

Alice and Bob didn't even have to know Steve, nor eachother.

Alice → Steve's wallet app → Steve's wallet app's node → mining pool → block → Steve's wallet app's node → Steve's wallet app → Bob

The transaction was technically valid. It obeyed every consensus rule. It paid the fee. It will persist in the UTXO set until spent. By every protocol-level measure, it was a real Bitcoin transaction. But by the measure that actually matters in P2P cash system transactions (two parties transacting without a trusted third party) it failed.

Alice doing thing's differently nowadays is quite possible given the median Bitcoin experience in 2026. There are 1.26 million BTC sitting in ETF cold storage. 194 public companies holding bitcoin on balance sheets. Exchanges custodying the vast majority of retail coins.

The protocol is peer-to-peer. The product is client-server.

Even if the tools for real P2P usage are better than ever: Sparrow Wallet builds transactions locally. Lightning settles in two seconds for a fraction of a cent. Running bitcoind on a Raspberry Pi costs less than a month of Spotify.

The exit from custodial dependency is always available. Alice could hold her own keys, run her own node, and send directly to Bob’s address without any corp. in the middle.

Still, she chose trust over work. Maybe as it's easier, maybe because of it's effortlessness.

And, maybe we then need to build on self-custody again.