old nodes accept p2spkh spends because they see witness version 3 as "anyone can spend." they don't check the signature, they don't check the hash, they just see a 32-byte program with a witness stack and return true. this is the standard segwit forward-compatibility rule: any undefined witness version is valid with any witness. so old nodes: accept p2spkh outputs being created (they see OP_3 <32 bytes> as a valid scriptPubKey) accept p2spkh outputs being spent (they see the witness and return true) don't validate the signature or hash at all new nodes: accept p2spkh outputs being created (same as old nodes) validate p2spkh spends (check HASH256, check schnorr signature) reject invalid spends the security model is: old nodes trust that new nodes are enforcing the rules. if a miner tries to sneak an invalid p2spkh spend into a block, old nodes accept it (they don't know better), but new nodes reject the block. the miner's block is orphaned by the new-node majority. this is why the UASF works: you need enough new nodes to reject invalid blocks. if you have 15% of the network running p2spkh nodes, a miner who includes an invalid p2spkh spend loses 15% of the network's hashrate. that's enough to make it unprofitable.