This BIP describes 2 new types of Bitcoin address to support native Segregated Witness (segwit) transactions. The first type resembles the original P2PKH base-58 address. The second type is a z-base-32 representation of a witness program with Damm algorithm checksum, which supports arbitrarily complex transactions, and is forward compatible to new version witness programs in the future.
The first new Bitcoin address format defined is specific to pay-to-public-key-hash segwit transaction. The scriptPubKey is an OP_0 followed by a push of version 0 witness program:
Version byte is 0x19 for a main-network address, 0x41 for a testnet address. The following 20-byte is the public key hash. And the 4-byte checksum is the first four bytes of the double SHA256 hash of the version and public key hash.
All addresses generated with this scheme will a constant length of 34 characters, with a "B" prefix for main-network and "T" prefix for testnet.
{|class="wikitable" style="width:40ex; text-align: center; margin: 0 auto 0 auto;"
!width="12%"|Value
!width="12%"|Symbol
!width="12%"|Value
!width="12%"|Symbol
!width="12%"|Value
!width="12%"|Symbol
!width="12%"|Value
!width="12%"|Symbol
|-
| 0 || y || 8 || e || 16 || o || 24 || a
|-
| 1 || b || 9 || j || 17 || t || 25 || 2
|-
| 2 || n || 10 || k || 18 || 1 || 26 || 4
|-
| 3 || d || 11 || m || 19 || u || 27 || 5
|-
| 4 || r || 12 || c || 20 || w || 28 || h
|-
| 5 || f || 13 || p || 21 || i || 29 || 7
|-
| 6 || g || 14 || q || 22 || s || 30 || 6
|-
| 7 || 8 || 15 || x || 23 || z || 31 || 9
|}
It is case-insensitive and includes all alphanumeric characters excluding 0, 2, l, v. The order of alphabet is chosen so that less ambiguous alphabet characters will appear more frequently than others.
The 54-digit raw address is then divided into 9 equal segments. For each segment, a checksum is calculated with Damm algorithm and appended to the end of the segment. This makes the length increases to 63 digits.
A second round of checksum is calculated by taking the digits in the same position of the 9 segments, and appended to the end of the address. This is the final address with 69 digits.
* A native witness program output is a scriptPubKey with a push of version byte followed by a push of witness program, and nothing else;
* A witness program in P2SH is a P2SH redeemScript with a push of version byte followed by a push of witness program, while the scriptPubKey looks like a normal P2SH output.
As the P2SH address has been defined in 2012, using witness program in P2SH allows most existing wallets to pay a segwit-compatible wallet without any upgrade. However, this method requires more block space and is less collision-resistance than a native witness program, and is only a short-term solution to make the transition smoother. Eventually, all users are expected to use the more efficient native witness program as the primary payment method.
The drawbacks of Bitcoin addresses have been extensively discussed in BIP13. Since then, better payment methods have been proposed or deployed, for example:
*BIP47 Reusable Payment Codes for Hierarchical Deterministic Wallets
However, none of these are as widely adopted as the suboptimal base-58 scriptPubKey template addresses, which is still a standard for the whole eco-system, from wallets, block explorers, merchants, exchanges, to end users. As P2PKH transactions are still dominating the blockchain, the author believes that the proposed P2PKH segwit addresses is the easiest way for wallets and services to adopt native witness program, which is particularly important in the context of scaling the capacity of the blockchain.
The P2PKH segwit addresses only allow simple payment to a single public key. For arbitrarily complex segwit transactions, the general segwit address is used. The use of z-base-32 eliminates case-sensitiveness and simplifies transformation to and from hexadecimals. The Damm algorithm checksum allows detection of all single-digit errors and all adjacent transposition errors. With the excellent upgradability of witness program, this proposal is also forward compatible to new version witness programs that is not longer than 32 bytes.
This proposal is not backward compatible, but it fails gracefully -- if an older implementation is given one of these new Bitcoin addresses, it will report the address as invalid and will refuse to create a transaction.
