Transaction Weight

A metric for measuring the "size" of a transaction.

With the introduction of BIP 141 (Segregated Witness), transactions were given a new unit of measurement called weight.

This weight measurement basically gives the validation part of a transaction (the signature data) a discount, so it doesn’t take up as much “space” inside a block. In other words, the signature data is less costly than the rest of the transaction data.

Calculation

The weight of a transaction is calculated by multiplying the size (in bytes) of different parts of the transaction data by different values:

Data	Multiplier
Version	x4
Marker	x1
Flag	x1
Input Count	x4
Input	x4
Output Count	x4
Output	x4
Witness	x1
Locktime	x4

Example

Transaction: c586389e5e4b3acb9d6c8be1c19ae8ab2795397633176f5a6442a261bbdefc3a

0200000000010140d43a99926d43eb0e619bf0b3d83b4a31f60c176beecfb9d35bf45e54d0f7420100000017160014a4b4ca48de0b3fffc15404a1acdc8dbaae226955ffffffff0100e1f5050000000017a9144a1154d50b03292b3024370901711946cb7cccc387024830450221008604ef8f6d8afa892dee0f31259b6ce02dd70c545cfcfed8148179971876c54a022076d771d6e91bed212783c9b06e0de600fab2d518fad6f15a2b191d7fbd262a3e0121039d25ab79f41f75ceaf882411fd41fa670a4c672c23ffaf0e361a969cde0692e800000000

Data	Bytes	Multiplier	Weight
Non-Witness	106	x4	424
Witness	110	x1	110
			534

FAQ

Why x4?

The factor of 4 helps to create more of a balance between the cost of creating an output and the cost of spending an output.

In typical transaction data, an output can be created using around 32 bytes, but spending it requires around 108 bytes:

With the old metric, every *byte* of data you included in a transaction increased its fee. So as a result, it cost more to spend an output than it did to create one.

However, by using the new weight metric, there is more of an even balance between the cost of creating an output and spending an output:

With the weight metric, spending an output “costs” roughly the same as creating one.

As a result, this creates a more balanced incentive to spend outputs, which should help to reduce the size of the UTXO set (which takes up valuable space in RAM on a computer running a full node).

Thanks

Thanks to luke-jr for explaining how multiplying non-witness data by 4 helps to create a balance between the costs of creating and spending UTXOs.
Thanks to Pieter Wuille for this helpful link: https://segwit.org/why-a-discount-factor-of-4-why-not-2-or-8-bbcebe91721e