What is a hash, and why do blockchains use them?

A fingerprint for data

A hash function takes some data and turns it into a short code of fixed length.¹ That code is called a message digest, or just a digest.² The data can be almost anything: a file, some text or an image.²

Think of the digest as a fingerprint for the data. The same data always gives the same digest, so anyone can hash it again and compare the results.² If they match, the data has not changed.²

Small change, big difference

Even the smallest change to the input, such as one bit, gives a completely different digest.² The US standard for these functions says any change to a message will, with very high probability, give a different digest.³

Three safety properties

Hash functions used for security are built to have three properties.^2,1 NIST's example of how strong they are: finding a SHA-256 collision would take the whole 2015 Bitcoin network about 36 trillion years.²

• One-way. Given a digest, it is computationally infeasible to work out an input that produces it. This is called preimage resistance.²
• Second-preimage resistant. Given one input, you can't find a different input with the same digest.²
• Collision resistant. You can't find any two different inputs that give the same digest.²

SHA-256

SHA-256 is a hash function used by many blockchains.² Its digest is 256 bits long, usually shown as a string of 64 characters using the digits 0 to 9 and the letters a to f.² The US standard FIPS 180-4 specifies it.^3,2

There are about 10 to the power of 77 possible SHA-256 digests.² Two inputs with the same digest must exist in theory, but finding one would take about 2 to the power of 128 tries on average.² The 2015 Bitcoin network made about 300 quadrillion hashes a second, which is how NIST reached its 36 trillion years.²

Bitcoin uses SHA-256. Ethereum uses a different hash function, called Keccak-256.² Keccak won a NIST competition to create the SHA-3 standard.²

What blockchains use hashes for

Blockchains use hashes for many jobs.²

• Giving things a unique ID. In Bitcoin, a transaction's ID is the hash of the signed transaction.^2,4
• Locking in a block's contents, by storing a hash of its data in the block's header.²
• Linking each block to the one before it, as the next lesson explains.²
• Making addresses.²

Hashes can age

NIST deprecated an older function, SHA-1, in 2011, which means it advised against using it. It disallowed SHA-1 for digital signatures at the end of 2013.¹ NIST also notes that if a flaw were found in SHA-256, blockchains using it would need a hard fork to switch to a new function.² Lesson 6 explains forks.