Introduction to bitcoin mining

What is bitcoin mining?
Mining is the process of writing pages (blocks) of bitcoin transactions into the bitcoin ledger, called ‘The Bitcoin Blockchain’, and getting rewarded with newly created bitcoins.

To understand this in more detail, the rest of the post describes:

How do bitcoin transactions work?
Why is mining needed in bitcoin?
Why do miners mine?
What is this ‘computationally expensive’ guessing game?
Why pay rewards in BTC instead of USD?
Who mines?
What can and can’t miscreants do?
If you are new to bitcoin, it might be worth having a quick read of “A gentle introduction to bitcoin”.

How do bitcoin transactions work?
The process is:

Make a payment (a bitcoin transaction)
Wait for it to be mined in a block (average 10 mins)
Wait for more blocks to be mined on top (average 10 mins per block)

1. Make a payment. When you make a bitcoin payment, the transaction message is sent to the network and passed around all the network participants (called ’nodes’), and remains in an ‘unconfirmed’ state. This means the nodes have seen that the payment has been initiated, and they have validated it according to certain technical and business logic rules, but it isn’t yet written into anyone’s bitcoin blockchain ledger.

Unconfirmed transaction = valid, known transaction, but not yet included in the ledger.

1. Wait for it to be mined in a block (average 10 mins). Miners take the list of unconfirmed transactions (specifically, those that they know about), and they bundle them into a block, which is just a list of transactions plus some other data.

They then get to work ‘mining’ the block which means playing a guessing game to find a random number (more later).

If they guess right, then the block is published to the rest of the network. The computers on the network validate that the block meets the criteria, and then ignore it or store it into their blockchains. The competition then starts again with the unconfirmed transactions that have accumulated since.

The network adjusts the difficulty of the guessing game to target a block being created every 10 mins or so, irrespective of the amount of computing power in the network.

1. Wait for more blocks to be mined on top (average 10 mins per block). The next block that is mined on top of the one with your transaction will refer to the previous block (hence, ‘blockchain’). The more blocks that have been built on top of the one with your transaction, the more ‘baked’ into the blockchain it is, and so the harder it is to unwind through block-reorganisation attacks (more later).

Unconfirmed transaction -> Confirmed transaction (1 block) -> Confirmed transaction (many blocks)

The current advice suggests that after 6 blocks, the chances of the transaction being unwound due to a competing longer chain replacing your blocks is very small. If you are receiving a payment, then the higher the value your payment, the longer you may want to wait to reduce the chance of your payment being unwound.

Why is mining needed in bitcoin?
There are two parts to this. First you need a way to get transactions into the ledger, secondly you need a way to make it expensive for miscreants to add dishonest blocks.

Ledger addition. Transactions are added to the ledger in blocks so as to create some sort of time order to the transactions. In bitcoin you can’t trust the timestamp of any particular participant, and there is no ‘master clock’ to trust, so block order is the equivalent of time order.

Financial deterrent. This is about the guessing game, called “Proof of work”. You don’t actually need the guessing game to add blocks to a blockchain. However, the guessing game makes it computationally expensive (therefore financially expensive) to add blocks. This cost acts as a deterrent to miscreants who would otherwise want to add their dishonest blocks. So long as most of the network is ‘honest’, then the dishonest parties will have a tough time creating rogue blocks.

“Why proof of work?”, in three acts:

Act 1

Anyone can create blocks on an “open” network.
As you can’t trust anyone specifically, each individual node has to assume that the ‘majority’ of the rest of the network is right.
So to dominate the network, you just need to create many aliases who are all under your control and all agree with each other. This kind of domination-by-numbers is called a ‘Sybil attack’.
Act 2

It is cheap and easy to spawn validators who all agree with each other.
Therefore it is very cheap to bully the network.
So for a network to be secure against this, you need to have a more expensive way to bully the network.
Act 3

Computational power is more expensive and requires investment and upkeep.
Therefore use majority-by-computational-power instead of majority-by-numbers.
So miscreants will need to spend a lot more money to dominate the network.
Finale

The name given to a challenge that is computationally expensive for the sake of it, is called a “Proof of work” challenge.
Why do miners mine?
Mining reward = Voluntary transaction fees + Block reward (currently 25 BTC per block)

When you mine a block, get to collect any voluntary transaction fees from the transactions you have included. You also get to write one transaction paying yourself some BTC (currently 25 BTC, and reducing to 12.5 BTC in the middle of 2016). This is called a ‘block reward’ or ‘coinbase transaction’ (not to be confused with the American company called “Coinbase” which operates under a UK legal entity “Coinbase UK, Ltd”).

This is the ‘minting process’ i.e. how bitcoins are created. The reward decreases with time, and in theory, transaction fees will replace the block reward.

tx_fees_replace_block_rewards
Transaction fees are not mandatory (hence the “bitcoin transactions are free” mantra) but miners will seek out transactions containing fees, and preferentially add them to blocks that they are creating. If there are more unconfirmed transactions than can fit in a block, rational miners will mine the ones with the highest transaction fees first.

What is this ‘computationally expensive’ guessing game?
Miners spend a lot of computing power trying to guess a number, which when added to a block and put through an algorithm, outputs a ‘hash’ that meets certain criteria.

A hash is a fingerprint of data. It’s easy to make a hash from some data but computationally impossible to create the data from the hash. Hashes look random compared with the data put in.
You can play with hashing here: Go to http://www.xorbin.com/tools/sha256-hash-calculator and type some data into the big box. You’ll see the hash in the smaller box. I typed “What does the hash of this look like?”:

hash1
It’s easy to generate a hash from some text, but impossible to re-generate the text from the hash.

If you change just one part of the data, the hash looks entirely different. I added a question mark:

hash2
Adding or changing just one characters results in a totally different-looking hash.

By changing the data slightly, try to find a hash starting with 0000000. Tricky eh?

By adding “-17” to the sentence, I found something that gave a hash starting with one zero:

What does the hash of this look like?-17 = 0fd82107e6e73b6f369853da3b53d4a93e8be1e5b3a4dd7da2b4ea644774bc80

I kept going, and to find something that gave a hash starting with a double zero, it took 272 attempts:

What does the hash of this look like?-272 = 00629a604a7ec6b1f05e7703c57197ed6119a6282e9b5f750e14a1500578d3fd

Bitcoin block mining. Bitcoin mining is essentially the same game, where you tweak the input data (the block header) so that you get an output hash that matches what is required by the network at that point in time.

A recent bitcoin block #372910 was ‘solved’ because the hash was 000000000000000000b037a61e47df14b035199b5a2d464691b9456394bc07da – this had enough zeroes to satisfy the network at this time*.

• More accurately (for pedants) the block header containing the nonce is hashed twice using the SHA-256 hashing algorithm, and had to meet a number smaller than the target number determined by the network difficulty of 54,256,630,327.89 (at block #372910).

Further fun. If you are up for some light programming, there is an excellent guide to playing the guessing game in Python on Alex Gorale’s blog.