Block time

The block difficulty formula we saw earlier uses a 10-second threshold to make sure that the difference between the time a parent and child block mines is in is between 10-20 seconds. But why is it 10-20 seconds and not some other value? And why there such a constant time difference restriction instead of a constant difficulty?

Imagine that we have a constant difficulty, and miners just need to find a nonce to get the hash of the block less and equal to the difficulty. Suppose the difficulty is high; then in this case, users will have no way to find out how long it will take to send ether to another user. It may take a very long time if the computational power of the network is not enough to find the nonce to satisfy the difficulty quickly. Sometimes the network may get lucky and find the nonce quickly. But this kind of system will find it difficult to gain attraction from users as users will always want to know how much time it should take for a transaction to be completed, just like when we transfer money from one bank account to another bank account, we are given a time period within which it should get completed. If the constant difficulty is low, it will harm the security of the blockchain because large miners can mine blocks much faster than small miners, and the largest miner in the network will have the ability to control the DApp. It is not possible to find a constant difficulty value that can make the network stable because the network's computational power is not constant.

Now we know why we should always have an average time for how long it should take for the network to mine a block. Now the question is what the most suitable average time is as it can be anything from 1 second to infinite seconds. A smaller average time can be achieved by lowering the difficulty, and higher average time can be achieved by increasing the difficulty. But what are the merits and demerits of a lower and higher average time? Before we discuss this, we need to first know what stale blocks are.

What happens if two miners mine the next block at nearly the same time? Both the blocks will be valid for sure, but the blockchain cannot hold two blocks with the same block number, and also, both the miners cannot be awarded. Although this is a common issue, the solution is simple. In the end, the blockchain with the higher difficulty will be the one accepted by the network. So the valid blocks that are finally left out are called stale blocks.

The total number of stale blocks produced in the network is inversely proportional to the average time it takes to generate a new block. Shorter block generation time means there would be less time for the newly mined block to propagate throughout the network and a bigger chance of more than one miner finding a solution to the puzzle, so by the time the block is propagated through the network, some other miner would have also solved the puzzle and broadcasted it, thereby creating stales. But if the average block generation time is bigger, there is less chance that multiple miners will be able to solve the puzzle, and even if they solve it, there is likely to be time gap between when they solved it, during which the first solved block can be propagated and the other miners can stop mining that block and proceed towards mining the next block. If stale blocks occur frequently in the network, they cause major issues, but if they occur rarely, they do no harm.

But what's the problem with stale blocks? Well, they delay the confirmation of a transaction. When two miners mine a block at nearly the same time, they may not have the same set of transactions, so if our transactions appears in one of them, we cannot say that it's confirmed as the block in which the transaction appeared may be stale. And we should wait for a few more blocks to be mined. Due to stale blocks, the average confirmation time is not equal to average block generation time.

Do stale blocks impact blockchain security? Yes, they do. We know that the network's security is measured by the total computation power of the miners in the network. When computation power increases, the difficulty is increased to make sure that blocks aren't generated earlier than the average block time. So more difficulty means a more secure blockchain, as for a node to tamper, the blockchain will need much more hash power now, which makes it more difficult to tamper with the blockchain; therefore, the blockchain is said to be more secure. When two blocks are mined at nearly the same time, we will have the network parted in two, working on two different blockchains, but one is going to be the final blockchain. So the part of the network working on the stale block mines the next block on top of the stale block, which ends up in loss of hash power of the network as hash power is being used for something unnecessary. The two parts of the network are likely to take longer than the average block time to mine the next block as they have lost hash power; therefore, after mining the next block, there will be decrease in difficulty as it took more time than the average block time to mine the block. The decrease in difficulty impacts the overall blockchain security. If the stale rate is too high, it will affect the blockchain security by a huge margin.

Ethereum tackles the security issue caused by stale blocks using something known as ghost protocol. Ethereum uses a modified version of the actual ghost protocol. The ghost protocol covers up the security issue by simply adding the stale blocks into the main blockchain, thereby increasing the overall difficulty of the blockchain, as overall difficulty of the blockchain also includes the sum of difficulties of the stale blocks. But how are stale blocks inserted into the main blockchain without transactions conflicting? Well, any block can specify 0 or more stales. To incentivize miners to include stale blocks, the miners are rewarded for including stale blocks. And also, the miners of the stale blocks are rewarded. The transactions in the stale blocks are not used for calculating confirmations, and also, the stale block miners don't receive the transaction fees of the transactions included in the stale blocks. Note that Ethereum calls stale blocks uncle blocks.

Here is the formula to calculate how much reward a miner of a stale block receives. The rest of the reward goes to the nephew block, that is, the block that includes the orphan block:

(uncle_block_number + 8 - block_number) * 5 / 8 

As not rewarding the miners of stale blocks doesn't harm any security, you must be wondering why miners of stale blocks get rewarded? Well, there is another issue caused when stale blocks occur frequently in the network, which is solved by rewarding the miners of stale blocks. A miner should earn a percentage of reward similar to the percentage of hash power it contributes to the network. When a block is mined at nearly the same time by two different miners, then the block mined by the miner with more hash power is more likely to get added to the final blockchain because of the miner's efficiency to mine the next block; therefore, the small miner will lose reward. If the stale rate is low, it's not a big issue because the big miner will get a little increase in reward; but if the stale rate is high, it causes a big issue, that is, the big miner in the network will end up taking much more rewards than it should receive. The ghost protocol balances this by rewarding the miners of stale blocks. As the big miner doesn't take all the rewards but much more than it should get, we don't award stale block miners the same as the nephew block; instead, we award a lesser amount to balance it. The preceding formula balances it pretty well.

Ghost limits the total number of stale blocks a nephew can reference so that miners don't simply mine stale blocks and stall the blockchain.

So wherever a stale block appears in the network, it somewhat affects the network. The more the frequency of stale blocks, the more the network is affected by it.