4 ms·

I understand the cryptographic concept of a blockchain but have a bunch of "real-world scenarios" type questions on how it actually works. Like for example, it

by atonse 9y ago

I understand the cryptographic concept of a blockchain but have a bunch of "real-world scenarios" type questions on how it actually works. Like for example, it's a chain. I get that. But if there are thousands of people adding to the chain all the time, how do they coordinate to make sure they're always adding to the last block? Because this might happen: - I sync and have the latest block - I build another block off the latest block (takes 1 second) - In that time, I no longer have the latest block. Do I have to get the newest block again from the "world" and then recompute? Wouldn't this just be never-ending? And how would this work as more people add to the chain? Would I constantly be out of sync? That's the kind of stuff where I'm completely lost and haven't found a good explanation for it.

hanbura 9y ago
This happens fairly frequently. Whenever you see a new block that forms a longer chain than the chain you currently know, you throw your work away and use that block as your new starting point. But you only do that if the new block has a longer chain, not for a chain of equal length. So if two miners both mine a new block that follows the same block, one will be seen first by some miners, the other will be seen first by others. That gives them different mining power, meaning the chains grow with different speed. After 1-3 blocks this mining power difference will lead to one of both chains being longer from everyone's point of view. At that point the longer chain won, the shorter is forgotten and all work put into it becomes worthless. That encourages miners to make sure they use new blocks as fast as possible. Unless of course you have more mining power than everyone else combined.
- annabellish 9y ago
  What happens to the transactions in the chain that were thrown away? Are they just lost now? Do they need to be added again?
  - hdhzy 9y ago
    Usually the second miner also saw and included the same transactions so they are not lost. If they didn't the transaction is undone in a sense that it just didn't happen. Of course it can still be picked by the next miner. The chain that is longest (by difficulty) is by definition a log of everything that happened. Transactions that are not part of that chain are just candidates, nothing less nothing more.
- jackhack 9y ago
  >the shorter is forgotten and all work put into it becomes worthless Then one can commit work to (what is believed to be) the current, longest chain, only to find later that a branch occurred and the work was lost? As such it is then incumbent upon the contributor to follow the chain for some period of time until other work has contriubted to the chain? So how long is long enough? one contribution? ten? Serious question. How does one know with certainty that work was accepted by group consensus? This uncertainty seems to undermine the idea that blockchain is a trustworthy, predictable, verifiable ledger of activity if the work contributed cannot be known as reliably preserved at the time it is submitted, only later in review. I admit I'm fairly ignorant of the details of this, just attempting to understand. I hope I am misunderstanding something; somebody please steer me back onto the path.
  - SomeStupidPoint 9y ago
    If you had more computing power than the network, you could (theoretically) undo all of BTC by completely replacing the chain. I think in practice, they would ignore such a rewrite of history.
  - hdhzy 9y ago
    If you find a valid block then it's more profitable to switch to this block as a base. Chain splits happen when someone finds a block but is not aware of other valid block.
  - JamesLeonis 9y ago
    > So how long is long enough? one contribution? ten? Serious question. The formal name of this is Proof Of Work (PoC). The idea is to make "mining" blocks take non-trivial amount of computational work in order to forge another transaction block. When a miner finds a valid nonce, it announces it to the rest of the network so others can begin working on that chain. When I say non-trivial amount of work, it would take my computer (nVidia 1080 @ 20MH/s) ~3.5 years to mine one Ethereum block as of now. Comparatively, the combined power in the Ethereum network (119.4 TH/s) mines blocks every 24 seconds. Additionally, the miner that finds the correct nonce that makes a valid block is rewarded. This incentivizes each miner to keep up to date with the latest blockchain and abandon lagging branches and work. > How does one know with certainty that work was accepted by group consensus? Through these rewards and the large computational work involved in mining new blocks, the majority of the miners aggressively gravitate to the longest chain. That means that the longer chain continues to grow longer much faster than the other branches. It's this aggressive growth combined with the reward incentives on the longest chain that keeps the whole network honest. Why mine a block that the network won't reward you for? For the individual miner, it's in their best interests to always be working on the longest chain. > This uncertainty seems to undermine the idea that blockchain is a trustworthy, predictable, verifiable ledger of activity if the work contributed cannot be known as reliably preserved at the time it is submitted, only later in review. This is part of the security of the blockchain. This prevents malicious agents from pushing "valid" blocks, even if they manage to get all the dots in a row. The computational work involved in building the next block is vastly outpaced by the general network; by the time you minted you "attack" block, many many many new blocks were already minted by the network and they won't accept your block. This "consensus" keeps these attacks from happening. However, this still can be attacked. If an agent gained control of the majority of the network computational power, then they can mine blocks faster than the honest miners, thus growing the longest chain and thus control exactly what transactions/blocks are mined. This is referred to as the 51% attack because they become the Consensus. But, using my calculations above, I would need somewhere around ~6*10^6 nVidia 1080 cards all working for me to gain majority in the Ethereum network. So I sleep comfortably at night knowing my ETH is safe.
rtkwe 9y ago
Currently that's solved because transactions aren't added live by every user but instead get batched together and added as a group so individually you just have to wait for some block to pickup your transaction from the pool of available transactions. What you're suggesting could potentially become a problem for a technology if new block propagation takes longer than new block discovery. Today orphan blocks aren't exactly rare there're several a month where two different pools will announce new blocks within a couple seconds of each other and one gets abandoned because it was the later one or because the other block had a new block mined from it faster.
- atonse 9y ago
  "What you're suggesting could potentially become a problem for a technology if new block propagation takes longer than new block discovery." But that's an inevitability. New block discovery only takes longer over time, and block propagation will only get more complicated as you add more users to chain.
  - rtkwe 9y ago
    Luckily every user doesn't require a full copy to submit transactions and use the chain just miners.
hdhzy 9y ago
Well in a distributed system you are in a way always out of sync. The primary achievement of Bitcoin was taking already exiting pieces (hash cash, Merkle trees, signatures etc.) and assembling them in a way that solves consensus in such a system (see [0]). What's interesting is the combination of these pieces. They were known a long time before. Actually I had CS lessons where we were creating virtual currencies with these exact elements long before Bitcoin happened (of course no one got Bitcoin in the process :(... or maybe they did but never told me :) ). How to know Bitcoin better? The best approach is the same as with everything else: reverse engineer and recreate a small coin in your language of choice (chain using hashes plus a simple proof of work using N leading 0s). Remember Satoshi's genius was in assembling pieces in a way nobody thought of, you just need to copy a small version of that. It's not that difficult. For me the most interesting aspect is smart contracts [1]. That's the one underutilized element of the system that is quite versatile. [0]: https://en.m.wikipedia.org/wiki/Byzantine_fault_tolerance#Byzantine_Generals.27_Problem https://en.m.wikipedia.org/wiki/Byzantine_fault_tolerance#By... [1]: https://en.bitcoin.it/wiki/Contract https://en.bitcoin.it/wiki/Contract

by atonse 9y ago

hanbura 9y ago

This happens fairly frequently. Whenever you see a new block that forms a longer chain than the chain you currently know, you throw your work away and use that block as your new starting point. But you only do that if the new block has a longer chain, not for a chain of equal length. So if two miners both mine a new block that follows the same block, one will be seen first by some miners, the other will be seen first by others. That gives them different mining power, meaning the chains grow with different speed. After 1-3 blocks this mining power difference will lead to one of both chains being longer from everyone's point of view. At that point the longer chain won, the shorter is forgotten and all work put into it becomes worthless. That encourages miners to make sure they use new blocks as fast as possible. Unless of course you have more mining power than everyone else combined.

annabellish 9y ago
What happens to the transactions in the chain that were thrown away? Are they just lost now? Do they need to be added again?
- hdhzy 9y ago
  Usually the second miner also saw and included the same transactions so they are not lost. If they didn't the transaction is undone in a sense that it just didn't happen. Of course it can still be picked by the next miner. The chain that is longest (by difficulty) is by definition a log of everything that happened. Transactions that are not part of that chain are just candidates, nothing less nothing more.
jackhack 9y ago
>the shorter is forgotten and all work put into it becomes worthless Then one can commit work to (what is believed to be) the current, longest chain, only to find later that a branch occurred and the work was lost? As such it is then incumbent upon the contributor to follow the chain for some period of time until other work has contriubted to the chain? So how long is long enough? one contribution? ten? Serious question. How does one know with certainty that work was accepted by group consensus? This uncertainty seems to undermine the idea that blockchain is a trustworthy, predictable, verifiable ledger of activity if the work contributed cannot be known as reliably preserved at the time it is submitted, only later in review. I admit I'm fairly ignorant of the details of this, just attempting to understand. I hope I am misunderstanding something; somebody please steer me back onto the path.
- SomeStupidPoint 9y ago
  If you had more computing power than the network, you could (theoretically) undo all of BTC by completely replacing the chain. I think in practice, they would ignore such a rewrite of history.
- hdhzy 9y ago
  If you find a valid block then it's more profitable to switch to this block as a base. Chain splits happen when someone finds a block but is not aware of other valid block.
- JamesLeonis 9y ago
  > So how long is long enough? one contribution? ten? Serious question. The formal name of this is Proof Of Work (PoC). The idea is to make "mining" blocks take non-trivial amount of computational work in order to forge another transaction block. When a miner finds a valid nonce, it announces it to the rest of the network so others can begin working on that chain. When I say non-trivial amount of work, it would take my computer (nVidia 1080 @ 20MH/s) ~3.5 years to mine one Ethereum block as of now. Comparatively, the combined power in the Ethereum network (119.4 TH/s) mines blocks every 24 seconds. Additionally, the miner that finds the correct nonce that makes a valid block is rewarded. This incentivizes each miner to keep up to date with the latest blockchain and abandon lagging branches and work. > How does one know with certainty that work was accepted by group consensus? Through these rewards and the large computational work involved in mining new blocks, the majority of the miners aggressively gravitate to the longest chain. That means that the longer chain continues to grow longer much faster than the other branches. It's this aggressive growth combined with the reward incentives on the longest chain that keeps the whole network honest. Why mine a block that the network won't reward you for? For the individual miner, it's in their best interests to always be working on the longest chain. > This uncertainty seems to undermine the idea that blockchain is a trustworthy, predictable, verifiable ledger of activity if the work contributed cannot be known as reliably preserved at the time it is submitted, only later in review. This is part of the security of the blockchain. This prevents malicious agents from pushing "valid" blocks, even if they manage to get all the dots in a row. The computational work involved in building the next block is vastly outpaced by the general network; by the time you minted you "attack" block, many many many new blocks were already minted by the network and they won't accept your block. This "consensus" keeps these attacks from happening. However, this still can be attacked. If an agent gained control of the majority of the network computational power, then they can mine blocks faster than the honest miners, thus growing the longest chain and thus control exactly what transactions/blocks are mined. This is referred to as the 51% attack because they become the Consensus. But, using my calculations above, I would need somewhere around ~6*10^6 nVidia 1080 cards all working for me to gain majority in the Ethereum network. So I sleep comfortably at night knowing my ETH is safe.

rtkwe 9y ago

Currently that's solved because transactions aren't added live by every user but instead get batched together and added as a group so individually you just have to wait for some block to pickup your transaction from the pool of available transactions. What you're suggesting could potentially become a problem for a technology if new block propagation takes longer than new block discovery. Today orphan blocks aren't exactly rare there're several a month where two different pools will announce new blocks within a couple seconds of each other and one gets abandoned because it was the later one or because the other block had a new block mined from it faster.

atonse 9y ago
"What you're suggesting could potentially become a problem for a technology if new block propagation takes longer than new block discovery." But that's an inevitability. New block discovery only takes longer over time, and block propagation will only get more complicated as you add more users to chain.
- rtkwe 9y ago
  Luckily every user doesn't require a full copy to submit transactions and use the chain just miners.

hdhzy 9y ago

Well in a distributed system you are in a way always out of sync. The primary achievement of Bitcoin was taking already exiting pieces (hash cash, Merkle trees, signatures etc.) and assembling them in a way that solves consensus in such a system (see [0]). What's interesting is the combination of these pieces. They were known a long time before. Actually I had CS lessons where we were creating virtual currencies with these exact elements long before Bitcoin happened (of course no one got Bitcoin in the process :(... or maybe they did but never told me :) ). How to know Bitcoin better? The best approach is the same as with everything else: reverse engineer and recreate a small coin in your language of choice (chain using hashes plus a simple proof of work using N leading 0s). Remember Satoshi's genius was in assembling pieces in a way nobody thought of, you just need to copy a small version of that. It's not that difficult. For me the most interesting aspect is smart contracts [1]. That's the one underutilized element of the system that is quite versatile. [0]: https://en.m.wikipedia.org/wiki/Byzantine_fault_tolerance#Byzantine_Generals.27_Problem https://en.m.wikipedia.org/wiki/Byzantine_fault_tolerance#By... [1]: https://en.bitcoin.it/wiki/Contract https://en.bitcoin.it/wiki/Contract