4 ms·

For POS transactions though don't you want it to be instant? 3-12 blocks is half a hour to two hours wait. The chart shows 30 satoshis/byte to take 40 minutes

For POS transactions though don't you want it to be instant? 3-12 blocks is half a hour to two hours wait. The chart shows 30 satoshis/byte to take 40 minutes to a 19 hours total wait.

gruez 9y ago
i think that's strawman at this point. even if bitcoin had 1GB block sizes, bringing the wait time to 1 block (10 minutes), it's still not going to be acceptable for POS. it's either going to be on-chain with no confirmations, or through layer2 protocols.
- xorcist 9y ago
  Orphan blocks are common, one block has never been considered secure. Early in Bitcoin, a best practice emerged to consider six blocks secure. So that takes an hour, best case. (Given the insane hash power now maybe that standard can be lowered, but on the other hand it's not common to get a spot in the next block anymore.) So point of sale was never a good fit for Bitcoin. What most real world merchants do when the amount is small enough is to eat the risk. Even if it's easy to double spend in practice most customers won't bother for a can of coke.
thephyber 9y ago
I think it's important to consider the alternative. In Zimbabwe, you could lose a significant percentage of your cash's value if you delay a purchase. Taking a risk that the other side of the transaction might not be as big of a problem if you can look at their account balance before the transaction begins. Sure, you risk getting caught in a double-spend that is rejected from the blockchain, but if the alternative is to hold onto cash or to lose out on a transaction in a tough economy, you might be willing to take on that risk.
nikcub 9y ago
Accepting zero confirmation transactions at POS is a different problem to fees (which are atm admitantly high but that should improve) Merchant picks a solution based on who they integrate with - something like BitGo Instant[0], GreenAddress[1] or an analytics/fraud API provider that adds a risk score to a tx no different to a credit card transaction [0] https://www.bitgo.com/solutions#instant https://www.bitgo.com/solutions#instant
SilasX 9y ago
You definitely do want it to be instant, but this isn't quite an apples-to-apples comparison. That two-hour figure is the amount to get the transaction to the point of irreversibility so it can be re-spent without worry. The equivalent figure for credit cards is ~45 days and checks can take a week to clear. But you're right, making transactions on chain is not going to be practical in an untrusted environment and it still depends on intermediaries.
- uncoder0 9y ago
  What you're describing sounds like zero confirmation transactions which bitcoin used to support before RBF. You can't reliably take zero-conf in btc anymore though. So you have to wait for settlement to turn over goods or services.
  - xorcist 9y ago
    Zero conf was never secure. Many did demonstrate double spends long before RBF was thought of. Still, that argument was made against RBF and was enough of a hindrade to consensus that RBF was made opt-in. And it still is. So you are absolutely free to accept zero conf transactions, and take the risk, as long as you don't accept those who flag for RBF. Most wallets won't even make them by default (which in my opinion is stupid with the fees we have today). The argument for a golden age before RBF isn't a good one. It was never reliable. And what little reliability it had is unchanged.

For POS transactions though don't you want it to be instant? 3-12 blocks is half a hour to two hours wait. The chart shows 30 satoshis/byte to take 40 minutes to a 19 hours total wait.

gruez 9y ago

i think that's strawman at this point. even if bitcoin had 1GB block sizes, bringing the wait time to 1 block (10 minutes), it's still not going to be acceptable for POS. it's either going to be on-chain with no confirmations, or through layer2 protocols.

xorcist 9y ago
Orphan blocks are common, one block has never been considered secure. Early in Bitcoin, a best practice emerged to consider six blocks secure. So that takes an hour, best case. (Given the insane hash power now maybe that standard can be lowered, but on the other hand it's not common to get a spot in the next block anymore.) So point of sale was never a good fit for Bitcoin. What most real world merchants do when the amount is small enough is to eat the risk. Even if it's easy to double spend in practice most customers won't bother for a can of coke.

thephyber 9y ago

I think it's important to consider the alternative. In Zimbabwe, you could lose a significant percentage of your cash's value if you delay a purchase. Taking a risk that the other side of the transaction might not be as big of a problem if you can look at their account balance before the transaction begins. Sure, you risk getting caught in a double-spend that is rejected from the blockchain, but if the alternative is to hold onto cash or to lose out on a transaction in a tough economy, you might be willing to take on that risk.

nikcub 9y ago

Accepting zero confirmation transactions at POS is a different problem to fees (which are atm admitantly high but that should improve) Merchant picks a solution based on who they integrate with - something like BitGo Instant[0], GreenAddress[1] or an analytics/fraud API provider that adds a risk score to a tx no different to a credit card transaction [0] https://www.bitgo.com/solutions#instant https://www.bitgo.com/solutions#instant

SilasX 9y ago

You definitely do want it to be instant, but this isn't quite an apples-to-apples comparison. That two-hour figure is the amount to get the transaction to the point of irreversibility so it can be re-spent without worry. The equivalent figure for credit cards is ~45 days and checks can take a week to clear. But you're right, making transactions on chain is not going to be practical in an untrusted environment and it still depends on intermediaries.

uncoder0 9y ago
What you're describing sounds like zero confirmation transactions which bitcoin used to support before RBF. You can't reliably take zero-conf in btc anymore though. So you have to wait for settlement to turn over goods or services.
- xorcist 9y ago
  Zero conf was never secure. Many did demonstrate double spends long before RBF was thought of. Still, that argument was made against RBF and was enough of a hindrade to consensus that RBF was made opt-in. And it still is. So you are absolutely free to accept zero conf transactions, and take the risk, as long as you don't accept those who flag for RBF. Most wallets won't even make them by default (which in my opinion is stupid with the fees we have today). The argument for a golden age before RBF isn't a good one. It was never reliable. And what little reliability it had is unchanged.