Faster transactions?

[bitdude]

Hi,

has there been a serious discussion on faster transactions in Bitcoin? Is a consensus that they are not needed or are there any technical problems that prevent this to be implemented?

What do I mean by faster transactions? From what I learned recently, it seems that to create an unconfirmed transaction out of nothing is possible and hence accepting any amount of payments with 0 confirmation is extremely risky. Rely on 1 confirmation seems to be OK for amounts under current block miner's revenue (say 25 BTC ~ 15000 USD), maybe the calculation should be made only a portion of a block revenue. Anyway, I stick with 15000 USD for this moment. 15000 USD sounds like a lot when we need something like 5 USD to confirm.

So, if we can confirm up to 15000 USD in 10 minutes, it might be interesting if we can confirm up to 1500 USD in 1 minute.

How to do that? We know that Namecoin can be mined simultaneously, so based on that knowledge I assume we can create a new blockchain similar to Bitcoin, with 10x faster block generation, meaning that we would accept as solution any hash that matches 10x smaller difficulty than Bitcoin's. But no new independent coins. We just use this new blockchain for confirmations of small transactions that we want to go faster. We can have a client that sends everything up to e.g. 1 BTC (600 USD now) using the faster chain.

Now my current knowledge of Bitcoin protocol does not allow me to go further in technical details on how to join faster chain coins with slower blockchain, but I believe this is just a technical and design obstacle that could be solved.

My questions:

1) To anyone with greater technical knowledge of Bitcoin protocol - is it really solvable as I believe or is there any super hard problem that would prevent that?

2) If this is possible and can be done, why is this not interesting? (I ask like that since I assume that if it was, someone would already came with that)

[franky1]

for store of wealth stick with bitcoin. for fast transactions, lets get merchants using litecoin for faster confirm times.

imagine bitcoin as the bank account that does wire transfers slowly and litecoin as the credit card..

[bitdude]

for store of wealth stick with bitcoin. for fast transactions, lets get merchants using litecoin for faster confirm times.

imagine bitcoin as the bank account that does wire transfers slowly and litecoin as the credit card..

Thanks for reply, I am aware of Litecoin and other coins, but that was not really my question. It was more about "can we make Bitcoin faster just with some kind of second blockchain?" and "is this question interesting?".

[Dalmar]

for store of wealth stick with bitcoin. for fast transactions, lets get merchants using litecoin for faster confirm times.

imagine bitcoin as the bank account that does wire transfers slowly and litecoin as the credit card..

That's a very risky path to take for bitcoin's future... Eventually average joes will just use Litecoin instead for everything as they get used to it, better not allow competition to get that close.

[DeathAndTaxes]

From what I learned recently, it seems that to create an unconfirmed transaction out of nothing is possible

This is a false statement.  Miners are not the only security mechanism in the Bitcoin network.  All nodes (every single one of the 100,000+) on the network independently validate transactions before they include them in the memory pool or relay them.

A transaction spending "nothing" would instantly be dropped by all peers on the network.  You would never even see it in your client.  It is very likely that your node wouldn't even see the tx to drop it, as every node between you and the sender dropped it and thus never relayed it on.

As for changing Bitcoin target interval.  In theory any change to Bitcoin is possible, as a practical matter it will never happen as it requires a consensus and a hard fork.  It is unknown if a shorter interval will be optimal as the network, transactions volume, and block sizes gets larger.  There is no free lunch.  Shorter block interval means higher orphan rates.  That means less effective security and the orphan cost rises.  With a higher orphan cost miners will be reluctant to make blocks larger, tx fees will rise, and lower paying or free txs will wait much longer to be included in a block.

[Melbustus]

Transaction security is a function of time-since-transaction and hash-power. You'll need more confs in litecoin to get the same security as fewer confs in bitcoin. Also, there are very few use-cases in which you're happy waiting 2.5mins for a conf, but not 10. Furthermore, these are statistical averages. A coin with a 30-second conf will still take many minutes sometimes.

Regardless, shorter confs also lead to more orphan blocks and more chain re-orgs. If your transaction finds itself in an orphaned block, that's just as much of a problem for the merchant as a double spend.

It's a balance/tradeoff... You can't just reduce the conf time and get a coin with "better" properties for free.

So, I'd say just use bitcoin with double-spend detection, and be fine with 0-conf transactions for small amounts of money. If you're dealing with a significant amount of money in *any* coin, you'll want to wait for many confs (6 for bitcoin, more for shorter-conf coins).


TL;DR: everything's a trade-off. Shorter-conf is not simply "better".

[bitdude]

From what I learned recently, it seems that to create an unconfirmed transaction out of nothing is possible

This is a false statement.  Miners are not the only security mechanism in the Bitcoin network.  All nodes (every single one of the 100,000+) on the network independently validate transactions before they include them in the memory pool or relay them.

A transaction spending "nothing" would instantly be dropped by all peers on the network.  You would never even see it in your client.  It is very likely that your node wouldn't even see the tx to drop it, as every node between you and the sender dropped it and thus never relayed it on.

As for changing Bitcoin target interval.  In theory any change to Bitcoin is possible, as a practical matter it will never happen as it requires a consensus and a hard fork.  It is unknown if a shorter interval will be optimal as the network, transactions volume, and block sizes gets larger.  There is no free lunch.  Shorter block interval means higher orphan rates.  That means less effective security and the orphan cost rises.  With a higher orphan cost miners will be reluctant to make blocks larger, tx fees will rise, and lower paying or free txs will wait much longer to be included in a block.

Thanks DAT! That is very interesting that you say my initial claim is false. In my other topic that I opened some days ago, I had a reply that stated

it's technically possible to create orders to send BTC from address that is empty

of course that this is just an opinion of an anonymous gentleman, so nothing I could rely on, but since then I have been a witness of ASICMINER's dividend being paid from an empty address and it took like 24+ hours for a first confirmation for that transaction to appear and that was because the address was empty and it took a while until it was funded. This tells me that it was indeed possible to create a transaction spending BTC from address with not enough balance.

Is this not the case that you are referring to?

[bitdude]

Just to stress it out - I am not talking about changing properties of Bitcoin blockchain. I was talking of creating a second blockchain to work in paralel with Bitcoin but without introducing new coins. That could be mined simultaneously and thus guarantee the precise security as wanted (e.g. 10 % of Bitcoin's). I was more talking about crypto algorithms being used to ensure that if we spend something in the faster chain, we will see that transaction later in the slower chain.

I am not really talking about litecoin, any other altcoin, new altcoin etc. Just Bitcoin with 2 different speed blockchains, the original one, very secure, "as is today"; and the faster one, for smaller transactions (up to 500 or 1000 USD) that would get 1 confirmation every 1 minute on average.

[Joshuar]

Litecoin will fail, 2.5minutes for a transaction is an absurd amount of time, there are already PoS coins that don't require mining(so no orphaned blocks) that do/will do transactions in a few seconds, just like real credit card companies.
You guys need to move on.

PCE

[DeathAndTaxes]

From what I learned recently, it seems that to create an unconfirmed transaction out of nothing is possible

This is a false statement.  Miners are not the only security mechanism in the Bitcoin network.  All nodes (every single one of the 100,000+) on the network independently validate transactions before they include them in the memory pool or relay them.

A transaction spending "nothing" would instantly be dropped by all peers on the network.  You would never even see it in your client.  It is very likely that your node wouldn't even see the tx to drop it, as every node between you and the sender dropped it and thus never relayed it on.

As for changing Bitcoin target interval.  In theory any change to Bitcoin is possible, as a practical matter it will never happen as it requires a consensus and a hard fork.  It is unknown if a shorter interval will be optimal as the network, transactions volume, and block sizes gets larger.  There is no free lunch.  Shorter block interval means higher orphan rates.  That means less effective security and the orphan cost rises.  With a higher orphan cost miners will be reluctant to make blocks larger, tx fees will rise, and lower paying or free txs will wait much longer to be included in a block.

Thanks DAT! That is very interesting that you say my initial claim is false. In my other topic that I opened some days ago, I had a reply that stated

it's technically possible to create orders to send BTC from address that is empty

of course that this is just an opinion of an anonymous gentleman, so nothing I could rely on, but since then I have been a witness of ASICMINER's dividend being paid from an empty address and it took like 24+ hours for a first confirmation for that transaction to appear and that was because the address was empty and it took a while until it was funded. This tells me that it was indeed possible to create a transaction spending BTC from address with not enough balance.

Is this not the case that you are referring to?

Bitcoin doesn't work on the concept of balances.  All transactions in their input reference a specific unique unspent output (by tx id and index) on the network.  That output either exists or it doesn't.  If it doesn't the transaction is simply invalid.

[DeathAndTaxes]

I am not really talking about litecoin, any other altcoin, new altcoin etc. Just Bitcoin with 2 different speed blockchains, the original one, very secure, "as is today"; and the faster one, for smaller transactions (up to 500 or 1000 USD) that would get 1 confirmation every 1 minute on average.

That would not provide any security.  Just because the tx is confirmed in the "second blockchain" doesn't mean the attacker can't still double spend it in the "main blockchain".  To enforce such requirement would mean a radical change of the Bitcoin protocol, and a change like this is simply not going to happen.

[bitdude]

From what I learned recently, it seems that to create an unconfirmed transaction out of nothing is possible

This is a false statement.  Miners are not the only security mechanism in the Bitcoin network.  All nodes (every single one of the 100,000+) on the network independently validate transactions before they include them in the memory pool or relay them.

A transaction spending "nothing" would instantly be dropped by all peers on the network.  You would never even see it in your client.  It is very likely that your node wouldn't even see the tx to drop it, as every node between you and the sender dropped it and thus never relayed it on.

As for changing Bitcoin target interval.  In theory any change to Bitcoin is possible, as a practical matter it will never happen as it requires a consensus and a hard fork.  It is unknown if a shorter interval will be optimal as the network, transactions volume, and block sizes gets larger.  There is no free lunch.  Shorter block interval means higher orphan rates.  That means less effective security and the orphan cost rises.  With a higher orphan cost miners will be reluctant to make blocks larger, tx fees will rise, and lower paying or free txs will wait much longer to be included in a block.

Thanks DAT! That is very interesting that you say my initial claim is false. In my other topic that I opened some days ago, I had a reply that stated

it's technically possible to create orders to send BTC from address that is empty

of course that this is just an opinion of an anonymous gentleman, so nothing I could rely on, but since then I have been a witness of ASICMINER's dividend being paid from an empty address and it took like 24+ hours for a first confirmation for that transaction to appear and that was because the address was empty and it took a while until it was funded. This tells me that it was indeed possible to create a transaction spending BTC from address with not enough balance.

Is this not the case that you are referring to?

Bitcoin doesn't work on the concept of balances.  All transactions in their input reference a specific unique unspent output (by tx id and index) on the network.  That output either exists or it doesn't.  If it doesn't the transaction is simply invalid.

In other words, there had to be a chain of transactions that guaranteed that that address will be funded, otherwise that transaction would not be accepted by nodes. Is this right? Thank you!

[bitdude]

I am not really talking about litecoin, any other altcoin, new altcoin etc. Just Bitcoin with 2 different speed blockchains, the original one, very secure, "as is today"; and the faster one, for smaller transactions (up to 500 or 1000 USD) that would get 1 confirmation every 1 minute on average.

That would not provide any security.  Just because the tx is confirmed in the "second blockchain" doesn't mean the attacker can't still double spend it in the "main blockchain".  To enforce such requirement would mean a radical change of the Bitcoin protocol, and a change like this is simply not going to happen.

OK, so you are saying that an answer to my question 1) is in fact NO - No, unless there are significant changes in the Bitcoin itself.

[DeathAndTaxes]

In other words, there had to be a chain of transactions that guaranteed that that address will be funded, otherwise that transaction would not be accepted by nodes. Is this right? Thank you!

Using the term "address funded" is still incorrect but yes.  However understand that a 0-confirm tx can be double spent (in theory it is harder to accomplish than most people imagine) and a 0-conifrm tx which relies on an unconfirmed tx for its input is doubly risky.  If the parent tx is double spent, fails to confirm (not included in a block due to insufficient fees), or has its tx id mutated then the child tx will never confirm (it now has an invalid input).

[DeathAndTaxes]

OK, so you are saying that an answer to my question 1) is in fact NO - No, unless there are significant changes in the Bitcoin itself.

Correct.  The rules for the "main chain" would need to be modified to enforce using blocks from the "secondary chain" for the purposes of validation.   Still even if those changes were possible you don't really gain anything from just using a smaller block interval on the main chain.  It is a lot of complexity to just reinvent a network with a smaller block interval.

[mmouse]

I just stumbled over this post while waiting for about 40 minutes now to get my $50 bitcoin transfer confirmed (just three confirmations needed, blockchain, go on, please, please...)

And this is not the first time I'm waiting. For me this clearly tells, that bitcoin isn't "ready for primetime" yet. It works well as a proof of concept, and beyond that. But imagine normal people doing normal daily payments with bitcoin. It's too slow, and besides that the whole blockchain mechanism certainly couldn't handle the amount of transactions if John Doe starts using bitcoin.

So the idea of having a merged second, considerably faster blockchain for small transactions seems quite useful to me. Perhaps you could even do multiple second chains somehow, to be able to handle enough transactions when usage keeps increasing.

And contrary to the proposed "just use another (Lite|Vert|Doge|Fast|whatever) coin" solution this mechanism should be included in the bitcoin wallet and be fully transparent to users. No mainstreet merchant will implement payment for 10 different coins or so. It should be just a on-click-bitcoin-payment and thats it.


Honestly, if you really tell people to use another coin, they will do exactly that in the long run. And there is no reason, why people should use bitcoin then at all. The "storage of wealth" argument is plainly wrong, as almost all other coins can be used for that, too. So people will store their assets in exactly the same coin they will use for daily payment. If on some day in the future this isn't bitcoin, it will get its place in history books as a nice proof of concept and will have no value any more.

Like the OP, I really think that bitcoin has a problem here. But as far as I can tell the big bitcoin buddys aren't interested in a solution (nor even in acknowleding the problem), though. We'll see, how this influences bitcoins future.

Just my 2 cent.

[franky1]

Just to stress it out - I am not talking about changing properties of Bitcoin blockchain. I was talking of creating a second blockchain to work in paralel with Bitcoin but without introducing new coins.

a scond blockchain is a new coin.

"coins" are not objects that sit on a blockchain.. the blockchain is the coin.. its just a ledger.

imagine it like a book with entries wrote in it. the book is bitcoin. making a new book, is making a new coin.. hense my use of mentioning litecoin.

swapping value between book 1 and book 2 is what exchanges do. so imagine you store your 100btc in bitcoin, and each day you move 0.1btc of value in exchange for.. well lets not use litecoin.. lets use another altcoin with 21mill cap, where coins are swapped 1for1. but this altcoin has a 2 minute blocktime.
(the bank/credit card anology)

you can even call this altcoin, bitcoin 2.0. but the point being, a separate ledger(blockchain) is a new coin

[bitdude]

I just stumbled over this post while waiting for about 40 minutes now to get my $50 bitcoin transfer confirmed (just three confirmations needed, blockchain, go on, please, please...)

And this is not the first time I'm waiting. For me this clearly tells, that bitcoin isn't "ready for primetime" yet. It works well as a proof of concept, and beyond that. But imagine normal people doing normal daily payments with bitcoin. It's too slow, and besides that the whole blockchain mechanism certainly couldn't handle the amount of transactions if John Doe starts using bitcoin.

So the idea of having a merged second, considerably faster blockchain for small transactions seems quite useful to me. Perhaps you could even do multiple second chains somehow, to be able to handle enough transactions when usage keeps increasing.

And contrary to the proposed "just use another (Lite|Vert|Doge|Fast|whatever) coin" solution this mechanism should be included in the bitcoin wallet and be fully transparent to users. No mainstreet merchant will implement payment for 10 different coins or so. It should be just a on-click-bitcoin-payment and thats it.


Honestly, if you really tell people to use another coin, they will do exactly that in the long run. And there is no reason, why people should use bitcoin then at all. The "storage of wealth" argument is plainly wrong, as almost all other coins can be used for that, too. So people will store their assets in exactly the same coin they will use for daily payment. If on some day in the future this isn't bitcoin, it will get its place in history books as a nice proof of concept and will have no value any more.

Like the OP, I really think that bitcoin has a problem here. But as far as I can tell the big bitcoin buddys aren't interested in a solution (nor even in acknowleding the problem), though. We'll see, how this influences bitcoins future.

Just my 2 cent.

Partially agree, but there exist solutions today for fast Bitcoin payments. Just saw a video how normal people buying normal food from normal seller using Bitcoins - they just scan QR code and that's it. Then I found this question on Stackoverflow that this is being done by relying just on existence of the transaction itself, without any confirmation. And that is "credit card fast". The thing is that the provider of this probably does accept a possibility of misuse and still goes OK since its fees are somehow good enough to cover those few cases. The seller is not in danger then and thus the whole concept is perfectly usable for common people.

Alternatively, you can have various "prepaid" products that would work "credit card fast" too and without even having a danger of being misused. Of course the trade-off here is to stick with a particular entity on both buyer and seller side.

And finally, Bitcoin is still extremely usable on the "bank to bank" level. Here you have oversees payments in matter of hours instead of 24hrs-10days banking system.

So I do not thing that the super-fast feature is crucial, I just say I would be interested in it if it is implemented somehow. I would even took the risk of hard fork for this feature. But that is just only me of course

[bitdude]

Just to stress it out - I am not talking about changing properties of Bitcoin blockchain. I was talking of creating a second blockchain to work in paralel with Bitcoin but without introducing new coins.

a scond blockchain is a new coin.

"coins" are not objects that sit on a blockchain.. the blockchain is the coin.. its just a ledger.

imagine it like a book with entries wrote in it. the book is bitcoin. making a new book, is making a new coin.. hense my use of mentioning litecoin.

swapping value between book 1 and book 2 is what exchanges do. so imagine you store your 100btc in bitcoin, and each day you move 0.1btc of value in exchange for.. well lets not use litecoin.. lets use another altcoin with 21mill cap, where coins are swapped 1for1. but this altcoin has a 2 minute blocktime.
(the bank/credit card anology)

you can even call this altcoin, bitcoin 2.0. but the point being, a separate ledger(blockchain) is a new coin

I would disagree and my argument would be that this is a matter of terminology and definitions. It might look like a coin, it might work as a coin, but unless it can be used as a coin, I would not call it a coin. For example, we can guarantee, that people can not get any coins from this fast blockchain to their wallets. We can set miner revenue to 0, fees to 0 (but not zero the actual BTC fees). Then really if everyone has no coins, can you call it a coin? Maybe by "no coin, still coin", but from a practical point of view, I would not call it coin, just a technical tool to achieve something.

[mmouse]

I would disagree and my argument would be that this is a matter of terminology and definitions. It might look like a coin, it might work as a coin, but unless it can be used as a coin, I would not call it a coin. For example, we can guarantee, that people can not get any coins from this fast blockchain to their wallets. We can set miner revenue to 0, fees to 0 (but not zero the actual BTC fees). Then really if everyone has no coins, can you call it a coin? Maybe by "no coin, still coin", but from a practical point of view, I would not call it coin, just a technical tool to achieve something.

What comes to mind is something like the share chain which p2pool uses to generate shares for internal payment distribution and blocks for the bitcoin blockchain at the same time. Probably some mechanism like that could be used.