Topic: What exactly is the reason bitcoin processes transactions so slowly? (Read 3271 times)

You forgot to add in the time it takes to dig the gold out of where you buried it for safekeeping, and travelling to get to where you are going to make the transaction, and maybe the guy you are meeting is running late, so that would add extra wait time too.

Yes. CHAPS. The UK's RTGS which settles in near real-time, with finality, on the books of the central bank.  Not sure why you put "same day" in quotes... it's near instantaneous.

http://www.bankofengland.co.uk/markets/Documents/paymentsystems/rtgsguide.pdf

However, you put your finger on the problem: there are only sixteen members of CHAPS, which means all other participants must route their transactions through one of those sixteen (and it can take time to get funds in to / out of these institutions). And it is expensive. And it is only open for certain hours.

Which is, of course, why payments for coffee in Starbucks are not settled across it and why almost all payments in GBP are aggregated in BACS/FPS/MC/VISA/etc and settled net in CHAPS at periodic intervals.

The same applies for Fedwire/USD and Target2/EUR, etc, etc.


CHAPS, Fedwire and Target2 are three counterexamples to that claim.  However, they are, I admit, practically irrelevant to this discussion since membership criteria are so strict and they suffer from so many other complexities.

So I think we're in violent agreement:  Bitcoin brings almost all of the hitherto inaccessible benefits of an RTGS to domains where, previously, only DNSs were available

But it does mean that Bitcoin finds itself in a strange hybrid situation -- not quite as fast as a true RTGS to provide assurances of transaction finality yet lacking the netting capabilities that prevent Starbucks payments cluttering up the same ledger that is used for multi-billion dollar inter-bank transfers!

(As for zero confirm transactions, I agree.... for most practical purposes, they are good enough and, on that measure, Bitcoin is almost certainly quicker than anything else out there.)

You're talking about CHAPS, the system that settles "same day" and that's considered good? The one that costs 20-30 pounds per transfer? The one that only operates during office hours?

I'm pretty sure there are no systems that move money irreversibly faster than Bitcoin. And bear in mind, zero confirm transactions work, and can be made even stronger in future.

Senior, I notice that you are burdened by a heavy load of gold coins.  Perhaps I can help you with them?

Gold coins in person.

If you run a bar & grill and a regular pays you, it's good enough when it shows up on blockchain.info.  Of course, this is a person who will eventually forget his wallet and have to go home before he pays a tab, and you will trust him that much anyway.

It seems like, after some study, that most of the crypto currencies act the same here. The ones that generate blocks faster just wait for more blocks to finish to call it confirmed. The ones which generate blocks faster will give you an initial low confidence confirm (inclusion in the first block) faster, but you still need to wait for multiple blocks before there is high enough certainty.

Oatmo

Not quite....  individuals don't have *direct* access to any currency's RTGS but they *can* make use of them indirectly.  e.g. for house purchasing in the UK, it is normal for a purchaser to transfer their deposit over CHAPS to the lawyer acting on their behalf.   The need for near-real-time settlement with finality is key.

The reason I brought this up is because I sense discussions about settlement speed, etc., in Bitcoin are based on a misunderstanding of how other currencies work.

To my mind, the Bitcoin network is an RTGS, rather than a DNS (gross settlement, settlement in "central bank money"*, finality, etc).  However, discussions on this topic talk about use-cases most usually addressed with DNSs (e.g. retail payments, online commerce, etc, etc).

Now, nobody would ever suggest using Fedwire for every credit card transaction in the world but people *are* implicitly suggesting that, should Bitcoin gain widespread adoption, every single transaction (every pint of beer, every pizza, every coffee) would be passed over the Blockchain....   in other words, there is an implicit assumption that Bitcoin (an RTGS) will gain widespread adoption for use-cases currently solved with DNSs.

It's just not obvious to me that this is the right answer.

Sure... a Deferred Net Settlement for Bitcoin using a traditional centralised architecture would be unattractive -- but one could easily envisage an architecture where low-value transactions take place on "auxiliary" side-chains before being periodically aggregated (and hence settled) net on the main BTC chain.   



* Clearly, there is no "central bank" in Bitcoin but the entire network can be seen as analogous to one (it controls the money supply, provides regulation (difficulty adjustments, rejection of invalid transactions, etc), etc, etc).

You need a trusted third party for this.

It's important to point out that this is of zero relevance to any individual, as they are not a bank, and thus have no access to any currency's RTGS system.

This thread: "how I learned to stop worrying and love the ~10 min conf time."

Perhaps there's a terminological problem here.   We need to distinguish between the concepts of clearing, settlement and finality.

Clearing in the context of payments is of most relevance to deferred net settlement systems (e.g. Visa/MC, EURO1, BACS, etc) where it is necessary for somebody to sort out the obligations (who owes what to whom) in order that settlement can then be effected efficiently between the parties.  And as merchants who have experienced chargeback from the Visa and MC networks will attest, "settlement" under these systems does not imply "finality".

However, the Bitcoin network doesn't do any netting: every transaction is settled gross.  As such, it is most analogous to a real-time gross settlement system such as CHAPS/Target2/Fedwire, etc.

In these systems, every transaction is settled gross (like Bitcoin) and is settled, with finality, in central bank money.  This happens in seconds and clearing isn't really a helpful concept/term to use.  Settlement is the key concept here.

Either way, it is clear that the Bitcoin network does *not* provide the level of settlement speed that the major currencies' RTGSs provide.... but it does provide guarantees of finality far quicker than the DNS systems that most individuals rely on for most of their transactions.

Alright, so maybe Bitcoin clears faster than any other financial mechanism ever, then

Unless you are trying to do a cash transaction at a distance. It might take a couple days for the cash to go through the mail, clearly bitcoins clear much faster than that.

To take your point ad absurdum, even cash transactions aren't instantly cleared. In fact, the vast majority are never cleared at all. There's a level of trust applied with a cash transaction that the bills exchanging hands are not counterfeit. If there were to be actual verification (and I don't mean by using a magic marker), then a cash transaction would take even longer to clear than a credit card.

Right.

If you accept bitcoin payments, there are basically two extreme scenarios:
1 - high-value sales such as mining rigs, jewellery etc.: wait for sufficient confirmations before giving away or sending the item.
2 - low-value sales such as pints of beer: accept zero-confirmation transactions (maybe checking that they include sufficient fees to be processed in reasonable time) and assume that nobody would go through the effort to stage a double spend fraud for a pint of beer.

If your scenario has both high value and the need for immediate reliable payment clearance, you most likely got a trust problem anyway, and fast transactions would not solve that.

Onkel Paul

So currently there is no good way to have both fast bitcoin transactions and double spending prevention?

10 min is just a magic number.

There are two things going on, the transaction must propagate through the network, and the transaction must be confirmed as valid. Both credit card and bitcoin transactions propagate almost instantly. For small purchases it could be acceptable to complete the transaction as soon as the bitcoins are seen with 0 confirmations. That is what credit card users do as well.

Bitcoins are confirmed when you get to the number of blocks you deem necessary, most places use 6 confirms, but I have seen 3 or 9 as well. Credit Cards, on the other hand, take much longer to fully confirm; the transaction can be reversed weeks later by the credit card company.

The main factor is decentralization.
You cannot have both; decentralization and quick judgement on who is trying to double spend.
If you choose to go for the quick judgement way, then you will end up in hell. Where you come from, so no worries here.