Topic: Not Bitcoin XT - page 3. (Read 21893 times)

Rats.  It is so embarassing to be predictable. 

... all this time with demand well below capacity and without a "fee market".  But now the lemonade stand is under a new management, who are is determined to change that...

Yes, those transactions that pay higher fees will go first.

However, the capacity of the network is ~0.8 MB (~1500 trasactions) every 10 minutes.  If there is 1.6 MB of transactions (~3000 tx) in the queue, half of them will miss the next block, even if each of them is paying 1 BTC of fee.  

Fees can only decide which transactions will get served first; no matter how high they are, they cannot reduce the average waiting time -- not even by 1 second.

Moreover, it will not be "a little longer".  If clients (all over the world, independently of each otehr) are issuing 3000 transactions every 10 minutes, at most 1500 of them will be processed in the next block.  The other 1500 will go into the queue, adding to the backlog; and will be confirmed only after the traffic again drops below 1500 tx/block.  Even then, if the demand drops to 1300 tx/block, the the backlog will be cleared only at the rate of 1500 - 1300 = 200 tx/block.  

So, a surge in demand that lasts a couple of hours can take many hours to clear.  A large percentage of the transactions will be delayed until the backlog is clearing. Even if most transactions are paying high fees. The average time to first confirmation, which was ~10 minutes before the jam started, will then jump to several hours.

Even if tools like replace-by-fee (RBF) and child-pays-for-parent (CPFP) are available, they will not reduce the average waiting time by a single second.  If you use RBF to push your transaction to the front of the queue, so that is gets mined in the next block, some other transaction that was expecting to be in that block will be pushed to the second-next one, and another that was to be in the latter will be pushed to the third next block, and so on.  

If your transaction got into the backlog (and there is a significant chance that it will, no matter what fee your smart wallet chose), there will be thousands of other clients, just as impatient and wealthy as you, that will try to get their transactions ahead of yours, using those same smart wallets and fee-adjusting tools too.  So, after you sent your payment for your shuttle flight, or to reserve the last room in that hotel in Paris, you will have to keep checking the queues, increasing the fee of your transaction, maybe for hours -- only to wait until the backlog is clearing, anyway.   With a high probability, you will pay a lot more, only to wait much, much longer.

If the average demand is greater than the capacity, the backlog of unconfirmed transactions will keep growing forever and will never clear.  A fraction of the transactions will fall off the end of the queue, after sitting there for hours or days, and will never be confirmed. The average waiting time will then be infinite.

To summarize:

If the average demand is much less than the capacity, there is no fee market: all transactions that pay minimum fee will be processed in then next 10 minutes.

If the average demand is close to capacity, it will be greater than capacity during peak hours or days; then a growing backlog will form until the demand subsides, and the average waiting time will shoot up from 10 minutes to hours.

If the average demand is greater than the capacity, the backlog of unconfirmed transactions will keep growing forever and will never clear.  The average waiting time will be infinite and some transactions will never confirm.

Any bitcoin developer who wants to see bitcoin develop a "fee market" is either incompetent, or does not care about its users.

I just pointed out that JorgeStolfi mentioned "fee market", not "free market", it seems like you misread that part (original quote below).

WTF ? Free market doesn't mean that the item listed in it is free.

Free market means it's free from control and corruption, and the price is determined by voluntary trading, and not manipulated by central banks or etc.

I hope you guys are not so naive to confuse "free" as in free from corruption, with the other meaning of "free" which is with 0 cost.

RealBitcoin,

Fee != Free

Of course, let's just give up gravity because we feel uncomfortable by not being able to fly. I`m really pissed at gravity, every time i jump out of the window from the 10th floor I cannot fly. Damn you gravity.

Free market is so nasty, let's just have corrupt control maniac oligarchy control the market because that will make it so much better...


You guys don't even realize how ridiculous you sound.

Stolfi's story is: A guy who spends an inordinate amount of his free time following those nerds and libertarians around, pointing at them, and shouting, "You're doing it wrong!", while they just look at him and chuckle.


Oh, I doubt it very much.

God damn I gotta love a lemonade stand worth over 3 billion dollars that's been in the market for almost 6 years ...

Yeeeah.  Right on the head.  I've given up talking to these clowns, but you're right.

Small blocks and high fees are NOT a viable way forward. 

Whenever the demand is greater than the capacity (supply), then the transaction price increases and those who will pay more get their TX executed, those who dont will have to wait a bit more.

Thus you can still send your TX in normal amount of time, you just have to pay more to get it cleared.

And it's not random transactions we talk about, but depending on who pays more. It's an auction market, and who bids more will get priority.

That kind of misinformation isn't useful.

It increases exponentially, not linearly, doubling every 2 years for 20 years. 10 doublings, giving a 1024x increase. On top of the initial 8x increase, making an effective increase of 8192x.

This guy gets it. +1 for making me laugh.

The nodes will drop that low fee transaction from its mempool and since the sender has gone offline, then there is no one to continue to broadcast that transaction and it is forgotten by the network. When the node goes back online, the transaction may have disappeared. If it didn't, then it should be rebroadcasting the transaction every 30 minutes.

Just an ordinary professor.  I cannot do miracles.  If students just repeat any bullshit that they read, without thinking critically, I cannot help them...


The "fee market" idea is nonsense in so many levels that I don't know where to begin. I give up.

Bitcoin's story is "nerds and libertarians learning the basics of how the world works, the hard way".  The "fee market" will be another lesson: "how to manage a lemonade stand."

I read it before and re-read it again now.

First thing, even Gavin said that he thinks this scenario is unlikely.

Anyhow.

The first half of the article is not really relevant because as he says memcache can be limited.

So that leaves the second part which has 3 main arguments:

The first two, are irrelevant because you simply have to pay a high enough fee. A fee market means there is always space if you pay a high enough fee. Thats how markets work. If the blocks start getting full the transactions get dropped of. There will always be space if you can pay a good fee. People of-course, will not attempt transactions with low fees just as nobody attempts to buy a mercades for $5.

Wallets can automatically estimate what fee is necessary.

The third argument; So let says it occurs in the scenario where a transaction was submitted but the fee estimated was too low (busy part of the day maybe). The transaction does not make it in.

Eventually it probably will make it because the market price can drop, and presumably the fee was not too far below the market price.

Replace-by-fee means that the transaction can be resubmitted with a higher fee. A wallet could do that automatically.


This part I can not understand. The wallet can create a new transaction automatically, to replace the old. Why does it have to be stuck?

If a wallet does not have that functionality, the user will have to resubmit it himself and not go to sleep.

If his wallet does not even allow to resubmit transactions, they should consider changing wallets. Some already allow that.

Please see the article I just posted above.

What are you talking about?

Miners can get paid from fees if either fees go up or more people use Bitcoin. With increasing block size limits, fees can both go up (but to a lesser extent) and gain more users thus generating more fees. By keeping the limit, at a certain point, fees can only go up by usership won't have an affect except by perhaps increasing the fees some more. At some point, people will become frustrated with the fees that they just quit.

But a hard fork should never be done in an emergency as that can cause problems. It is best to do them when a problem that has potential to become an emergency has been identified.

I agree. I prefer BIP 100. I am also in favor of action in the near future instead of waiting for this to become an actual problem.

BIP101 does not immediately jump from 8 mb to 8 gb. It increases exponentially over the course of several years.

edit: fixed linearly to exponentially. Thanks dooglus for pointing out my mistake.

Please read the following. It covers the problem with this: https://medium.com/@octskyward/crash-landing-f5cc19908e32

You will need fees at some point to pay miners. Thats the reality.

To change that you need to:
Somehow move/change proof of work for Bitcoins blockchain security
Have some sort of constant inflation (i.e. get rid of 21 million limit).

Otherwise we will need fees, that will add up to millions per day.