Montreal scaling Bitcoin workshop recap.

[adamstgBit]

fear of being proven wrong.

They're not wrong. Centralization is already a problem.

ok show me a part of the network that is Centralized and the problem it's causing.

[Peter R]

They have no other arguments any longer.  The only remaining talking point they have is that if we raise the block size limit too much that it will result in centralization (without a clear definition for what this means).

So we don't have 1% orphan rates at 1MB blocks? So the orphan rates would not grow if we had e.g. 8 MB blocks right now? It really seems like they don't really have any arguments. You should really discuss this with other people (not if you're going to be closed-minded) before making slides that are incorrect (hint: Luke-Jr said something about one on IRC if I recall correctly).

Of course orphans would increase if miners published larger blocks [all other variables held constant].  That is what my talk was all about!  It is the risk of having a block orphaned that creates the cost.  Do you see what I mean?  If I'm a miner, why would I bother to make an 8 MB block if it had (e.g.) a 25% chance of being orphaned?  I would only do so if it included enough fees to offset this risk.  This is the essence of my fee market paper.  

There were 155 orphaned blocks in Q1 2015 and 97 in Q2.  The average orphan rate is about 1%.  It is the fact that block propagation is slow that makes it costly for a miner to even attempt to produce an 8 MB block.  I estimate that, given the current network propagation impedance, a miner would only be wise to attempt to publish an 8 MB block if it contained 3 to 6 BTC of fees (due to his increased risk of orphaning).  



As block propagation improves, it gets cheaper for miners to build larger blocks.  There is a natural balance that occurs without the need for a tight limit.  This is how Bitcoin has always worked.  The free market solves the block size problem without centralized intervention.  

source: https://scalingbitcoin.org/papers/feemarket.pdf

[Lauda]

ok show me a part of the network that is Centralized and the problem it's causing.

I never said that it was causing any problems right now. Stop diverting the argument.

If I'm a miner, why would I bother to make an 8 MB block if it had (e.g.) a 25% chance of being orphaned?  I would only do so if it included enough fees to offset this risk.  This is the essence of my fee market paper.  
I estimate that, given the current network propagation impedance, a miner would only be wise to attempt to publish an 8 MB block if it contained 3 to 6 BTC of fees (due to his increased risk of orphaning).  
-snip-

Yes, I'm aware of your talk. You were actually one of the few that I saw on livestream as I did not have enough time. Honestly I do not have time to asses your paper myself, nor am I that interested in it (papers about fees in general). I do have a question though. If there are no incentives for 8 MB blocks (since fees are lower than 3 to 6 BTC ) then why should we implement them at all? If there aren't going to be bigger blocks often (that would fit more transactions - one of the main arguments of big blockists and XT supporters; i.e. need for more room) then it is not worth the risk, or is it?
People need to shift their focus towards making propagation times better rather than explaining potential situations relating to fees. If we had a better way of propagation right now, we would not be having this discussion.

[knight22]

ok show me a part of the network that is Centralized and the problem it's causing.

I never said that it was causing any problems right now. Stop diverting the argument.

They're not wrong. Centralization is already a problem.

wtf is wrong with you?

[Lauda]

ok show me a part of the network that is Centralized and the problem it's causing.

I never said that it was causing any problems right now. Stop diverting the argument.

They're not wrong. Centralization is already a problem.

wtf is wrong with you?

Centralization is a problem != centralization is causing problems. Stop grasping straws trying to undermine my arguments. If you do not want to properly contribute, then do not do so at all.
Clarification update:
Centralization is a problem - In other words, the node count is going down and the mining is centralized via a few pools (China being the one with most hashrate). (bad) Example of centralization causing problems - Nodes are impossible to run on average hardware; i.e. they need a data center (might as well use Google). Decentralization is one of the main feats of Bitcoin. Rhetorical question: Would you feel comfortable if the whole network was running on Amazon and Google cloud services?
Update 2:
No, that is not a datacenter. There is no incentive to run a node, what exactly are you talking about?

[valkir]

People need to shift their focus towards making propagation times better rather than explaining potential situations relating to fees. If we had a better way of propagation right now, we would not be having this discussion.

Totally agree with you. If we find a way to make propagation times better we will than be able to increase the block size. If not, block size increase, in my opinion, with make new nodes harder to exist and then result in a centralisation of the actuel node. There was a talk about that this week end, short and clear. We need to resolve propagation times before talking about block size.

Im curious to know what was said during the dev meet up during roundtable 2 of day 2.

[knight22]

ok show me a part of the network that is Centralized and the problem it's causing.

I never said that it was causing any problems right now. Stop diverting the argument.

They're not wrong. Centralization is already a problem.

wtf is wrong with you?

Centralization is a problem != centralization is causing problems. Stop grasping straws trying to undermine my arguments. If you do not want to properly contribute, then do not do so at all.

Well maybe you should clarify how is this a problem if it's not causing any problem...

First there is no incentives for a mining pool to hijack the network. Second, the Slush and Ghash episodes demonstrated that the market badly react when a pool gets near of 50% hashrate until things rebalance.  I don't see why the actual situation should be considered as a problem based on past market dynamics.  

[jonald_fyookball]

They have no other arguments any longer.  The only remaining talking point they have is that if we raise the block size limit too much that it will result in centralization (without a clear definition for what this means).

So we don't have 1% orphan rates at 1MB blocks? So the orphan rates would not grow if we had e.g. 8 MB blocks right now? It really seems like they don't really have any arguments. You should really discuss this with other people (not if you're going to be closed-minded) before making slides that are incorrect (hint: Luke-Jr said something about one on IRC if I recall correctly).

Of course orphans would increase if miners published larger blocks [all other variables held constant].  That is what my talk was all about!  It is the risk of having a block orphaned that creates the cost.  Do you see what I mean?  If I'm a miner, why would I bother to make an 8 MB block if it had (e.g.) a 25% chance of being orphaned?  I would only do so if it included enough fees to offset this risk.  This is the essence of my fee market paper.  

There were 155 orphaned blocks in Q1 2015 and 97 in Q2.  The average orphan rate is about 1%.  It is the fact that block propagation is slow that makes it costly for a miner to even attempt to produce an 8 MB block.  I estimate that, given the current network propagation impedance, a miner would only be wise to attempt to publish an 8 MB block if it contained 3 to 6 BTC of fees (due to his increased risk of orphaning).  



As block propagation improves, it gets cheaper for miners to build larger blocks.  There is a natural balance that occurs without the need for a tight limit.  This is how Bitcoin has always worked.  The free market solves the block size problem without centralized intervention.  

source: https://scalingbitcoin.org/papers/feemarket.pdf

Very interesting. 

I feel this deserves this deserves its own thread.

[knight22]

People need to shift their focus towards making propagation times better rather than explaining potential situations relating to fees. If we had a better way of propagation right now, we would not be having this discussion.

Totally agree with you. If we find a way to make propagation times better we will than be able to increase the block size. If not, block size increase, in my opinion, with make new nodes harder to exist and then result in a centralisation of the actuel node. There was a talk about that this week end, short and clear. We need to resolve propagation times before talking about block size.

Im curious to know what was said during the dev meet up during roundtable 2 of day 2.

Isn't pruned nodes supposed to help increasing propagation times?

[Peter R]

They have no other arguments any longer.  The only remaining talking point they have is that if we raise the block size limit too much that it will result in centralization (without a clear definition for what this means).

So we don't have 1% orphan rates at 1MB blocks? So the orphan rates would not grow if we had e.g. 8 MB blocks right now? It really seems like they don't really have any arguments. You should really discuss this with other people (not if you're going to be closed-minded) before making slides that are incorrect (hint: Luke-Jr said something about one on IRC if I recall correctly).

Of course orphans would increase if miners published larger blocks [all other variables held constant].  That is what my talk was all about!  It is the risk of having a block orphaned that creates the cost.  Do you see what I mean?  If I'm a miner, why would I bother to make an 8 MB block if it had (e.g.) a 25% chance of being orphaned?  I would only do so if it included enough fees to offset this risk.  This is the essence of my fee market paper.  

There were 155 orphaned blocks in Q1 2015 and 97 in Q2.  The average orphan rate is about 1%.  It is the fact that block propagation is slow that makes it costly for a miner to even attempt to produce an 8 MB block.  I estimate that, given the current network propagation impedance, a miner would only be wise to attempt to publish an 8 MB block if it contained 3 to 6 BTC of fees (due to his increased risk of orphaning).  



As block propagation improves, it gets cheaper for miners to build larger blocks.  There is a natural balance that occurs without the need for a tight limit.  This is how Bitcoin has always worked.  The free market solves the block size problem without centralized intervention.  

source: https://scalingbitcoin.org/papers/feemarket.pdf

Very interesting. 

I feel this deserves this deserves its own thread.

Thanks! 

I think I need to write a more accessible version of my paper--more like my talk but with a bit more depth. 

[LiteCoinGuy]

I really wish you two would have met and found some common ground. Here we have two people who were front row, in person, live action at the workshop that this thread is about, and instead of hearing your stories of who you met, what you saw, and if you think it was a productive with respect to the block size debate at large, we have to endure a bickering back and forth about hurt egos. 

I've been reporting in the uncensored Bitcoin Forum http://bitco.in/forum/threads/gold-collapsing-bitcoin-up.16/page-26

************
This conference has really opened my eyes to what I see as a huge divide in ideology between the small-block camp and the large-block camp.

The small-block camp seem to want to create rules to micromanage every aspect of the system; the large-block camp want the consensus rules to be as simple as possible and allow the market to sort out the details.

Here's a perfect example from maaku:



Here he's arguing with Gavin that propagation time is too long to support 8 MB blocks.  However, just last night he was arguing with me that propagation time is too short to support a fee market.

At first I couldn't understand how they could argue from both sides of most topics: how can they honestly think that there are too many orphans AND not enough orphans?  Well, it's simply that they can't see how the market will come to an equilibrium.  They actually think they need to micromanage everything in order for the system not to break.   This is why orphans can be too high (for safety) but too low (for a fee market) in their minds.

************

thanks for that. and gavin made the right choice to release Bitcoin XT because that is our fallback system.

[valkir]

People need to shift their focus towards making propagation times better rather than explaining potential situations relating to fees. If we had a better way of propagation right now, we would not be having this discussion.

Totally agree with you. If we find a way to make propagation times better we will than be able to increase the block size. If not, block size increase, in my opinion, with make new nodes harder to exist and then result in a centralisation of the actuel node. There was a talk about that this week end, short and clear. We need to resolve propagation times before talking about block size.

Im curious to know what was said during the dev meet up during roundtable 2 of day 2.

Isn't pruned nodes supposed to help increasing propagation times?

Suppose to but still not perfect. The communication with the china and there great firewall make propagation a big problem for big miner.
I was in scaling montreal and I did try to learn more about network problem with Matt Corallo and Rusty Russell and the network problem seems to be the place we should focus before working on bigger block size. A big farm ceo even said he created a node system to be sure to reach more node quickly to eliminate orphan.

[knight22]

Clarification update:
Centralization is a problem - In other words, the node count is going down and the mining is centralized via a few pools (China being the one with most hashrate). (bad) Example of centralization causing problems - Nodes are impossible to run on average hardware; i.e. they need a data center (might as well use Google). Decentralization is one of the main feats of Bitcoin. Rhetorical question: Would you feel comfortable if the whole network was running on Amazon and Google cloud services?

By data center you mean this? https://bitseed.org/

I understand what you mean but I think this fear is overblown. There is an incentives for individuals, businesses and miners to run full nodes at the lowest price possible so the market will come up with cheap solutions.

[valkir]

You should go see Patrick Strateman talk about Initial Block Synchronization.
Show us that nodes will get harder to get the more the size increase.

[itsamemario]

They have no other arguments any longer.  The only remaining talking point they have is that if we raise the block size limit too much that it will result in centralization (without a clear definition for what this means).

So we don't have 1% orphan rates at 1MB blocks? So the orphan rates would not grow if we had e.g. 8 MB blocks right now? It really seems like they don't really have any arguments. You should really discuss this with other people (not if you're going to be closed-minded) before making slides that are incorrect (hint: Luke-Jr said something about one on IRC if I recall correctly).

Of course orphans would increase if miners published larger blocks [all other variables held constant].  That is what my talk was all about!  It is the risk of having a block orphaned that creates the cost.  Do you see what I mean?  If I'm a miner, why would I bother to make an 8 MB block if it had (e.g.) a 25% chance of being orphaned?  I would only do so if it included enough fees to offset this risk.  This is the essence of my fee market paper.  

There were 155 orphaned blocks in Q1 2015 and 97 in Q2.  The average orphan rate is about 1%.  It is the fact that block propagation is slow that makes it costly for a miner to even attempt to produce an 8 MB block.  I estimate that, given the current network propagation impedance, a miner would only be wise to attempt to publish an 8 MB block if it contained 3 to 6 BTC of fees (due to his increased risk of orphaning).  



As block propagation improves, it gets cheaper for miners to build larger blocks.  There is a natural balance that occurs without the need for a tight limit.  This is how Bitcoin has always worked.  The free market solves the block size problem without centralized intervention.  

source: https://scalingbitcoin.org/papers/feemarket.pdf

Very interesting. 

I feel this deserves this deserves its own thread.

I feel like you should do your research, or at least provide valid information behind your negatvie trust. I am not an account farmer, and now my reputation on here is wrecked because you have decided to lie on my trust.  In addition, it's ironic that I cannot provide negative trust back although you are clearly not trustworthy as you have lied in regards to my reputation. This place is gettin bad theymos, you should probably chill on the trigger pulling. I am no account farmer, and have done nothing wrong except create a post to gauge the communities input on your gold fist.

[brg444]

BIP101 solves the scalability issue once and for all and is currently the only "up and running" implementation.

This proposition has never been more dead than after this weekend. It is now certain that no implementation proposing a set schedule for anything more than 1 or 2 years will not be considered.

XT, by the way, has its last remaining credibility shot after Mike's notable absence.

[knight22]

BIP101 solves the scalability issue once and for all and is currently the only "up and running" implementation.

This proposition has never been more dead after this weekend.

What is dead is Core releavence.

[brg444]

BIP101 solves the scalability issue once and for all and is currently the only "up and running" implementation.

This proposition has never been more dead after this weekend.

What is dead is Core releavence.

 

Core is the motor of growth in Bitcoin.

No one takes you seriously trolled, you'd have gotten laughed out of the room saying such idiocy during the conference

[knight22]

BIP101 solves the scalability issue once and for all and is currently the only "up and running" implementation.

This proposition has never been more dead after this weekend.

What is dead is Core releavence.

 

Core is the motor of growth in Bitcoin.

No one takes you seriously trolled, you'd have gotten laughed out of the room saying such idiocy during the conference

Sorry but Core doesn't allow any growth at all. Hence my point.

[lunarboy]

 The level of hostile emotional appeal and logical fallacy in this thread is astounding.

Why is it so difficult for a group of intelligent people to discuss ideas on their own merit?


there is no tax on being respectful.