Topic: LN+segwit vs big blocks, levels of centralization. - page 8. (Read 8897 times)

You can only do this if you delegate authority and trust, in other words, if you build hierarchies with power and enforced trust.  The DNS system is a hierarchical system, where the top nodes have a lot of power ; but in a no-double-spend system, this power becomes financial.  If you need a hierarchy of trust that you will get the correct information that Jack, who is trying to pay you, has NOT double-spent, how can you know that this hierarchy is not colluding to make you think that indeed, he didn't, while in fact he did ?
How can you be sure that one is not HIDING this former spending of his, the time that you accept the payment ? 
How can you check that nobody is putting more coins in circulation than is officially announced ?
If you have to trust specific hierarchical entities to tell you about that, the decentralized and trustless system is out of the window.

The whole problem of a truly decentralized and trustless system, is that you are not to depend on ANYBODY to be able to check the validity of the payment one proposes you, and the amount of coins in circulation.  It means that anyone, at any time, must be able to check this independently if he wants to.  This is not possible if there is a hierarchical system in place, because in such a system, you are DEPENDING on these centralized authorities, that can tell you whatever they want.

Yes, you can think of a system with different hierarchies of COINS, where you have a master coin that is the reserve currency of master nodes, who each of them are in a constant exchange rate with sub-coins of a different nature which can themselves be the reserve currencies of still other nodes with sub-sub coins.    However, if there are random payments from sub-sub-coin A of subcoin B, to owners of sub-sub-coin C of subcoin D, then, if you want this to be trustless and distributed, the users still need to have all these chains, to be able to check the right-to-spend of coin A, the right to exchange to coin B, the right to exchange to the master coin, the right to exchange of coin D and the right to exchange of coin C.  Yes, you might think that you don't have to bother about subcoins F, H, J etc.... but even that is not true, because you want to check their quantities in circulation.

So in the end, if you want this to be trustless, this is just a different way of organizing the transactions, but in the end, you have to know all of them, if you want to check the total liquidity.

I have been trying these kinds of things for quite a while, and I'm coming to the conclusion that there is no real way to have a system that is truly decentralized and trustless, and at the same time, scales without having an increasing burden on the individual user, who has the choice between delegating more and more trust to central authorities, or having more and more technical costs.

Well, there IS such a system, which I'm favourable for, but I'm not sure it is stable: that is: many small *independent* currencies, with floating exchange rates, and connected through decentralized exchanges.  But my fear is that speculative forces will put a hierarchy into these currencies, bringing us back to the current situation.  Nevertheless, at least, that is a system where the burden per user doesn't increase with adoption: a user decides to use just a few small crypto currencies.  If he needs to pay another user, he has to find a way through exchanges to get his coins, with several intermediaries, converted into the tokens of his counter party.  That's clumsy, but at least, it scales, because the number of steps is logarithmic with the number of users.

You see, this kind of evading dogmatic answer is typical.  If *theoretically* non-mining nodes matter, and have power, you must be able to argument against my Gedanken experiment, by indicating how non-mining nodes, in that case, can enforce the 1 MB rule against miners and users.  If you can't indicate how that would happen, you can only admit that my Gedanken experiment has no logical failure, and hence, that even 99% of the non-mining nodes cannot impose anything, from the moment that the miners agree amongst themselves, and the users want to continue to transact.

The nice thing of that Gedanken experiment is that it separates the "user power" in the market, and the "non-mining node power", which is most of the time, like in your reply, confused, to attribute the power of the users in the market, to the non-mining nodes.  By clearly separating them, you can pinpoint where the power resides, and it is NOT with the non-mining nodes, unless you can technically explain how my example fails.


Well, your argument of authority fails, because Satoshi was of my opinion in fact.... But again, rhetoric arguments like those (authority) have no logical meaning.  Tell me simply how, in a hypothetical situation, where there is a backbone of miners, that essentially ALL agree upon a 2 MB block, and users, that only want to be able to transact, and exchanges, that only want to allow people to transact and get their fee on it, how 99% of the "non-mining node army" that would have decided to keep 1 MB blocks, is going to enforce its rule. 

If you can find a way, then you are right, these nodes can keep the users and the miners in check, and are the true guardians of the protocol.  If you have to admit that users will happily continue transacting with their light wallets, connected directly to the miners, who can happily continue to build the block chain, and all the disgruntled non-mining nodes coming to a halt and nobody is caring, then you must admit that they don't serve any power purpose, and certainly cannot keep in check the miners and the users.



Which he was.  The error he committed, if his purpose was to have a decentralized development (which is not sure), to make a heir to his central power: Core.

Sidechains are like black accountability/shadow banking like keeping track of assets motion on a sidechains in a private network, the motions on the sidechains are not useable from main chain until they are "settled down".

I guess there can be different approaches

After all, we are discussing whether it is possible to make a payment network internally scalable conceptually and how to implement that approach in practice if it is proved possible, right? If these sidechains cannot be accessed from outside, they cannot be called sidechains at all, as I see it. If what you say were true, that would essentially mean that they should remain impenetrable from other such sidechains as well. But that, in its turn, would effectively mean different networks, not "subsets of the whole set", i.e. not integral parts of the payment network

The pb with sidechains is the address they manipulate cant be accessed from outside, operation on their data cannot be "routed" to them from outside, and they operate on whole different chain that has no real indexing outside of the "subnetwork".

now your just trying to make fake assumptions to pretend thats how things work

nodes matter, always have. they keep the pools inline. without the nodes then pools would collude and change the rules every day and users/merchants wont have choice.

yes some/alot of users dont care and will just use lite wallets and be sheep followers.

but trying to pretend nodes dont matter such as trying to make it sound like people should just shut down their node because they dont matter and instead just run a lite wallet. is sounding more like a strategy which you want to cause centralisation

sorry but those that have been around longer than you know how the network works and will continue running nodes as the symbiotic relationship to keep pools inline.

we dont want to centralise the network.
only core do by trying to bypass nodes (which has failed so far)
only core do by trying to by setting up their desire of a tier network structure rather than the peer network structure that has and currently still exists 2009-2017

sorry but nodes matter.. and yes it can cause changes in the network to not simply happen within a month.. but guess what. that should be a good thing. it keeps everyone inline

if the community say no / abstain. then the devs should ask the community (nodes) what features should be added that would make the community happy.. not find backdoors to bypass and try pushing through changes that can cause issues/dont fulfil promises and end up centralising the network.

put it this way. if we all stopped running nodes and did let one team of devs have the ability to change rules just by bribing pools with free all inclusive weekends like core have done (round table weekends at 'exotic' locations) then what would stop nefarious devs doing the same

P.S
even back in 2009-2010 satoshi knew there were many different node implementations otherwise he would have just implemented the 2mb in block 210,000. he knew it had to be a consensus event of all the nodes agreeing to upgrade the patch ready for block 210,000

he disappeared because some people started to think that he was a lord and god of bitcoin. so left to let nodes independently work and devs to independently make their own brands. and rely on consensus of the community decide when things should change.

P.S core should not be the only brand running/controlling.. pools should not be the only power house.
for good diverse decentralised peer network we should remain with the diverse nodes and then make devs ASK "whats needed" not "do this or else codebomb"

As I understand, it is only a matter of terminology

I'm not very familiar with the approach that sidechains are using (so bear with me), but that is what they seem to be doing, i.e. dividing the payment network into somewhat independent segments, each basically with its own blockchain. So this system should operate in a more or less same way as IP networks are organized, i.e. payments ("IP packets") won't go outside a subchain unless they are made for wallets beyond the scope of addresses of this particular subchain. Payment channels work essentially along the same path only at a higher level, i.e. the payment network is structured not into sidechains but by payment nodes which are made known and then used by users

I'm trying to make see that no "avoid double spend" system can be like that, simply because if you want it to be trustless and decentralized, an individual user needs to KNOW, and hence, needs to be able to check for himself:

1) the rights to spend of every other user he could potentially receive coins from
2) the rights of creation/total balance of the system

and this knowledge grows with the network size, so there's no escaping for this user to have to learn more, update more, check more as the network grows.

This comes about because a "no double spend" system needs to be sure that something DID NOT HAPPEN, not that something did happen.  You need to be sure that someone *did not spend*, not that someone did spend.  The only way to know this, in one way or another, is to get regular updates on spending rights/balances/.... of other users.

This is entirely different for, say, a communications network.  You really don't care that two remote nodes had a communication.  It is no burden for you.  

This is somewhat related to the Gibbs paradox in thermodynamics.  Suppose that you have two separate networks.  If these two networks are communication networks, and you link them together, then this new link doesn't put any burden on each of the participants on each of the subnets ; on the contrary, it opens up new possibilities of interaction between users of both sub nets.  This si why the internet is such a big success: hooking up a new subnet doesn't cost anything to other users.

But suppose now that these two networks are two different crypto currencies.  "linking them together" would mean, making the tokens of one exchangeable for the other one at a fixed rate (say, 1-1), so as to make them fungible.   What happens when we do that ?  Each user now has to have TWO block chains, somehow, united (but the data on both of them is now necessary).   Each user will now receive twice as much blocks/transactions .... So the technical burden on each user of each subnet doubles.

Well, it is interesting that you write "psychological" and not "political".  
There is indeed a very large political incentive to go for decentralized systems, because a centralized systems give a huge amount of power to the central authority.  However, in a system that is supposed to work on an economical incentive base (like "mining rewards"), the laws of economy apply, and I just wanted to point out that the "scaling wars" are using essentially economical arguments, like "it will be too expensive for people to set up nodes", and "LN nodes are economically incentivized by the fees that they can obtain".  If the system is supposed to work by economical incentive (and a monetary system usually will be !), then one cannot escape from economical laws.


Of course, but this hidden cost of central power will not offset the centralizing forces in a decentralized system, on the contrary.  The central forces, knowing they will benefit from this central power, will even be willing to "centralize at a loss", giving even more impetus to the centralizing forces.  So the *arguments* that a specific form of scaling would lead to economic incentives to centralize, are moot, because ALL scaling will have economical incentives to centralize beyond a certain size.

I have though along those lines, but there are difference with bitcoin network.

Already node are not identified on the network, there is  no "ip" or address associated with a specific node on the network.

And with internet, ip are hierarchised, meaning there can be subnetwork, at ip level, and dns also added ability to create domain and subdomain, this concept is totally lacking in bitcoin network.

To me the best (only way) to scale the network is to do something like inter-net , with sub networks/domain , or adding some kind of hierarchised structure to the network to be able to have more macro management and compartimentalization of processing into subnetwork.

But bitcoin protocol doesnt really allow this, I guess for sake of anonymity and fongibility, but that also prevent true scaling and efficient routing / masking of bitcore packets.

There is still the psychological factor of not trusting centralized financial institutions that can justify the additional cost.

To me it's same pb with oil, we know it's dirty, at the center of many wars and corruption, but it's still being used because it's  the cheapest,  and the "side effects" are secondary because they are not integrated into economic model.

But there can still be "hidden cost" with centralized network to garantee the trust and security, and avoid corruption and shadow banking which still had significant cost and showed to be an issue with the 2008 crisis.

It seems that you have finally written some substance worth of attention

Though it doesn't mean that it is applicable to an entire set of cases. As you would say, it is only a subset of the whole set of options. If we take Internet again as an example, does a switch or router need to know something about other existing networks out there? I guess that it doesn't. All it needs to know is to which port forward the packets if they belong to the network it is set to manage, or forward them to a specific port (which is an uplink to another router or switch) if these packets don't belong to this network (or drop them entirely if it hasn't been given an uplink address). This is how Internet scalability works in practice. I suspect this is basically how Bitcoin sidechains or payment channels are set up as well. Therefore there is no reason to think that the overall cost is necessarily set to (linearly) increase with the expansion of the payment network since these payment nodes can be set up along with the expanding network as required and using already existing hardware. You can use a typical computer as a router just by adding enough network cards to it, though that could in fact be more expensive than to buy a switch but in case of a payment node you don't need to buy anything at all

I wonder what is the origin of our inability to communicate.  I guess we both think that the other party is being deliberately of bad faith, but there must be another reason.

I'm telling that the peculiarity of a decentralized, trustless "no double spend" network, is that the technical cost to each user has to grow (essentially linearly) with the number of users on the network, contrary to other types of decentralized networks, like the internet, which was taken as an example.

This is because the big difference between a class of decentralized networks like the internet (where the cost can be almost constant per user when it grows) and a cryptocurrency type of network, where the trick is to know that there haven't been double spends, is that each user needs to know the spending rights of ALL OTHER users ; so the more users there are, the more each user, individually, needs to know, which augments the costs of storage, computing and bandwidth for each user.  The only solution is that users delegate more and more this knowledge to centralized authorities, that do this checking for them and which they have to trust.

I would think that that argument is quite simple, clear to understand, and evidently true, and indicates why decentralized systems like bitcoin cannot scale at constant user cost, and that "increasing the network" doesn't pay for itself, contrary to a network like the internet, where more users don't put a burden on existing users.

The fact that decentralized "no double spend tokens" networks have a cost per user that scales more or less linearly with the size of the network, and a centralized, similar network, can essentially work at constant cost per user, means that there will be 'centralizing forces' in a decentralized network, from the moment that the rising cost-per-user of the growing decentralized system crosses the fixed cost of a similar, but centralized system.

This also means that all scaling "solutions" will have in them, automatic economies of scale that push towards centralization of some kind.   The type of centralization will be different, but its effect will be equivalent.

There is simply no solution to scale a decentralized no-double-spend system beyond a certain size, because it becomes non-competitive with respect to a centralized version of it, towards which it will naturally evolve.

I'm curious if you really understand how irrelevant your remark is?

I brought an example of Internet only to show how the system can be internally scalable, that this effectively infinite level of scalability is conceptually as well as practically possible (simply because we don't yet have payment channels and thus cannot say with certainty if they are really scalable as the theory behind them assumes). I never said nor implied that Internet is trusted or distrusted, or whatever. I'm also curious if you really misunderstood this example so much?

That's why I started my post with the suggestion to get back to reality

What's the purpose of your thought experiment if we will never even come close to that in practice? It should be obvious to anyone with half a brain that if there were no more mining (in respect to one branch), there would be no confirmations, and thus the coin would be essentially left for dead. In other words, you don't need a Gedanken experiment to understand that since common sense would suffice just as well

You see, that's a valid argument to say that if miners are divided, users can chose, but it is not an argument against my Gedanken experiment that tries to demonstrate, theoretically, the futility of non-mining nodes as a power structure that "keeps miners in check". 

You are now arguing about a fork by divided miners.  A fork is simply decided upon in the market, like ETC/ETH, and also has nothing to do with the "power structure of non-mining nodes".  So it is no argument against my statement that "non-mining nodes have no power at all, and don't keep miners in check". 

My texts are coherent.  But they are not polished, they are long, because raw.  The effort to polish them would be 10 times the time I take to write them (I type about as quickly as a speak, to the time it takes for me is about the time it takes to read them out loud).

I don't really need people to read them.  They help me formulate, for myself, arguments that would have remained implicit in my head, and by having to write them up with the idea that *someone* might read them, I am pushed to formulate them explicitly, which helps me get the logic, for myself, right.     At the same time, it is also a challenge, and if people are interested - sometimes they are - they can argue against my positions.  In as much as I understand and appreciate their rebuttals, I learn something, and I re-organize my own understanding of the whole system.  That's the utility of this forum to me: as a test bench for my own understanding.
My customers get my polished understanding, where I have to be much more consensual, careful and precise.  But that's not free.  However, I need to be aware of the potential arguments and rebuttals of certain positions.  That's what I try to get out of this forum.


You see, this is an empty argument.  My argument is technical and economical, because the entire question is economical: the cost of growth in a decentralized versus a centralized network, and that cost is technical, because the arguments put forward are technical/economic arguments: "the network/disk/computing cost of big blocks" for instance, and your argument that a decentralized network pays itself for its expansion.


Of course not, both are most intimately related, as you should be able to see when you read more than the first paragraph of one of my previous posts.  The *technical cost* of a centralized network always scales at least as favourable, and in most cases, much more favourable, than the scaling of the cost of a decentralized network ; simply because the cost *per user* of a decentralized (trustless transaction) network has to rise (often almost linearly) with the scale of the network ; while the cost *per user* in a centralized network can essentially be kept constant as a function of the scale of the network.



Well, if you read what I just wrote, you see that I argue exactly the opposite, so this is perfectly on topic.

You cannot compare this with "the internet" because the internet is not a trustless network where every node needs to know the balances (or other "still right to spend" proofs) of ALL OTHER USERS, which is inherent in a trustless "no double spend" system.

If we have a global internet mesh, when Joe is connecting with Jack, then Mary doesn't need to know anything about that.  However, in a trustless double spend system, Mary DOES NEED TO KNOW, at the end of the day, that Joe spent his coins to Jack, so that she doesn't accept them.  So the  more Joes and Jacks there are on the network, the more Mary needs to know.  While Mary doesn't need to know if there are many Joes and Jacks when it is about communication.  Mary doesn't need to know that Joe sent a big file to Jack.  Mary does need to know that Joe spent his coins on Jack.

This is the big difference between a communications network, and a trustless, decentralized "no double spend" system, and the reason why the technical cost, in such a network, has to rise, per user, with the size of the network.  While a centralized network doesn't have that problem.

Let's get back to reality here

And the reality so far stands for Bitcoin Core just like most nodes and half the miners. So if the other half of them chooses to collude and produce a funky block, they will most likely find themselves in a sort of vacuum, with no users and no nodes. So the picture you are trying to draw here doesn't represent reality quite well. In fact, it seems to be distorting it heavily. Further, I wouldn't consider simple users and exchanges as sheep just because they've been pretty vocal about their position. I refer, or example, to a letter signed by major exchanges where they basically warned about rejecting any such attempts


Yes, just like I said above they are not sheep and have long chosen which side to stick to