Normally, a logical argument stands or falls by itself, and has no bearing to its author.  That said, I have no motive.  I'm not involved in crypto, apart from studying it.  I discuss crypto essentially to improve my own understanding of it, which is my only motive.  

A mathematical proof can be checked by itself, just as a cryptographic proof.  In the same way, a logical argument, even though not watertight, is of the same nature.  The arguments I'm putting forward are exactly those that convince ME of what I'm saying.  If you can punch a hole in it, you might convince ME that I'm wrong.  I would then be grateful.


Yes.  I think that is logical, no ?  If you are waiting for blocks that are not mined, your node simply stops.  And if you try to mine them yourself, you *become a miner*, and you've made a new coin.


I've been following the ETC/ETH split as it unrolled, because it was an eye-opener for me.  It was a fascinating time to follow.  I don't know if you "were there".  I had bought 5 ETH to play with a month before or so, and then found myself being owner of ETC and ETH.

When there are TWO coins, that is a full hard fork, the end result is determined by the MARKET.  If the market makes an obvious choice for one or the other, one might die ; but ETC survived on less than 10% of mining power and market cap.   This is probably why people talk of 95% (hopeful) consensus before forking if you do not want 2 prongs.  

That said, bitcoin is nastier in this respect than ethereum, because ethereum, like most more modern alt coins, have continuously adjustable difficulty ; while bitcoin has these 14 days which can be very, very long if your hash rate available is lower.  So if a hard fork keeps this aspect of bitcoin (a hard fork can change ANYTHING, so just as well the difficulty adjustment), and you are left with only 10% of the hash rate, that branch will have a hard time surviving, with a block only every 100 minutes on average, and 5 months for the next difficulty adjustment.  So most probably, bitcoin has a build-in forking protection unless the forkers adapt this.

Those that "do not want this" are those that lined out the current protocol, in other words, the whole bitcoin community.  By putting a hard limit on block size and by programming block reward halvings, they made it into a speculative asset with a limited number of possible transactions.  The idea that one "would adapt it when necessary" is a bit naive if at the same time, one is firmly convinced that the block reward halving is firmly unmodifiable.

It took me some while to understand that too.  But once you see it, it is quite obvious in fact.   I guess people are mentally locked in on this dogma, because they have been making decisions based upon it, which would bring too much doubt and regret if they allowed themselves to understand this.

It blinds them to two aspects of bitcoin: the fact that true centralization is mining pool centralization (the deciders of the block chain building), and that this centralization is already well advanced ; and the fact that bitcoin is going to lose its status as a currency to become a reserve currency because of an erroneous argument over the importance of non-mining nodes and the cost of disk space versus the emerging fee market and the economies of scale that proof of work induces.  

I think it is too late now to change this.  Block size now being an economic parameter determining fee market, the consensus mechanism will most probably make that this cannot be changed any more, especially because the miners are the ones that are on the receiving side of the fees.  It could still be changed, if mining centralization is sufficiently advanced, but I don't think that that is the case for the moment, to come to an oligarch consensus without sufficient dissidents.

As to the fact that "important bitcoin people" could fail to understand this aspect, there are only two conclusions: they understand much less about the dynamics of the system than one would expect them to have ; or they have a good reason to pretend not to understand it.

"if your income depends on you not understanding something, chances for you to understand it are low"

I said this somewhere else, but defining the word "consensus" without having it referring to a genuine phenomenon, is a useless exercise.  Consensus is a phenomenon.  Not something you can define as "what should be" or "the will of the people" or any other abstract definition, devoid of relationship to reality.

Consensus is the observable phenomenon that a lot of different people, outside of any hierarchical constraint, use the same rule set to come to the same data set.  In the case of a crypto currency, this is even more amazing in the sense that the rule set, and the data set, are not what is the most profitable for each of the participants.  But nevertheless, the *phenomenon* of people settling to the same protocol, and the same block chain, is the "consensus" phenomenon.  It has two aspects.  You could say that it has some "ergodicity property".  The consensus phenomenon applies just as well to the "different participants at a given time", as "over time".  The phenomenon over time is called "immutability". 


I don't deny this.  But now, you cannot talk any more about "majority", because ETC and ETH are different currencies, different protocols, different block chains and different users.   Whether the market share of ETC is bigger or smaller than ETH does not matter any more than whether ETH market share is bigger or smaller than bitcoin's or litecoin's.  They are different crypto currencies, with different consensus, different protocol, different block chain.

Mining always adjusts, in the end, to profitability.  That is true after a hard fork from a single currency, but it is true also between totally different currencies (at least, if the mining is compatible with the investment in hardware).  Mining hash rate is normally proportional to (block reward per unit of time) * (market price of coin) / difficulty cost price, over different currencies with compatible mining in the short time, and over all crypto currencies in the long term (after mining hardware can be written off).

Thanks for understanding that too.  Bitcoin not being a "proof of node" coin, nodes aren't the boss, like you say.


I learned a lot too during the ETH/ETC split about the dynamics of consensus.  It was a fascinating experience. Maybe bitcoin people didn't follow this drama very closely and didn't learn as much about how proof of work consensus networks really work.

Yes, against a *communication* attack.  But that has nothing to do with what the consensus over the ledger, the block chain, is.  That block chain is built by a community of miners.  And is used by user wallets.  As long as there is reliable communication between the miners and the user wallets, that's all a crypto currency needs. 

The distributedness of the *communication network* matters only in that respect.  If unrestricted network access is possible, then it doesn't serve much of a purpose.  Yes, I can use Tor to connect to a remote web site.  Yes, it will be difficult for an attacker to sensor that.  But if my only reason using Tor was to access the web site, and not my anonymity, then I can also connect DIRECTLY to the web site.   That is like using a VPN server.  Yes, if a direct connection is sensored in the country where I live, a VPN is useful, to avoid the "network censoring".  But if I can connect directly, and I don't care about them knowing my IP number, I don't need a VPN.

The other aspect of the distributed node network of bitcoin is its "block chain memory".  Yes, the block chain will never be eradicated from earth, because so many people have a copy of it, that's true.  But for SURE, the miner that is mining the chain has the correct copy he's mining on.  So I don't NEED that "global distributed memory" of block chain, because the "server" I'm using has it automatically.

You can say, yes, but that miner, as a "centralized server", may trick you.  No, he can't.  Because to trick me, he has to produce a block every 10 minutes or so on that chain.  Otherwise, my node following him, will stop.  If this is not the right chain HE thinks honestly he should be working on, then to trick me, he is building a huge orphaned chain and wastes mountains of hash rate.  The risk of me discovering another miner, and leaving his chain behind, while he wasted all this hash power just to trick me, is too big for him.  So I KNOW that a miner who is putting out a growing chain, is providing me with the "best chain according to him" on which he builds.

The best information about the block chain is provided by an active miner.  The "distributed network" only makes sure that this is not forgotten if ever that miner disappears, and makes sure that in case of network censorship, one can still get at it. 
But it doesn't play a role in any form of consensus.  If the internet works well, it is not needed.

This is remarkable as a brainwashing, isn't it.  A long, detailed explanation is swiped away with a single reminder of the Dogma without the slightest bit of logic.  Learn to think for yourself!

Again, the consensus mechanism is proof of work.   And the very proof that nodes do not determine much, comes from the simple argument that one can fire up sufficient nodes on Amazon to swamp every vote.  Someone called me out on that.  That's not the point.  I won't do it, of course, because 1) I know it is useless, and 2) it costs a lot of money.  Much less than a miner spends, but more than I spend on crypto (a few 10 dollars to play with some coins, and more if I want to buy something on the internet).

It does cost money to fire up 100 000 nodes on amazon.  But if one has strongly vested interests in one bitcoin flavor or another, (I'm not), then this cost is more than reasonable.  Why doesn't Roger Ver, or another antagonist in this debate, does so if it were so simple to "vote by node" ?

Why didn't it happen yet ?  Exactly because proof of work is the consensus mechanism, and not proof of node.

Now, I know that an "accepted truth" is hard to change in one's mind.  I also believed that the "bitcoin node network" was important.  I set up an empty node at home to "support the network" (and to study the block chain).  I'm fascinated by the dynamics of decentralized systems.  I study it since quite a while and when I learned about bitcoin, that was a great example.  My main motivation for studying it, is political: I'm an anarchist.  But the more I studied it, the more I saw that things that were "taken for granted" don't fit with the dynamical laws of the system.  And "non mining nodes are important" is one of these dogma's.   It takes a while to understand the exact dynamics of the whole.  And it is very interesting.  It is also very interesting to see how initial misunderstandings can be "locked in" because everybody repeats it.  Maybe there is a "consensus dynamics" over what is to be told about bitcoin too

No.  Because that is the Proof of Work consensus mechanism.  There is no other "voting" mechanism.  Proof of work is the voting mechanism on every part of the consensus system, and is proportional, well, to proof of work.   The "number of nodes voting" has explicitly been avoided, because that's open to a Sybil attack.  The "amount of bitcoin you hold" is not the consensus mechanism chosen (called, proof of stake).

However, the user DOES have of course a strong voting power *if there is a hard fork*: namely, in the market, by putting value on one coin, or another.   And "proof of work" (mining power) will follow the most lucrative block reward (and hence market cap).

But for that to happen, there first needs to be a hard fork: if there are no two block chains available, and not two coins, the user can only accept the single, existing chain, or decide not to use that coin.  But that's it.

Of course, miners will be *sensible* to any market movement, so indirectly, users deciding to leave bitcoin might have an effect on miners.  Miners will be hesitant to make a hard fork, because they may be afraid that the new coin they make, will have a lower return on hash rate than if they continue mining the old chain.   Or they may hesitate remaining on the old chain if they think that users will massively switch to the new coin.     So yes, users have MARKET INFLUENCE on miners.  Direct market influence if there is a hard fork and two coins (there, the users really vote with their money), or indirect "market sentiment" before miners decide to fork or not.

At no point, non mining nodes play a role in this, because the consensus mechanism is not "vote by node" but "vote by proof of work".  Satoshi uses both interchangeably because he considered that every node would also mine.

I honestly think the brainwashing is on the other side.  I present logical arguments, you only present rethoric and ad hominem.  I indicate the way I see how it functions and I explain it, and you only present (totally erroneous) conspiration.   That's like someone writing open source code, and you arguing that it must for sure contain a virus because of some conspiracy.  Read the code !  Read the arguments and tell me where you think it is wrong, and I'll make you see that you are mistaken (or I will see where I'm mistaken, and I will learn something).

Bingo. That's my point.


Absolutely.   We should clearly define what a "miner" is in this respect.  A "miner" is the entity that decides *what* block to build on *which* other block out there, and puts that block at disposal on a node.  A "miner" can be a complex entity, like a mining pool and its customers with mining hardware, selling hash power to the mining pool, or a single entity (solo miner).  The "providers of hash power" are not, in this respect, "miners" as we define it here, because they simply provide hashes, and do not decide upon the block to be built (the contents of the block, the transactions ; the previous block hash, and the protocol according to which it will be built).  The miner, in this conversation, is the entity deciding that, and *hence has a node*.  

As I said before, given the cost of mining, the cost of setting up an efficient node is minuscule compared with it.  There's no point in investing a huge amount in hash rate capacity, and wasting a lot of electricity, and not have a sufficiently efficient node that has strong connections to other miner nodes, because the time wasted to learn about blocks made by other miners, wastes much more money by lost hashing (on orphaned blocks) than the bit of cost of good infrastructure in a strong node, with good network.

So we can presume that a miner node will do everything, if the "proxy network of non mining nodes favorable to him" is too weak, to set up somewhat better node infrastructure that can serve directly its clients, the users, that want to use the block chain he (and his fellow miners on the same protocol) are mining (in other words, his coin).  Especially if that user is an exchange that lists his coin !

This is a bit like torrent: if you want to distribute a file, you can count on other torrent nodes putting your file at disposal (distributed network).  But if your income depends on having this file at disposal, you will make sure that you have enough server capacity to distribute the file yourself on the torrent network, independent of all the other torrent clients refusing your file and/or putting other files at disposal.

Now, as to the users sending transactions, that's the same.  If you are a user that *wants your transaction on the block chain of your choice* and you cannot find enough distributed nodes doing so, then you can connect your WALLET directly to a miner's node, or you can connect your NODE directly to a miner's node, to send your transaction.

Again, the distributed NETWORK of bitcoin nodes is a nice thing that protects against a *communication attack*, but if direct internet links are possible to miner's nodes, and if those miners have installed sufficient network capacity, then the distributed network is *optional*, and cannot impose its will.

Coinbase's node is important of course, because coinbase is a USER: what is important with coinbase, is the wallet on it, because coinbase's wallet contains the bitcoins which backs the bitcoin IOU that people trade.  Coinbase is hence a BIG USER, and hence a BIG MARKET PLAYER.  Its wallet is extremely important.

But coinbase could directly connect its node to one of the nodes of a big mining pool.  Whether YOUR non-mining node (and 1000 other non-mining nodes of others) is running, or not, doesn't mean anything to coinbase.  However, the chain that coinbase decides to follow (by choosing a protocol), is the coin that coinbase considers being its wallet.  If miners make two coins, and two chains, then coinbase will decide *what coin* it is trading.

The importance of coinbase's node, its choice of protocol, and hence its choice of chain (if that chain exists and is produced by a miner !  Otherwise, coinbase's node simply halts) comes from the fact that coinbase's wallet is the wallet of a big user.

Myth.   Bollocks.  Non-mining nodes are simply miner-chain proxies.  You are right that a "decentralized network" consists of independent nodes, but a crypto currency is not a "decentralized network" but a "decentralized block chain".  And that is built by miners.

After all is said and done, the "network" doesn't matter, the block chain does.  The crypto currency is the block chain, not "the network".

A (hopefully) decentralized network is for instance, Tor, and matters for communication.  But a crypto currency is not about communication but about a consensus ledger: the block chain.

The only way in which a "decentralized network" matters, is if there were some difficulty in the *network connection* to *obtain the block chain*.   For instance, if the Great Wall of China were blocking some miner nodes, and one has to get through via a P2P way in some or other way.  THEN the "decentralized network" matters.  But if there are good internet connections, and you don't need anonymity of communication (like with Tor), then that decentralized network doesn't matter.

Because the crypto currency is the block chain: the consensus ledger.  And that is built by miners.  Not by non-mining nodes.
And you can obtain it directly from the miner if you want to.  You don't need the P2P network for that.

Miners make block chains.  Users (with wallets) use them, and connect to nodes.  They can run their own full node (the only real reason to run a full node), connect to a "random node on the network" with a light wallet (and trust that node), or connect directly to the miner's network.  If the node to which they connect, runs the same protocol as the miner's (or a more permissive one), they will obtain a copy of the block chain made by those miners and their wallet will do things according to that block chain.

If they don't, well, they have a stalled node, connect to a stalled node, will not see the right balances, and may produce invalid transactions.  

If there are competing miners, with competing block chains, then the protocol choice of the node you use (one of the miners' protocols, otherwise, see above) will select the block chain corresponding to that protocol.  It is as if you used "that specific coin".  We have a hard fork and two coins now, because there are two living block chains.

So, essentially, miners make block chains, and users (with wallets) connect to a node (their own, or another one) that has a protocol that corresponds to the block chain at hand.  If there is only one block chain, they have no choice, there is only one chain.  If there are several chains, then there are several coins, and their choice of node selects their choice of coin.

(warning: their transactions can be valid on both chains in some circumstances).

Miners serve block chains.  Wallets browse block chains.  Nodes are proxies to block chain servers.

There is no such thing as an orphan hell in the network.  Let us define clearly what an orphaned block is: it is a valid block, but on which other miners decided not to build any more.  So only MINERS can orphan blocks.

BTW, miners (of same protocol) have good reasons to have very fast connections between them, because they need to know very quickly when a new good block (according to their rules) is mined.  They will not wait for it to be randomly propagated through the P2P network, because the time they need to receive it is lost hash power for them. It is this strong, fast network of equally-minded miners that is the "block chain principal server" (for that protocol).

What you can have, is that nodes come to a grinding halt, if their validation rules don't find any valid chain.  This happens if no miner produces such a chain.   Imagine that I write a modification of bitcoin core, with a soft fork, so that only blocks less than 500 KB are valid.  If I install this software, and connect it to the bitcoin network, it will download a certain part of the chain, until it finds the first block bigger than 500 KB.  And then it stops.  Because no other block, less than 500 KB, and directly built on that last block, is around.  I simply have a stalled node.  Did I "orphan" the rest of the bitcoin block chain ?  Not at all.  The rest of the chain is simply invalid according to my node, which will wait forever for the next valid block that no miner will make.

Suppose now that I fire up 100 000 nodes on amazon with this software.  Of course, I will mess up somewhat the bitcoin network, because many nodes will ONLY have connections to my amazon nodes which have stopped.  But after a while, the normal nodes will find out that my overwhelming majority of nodes do not use the same protocol, and that a GOOD chain does exist.  They will end up finding one another, banning my nodes as confused, and I won't be able, even with my overwhelming node majority, to bring down bitcoin to accept my soft fork, wind back the chain to my last valid block some years ago, and restart from there.  Even though I have more that 95% of the nodes in my hand on Amazon.

If it really were that simple that one can impose rules by having a large majority of non mining nodes, then I'm sure that someone would already have fired up 100 000 nodes on Amazon doing that.  The very fact that bitcoin has never undergone this, simply means that non-mining nodes are not that important.

And as a consequence, that the argument against bigger blocks, "because of less nodes and hence more centralization" is totally bogus.  But it is a hard myth that will not die.

Bingo.  

However, consensus dynamics is such that it IS now part of the protocol rules.  I fully agree with you that it was *extremely stupid*.  But now that block size became an *economic parameter*, whether one wants it or not, it IS locked in as a protocol rule.  The same rule that makes that one cannot find consensus over "redoing the last few blocks and leave out a few transactions", the same rule that makes that one cannot "jump a block reward halving", the same rule that makes that one cannot change the 10 minutes into 2 minutes.  Because all these parameters are economic parameters, meaning: if you change them, some get an advantage, and others, a disadvantage.   2 or 3 years ago, most probably the block size was still a technical parameter and could have been changed under "neutral consensus".  But now it is an economic parameter.  It is locked in.  Only centralization or a hard fork with two coins will be able to change it if the immutability mechanism means anything.

My idea is that people didn't really understand "consensus dynamics" and all its consequences.

What does that mean "validate" ?  It means: write it on their own copy of the block chain.   So if you own a node, you check for yourself that the block chain is validated, instead of having to rely on another node telling you so (with a light wallet).  But that's it.  

But they don't prevent miners anything.  Miners make blocks and set them on their nodes to be distributed on the network.  All USERS agreeing with those miner's rules will take the block chain and the new blocks from those miners' nodes, eventually directly if no other nearby nodes propagate them by configuring their nodes so that they connect directly to the miners' nodes, or by having their wallets connect to one of these nodes.

If you install a node that doesn't validate a block from a miner, that will not stop that miner from making that block and from getting it to the users agreeing with him ; more importantly, it will not stop miners agreeing with the previous miner from building on that block.  

Now, if your node doesn't agree, it stops downloading that block chain.  And that's it.  It won't download it, it won't propagate it.  But that won't stop users agreeing with these miners, and those miners amongst themselves, from building this chain, validating it amongst themselves, and use it as *the* block chain.

"non-validation" by a non-mining node doesn't stop that block from being mined or used.  It is as if your node simply didn't exist.  And it certainly won't stop other miners from mining on top of that block if THEY validate it.  If your node is waiting for a block that is never mined, it simply stops.


No, their fear is that their peers, the other miners, will orphan it.  The only entities in a crypto network that can "orphan" a block are other miners, who build on a previous block to make a longer chain according to their rules.  No non-mining node can orphan whatever block.  It can refuse a block, or it can find a longer chain.  It is the existence of the longer chain (made by a miner) that orphans a block.  Not a non-mining node.

What makes "anon" coins special in this respect ?  Why wouldn't other CPU coins be mined with a bot net, and in what way is a developer team the only one that can deploy a botnet ?   Isn't competition with botnets similar to competition with asics ?  

Can't you accuse every CPU coin to be botnet mined ?  Maybe bitcoin too in the first years ?  Maybe Satoshi had a botnet running that gave him all the coins until 2011 ?

I fail to see the logic behind this argument.  The only few XMR I ever had, was through mining them myself on my laptop !
It is the only coin I ever mined on my laptop.   So I don't see where this idea, that because it can be mined with a PC, it is of course mined by *one single botnet under the dev's control*.  If this took a long time (more than 48 hours say ) I would think that there was possibility of fair competition.  If a quarter of the current stash is mined before the most astute co-miners are even aware that the launch took place, with a difficulty "bug" so that you can do it on your PC, chances are it is the same guy launching the genesis block that got most of it, no ?  If on top of that, there is a PoS mechanism that amplifies initial stash, the word "fairness of distribution" becomes difficult to use.  If on top of that, there is a PoS voting system to use 10% of the mined coins on "development", you can imagine how centrally controlled this thing is.

However, it is very well possible that all this happened by accident, and that Evan Inc sold their initial stash on cocaine and prostitutes when DASH was less than a dollar, and are now biting off their nails when they see the price of DASH rise, not holding a single penny of what they instamined.  But allow me to doubt it.

Indeed.  Although we may be thinking about slightly different aspects.  Non-mining nodes are indeed a form of distributed proxy server of the block chain(s), but the real sources of those chains are the miners.  The distributedness of the network (many nodes) make for an easier downloading, communication and so on, but as long as the principal servers are "up and running", their proxies (the non-mining nodes) are only that: proxy servers.  They copy the block chain of their flavour from their principal server, and help speed up its download.  They also help sending transactions up the chain to their principal "miner" server, but those wanting their transactions to get there, can just as well send it directly.

Again, the amount of non-mining nodes is essentially immaterial - apart for their owners, who use it to verify for themselves the chain they want to use (in as much as one is available by miners), and do not want to depend on unverified third-party information.

That is like saying that having more people running firefox looking at your web pages makes your server distributed.

Non mining nodes don't contribute anything (or almost anything) to the "decentralized" nature of the network.  And the simple proof for that is that it is extremely easy to "Sybil attack" this amount of nodes.  One can easily fire up 100 000 nodes on Amazon and swamp the entire network.  

It is exactly because of this *unimportance* of non mining nodes, that Proof of Work exists.  Of course, in Satoshi's paper you will read about the importance of many nodes, but that is because for Satoshi, nodes and miners were the same.  All nodes were miners.  THEN of course, it is important.  But non-mining nodes don't contribute (almost) ANYTHING to the decentralization of the coin.  The ONLY thing they do contribute, is communication paths, so that no censoring of communication between the user wallet and the miner can take place.  But with direct connections between miners (who have to be interconnected to not get their blocks orphaned) and every user wallet, no intermediate nodes play any role (apart if the internet were censored and they provide extra pathways).

If the amount of nodes indicated anything about the decentralisation of bitcoin, proof of work would not have been necessary.  It is because it is meaningless, that one needs to introduce it.  

This is a simple aspect of the dynamics of crypto currencies: nodes that don't "vote" (no proof of stake), that do not "mine" and that are only communication hubs that can be bypassed between user wallets and miners, are useless if the direct internet connections are good enough between wallets and miners.  However, full nodes ARE important for their owners, so that they can check FOR THEMSELVES that what their wallet sees, comes from the right block chain and that nobody is talking shit to their wallet.  But it is the only use.

This funny little fact is of course full force denied by people wanting to keep small blocks for other reasons and claim that it would bring "centralization" because there would be less hapless PCs running on the network, downloading the block chain from the miners.  But the "servers of block chain" are the miner related nodes.  And whether more or less "firefox" browers (non-mining nodes) are downloading them doesn't matter.

I run an (empty) bitcoin node at home.  I could fire up 20 of them.  What difference would that make ?  I can switch off my node.  What difference would that make ?  I can run a core 0.14 and a BU node on the same machine (which I will probably do soon).  What does that even mean ?