When the segwit part of seg2x activates, it still will be the original BTC though.

Not if Craig Wright’s claimed 20% of network hashrate miners fork off so that the original BTC does not include any SegWit blocks. And that mining minority wins if the “million buttcoins” whales sell Jihan’s fork and buy the original BTC.

Essentially (afaics) Craig Wright is correct that the original Bitcoin can scale somewhat by having some grouping (even a minority of the hashrate) of miners agree to increase the blocksize and assuming the economic majority (which is always a follow the (sic) “million buttcoins” whales Schelling point) sell the fork which refused to increase and buy the fork that did. The generative essence is my point from months ago that the “ultimately, the miners and the whales are economically the same entities”.

However, Craig Wright’s plan has limitations on scaling depending on how much decentralization we want. I need to write a blog about this to go into more technical detail, as such an explanation will justify the remaining relevance of my (unpublished) blockchain design.

Bottom line my stance is sell all Bitcoin forks which are not the original Bitcoin, because the (sic) “million buttcoins” whale wrote Fork off and lose (some recent discussion of that). Remember Mircea Popescu also had a strong claim that he was the DAO attacker.

But if Craig’s leadership can attain this control over the blocksize (or equivalently minimum transaction fee as a control over blocksize), this exhibits that Bitcoin is not immutable and that someday its 21 million coin limit could be violated. Yet it may be the least worst of the options forward, although it still will not be truly decentralized unlimited scaling. I am not convinced he can gain enough support for periodic increases in blocksize, although a one time increase might happen.

The limitations on decentralization in contention with scaling are not just on attaining periodic consensus on block size increases. It also has to do with the distribution of the hashrate over many mining nodes and the cost of validation versus block reward (solely from fees in the future) for a 0.001% hashrate miner. This sets a lower bound on the transaction fees which would (at the current time) I think prevent the nanotransactions which Steem uses each time we submit a transaction to the blockchain. That is unless we calculate that cost and delay time for validation of transactions is epsilon. I have not run that calculation yet, but afaics that will not be the case for arbitrary smart contracts (both in terms of cost and also because they can’t always be parallelized). Craig argues that eventually Moore’s law overcomes (validation cost goes to epsilon even with 100k nodes), but this depends on growth of number of transactions multiplied by the complexity of transactions (i.e. smart contracts) not growing faster than Moore’s law which eventually must be the case if Moore’s law does not abate, but it could be decades further out than he is charting.

Without a limitation on block size then miners can spam other miners, to amplify the advantage of having slightly greater hashrate so that over time hashrate further concentrates amongst those miners who have more hashrate. This is chaos (i.e. change one thing and then there can be unpredictable Butterfly impacts). There is a precedent that Craig Wright may be a quant who may not be respectful of chaos but I would need to research the issue more.

As validation costs go to epsilon due to Moore’s law, who decides what the block rewards from fees (as minted/coinbase rewards diminish) should be and thus the amount expended on securing the block chain? This is a power vacuum. The more profitable miners can starve the marginal miners of income by accepting lower fees. I doubt Craig has entirely thought this out. He did not address this in his presentation and they banned me from his Slack when I started asking difficult questions.

Essentially all existing blockchain designs are winner-take-all economies-of-scale power vacuums. Bitcoin can scale with centralized control of the whales, as can Steem. But then we all realize it is not decentralized and is thus a fiat. And we know that eventually that fiat will be controlled by the one world government. We did not really achieve our goal of making a currency that is controlled by no one. Note I am not against scaling Bitcoin interim with block size increases, rather just pointing out that afaics it is not a decentralized scaling solution.

Decentralized scaling would have much greater network effects. Centralized shit always ends up monopolized in stupid directions and becomes a retarded clusterfuck eventually. Evidence Microsoft Windows.

Edit: Bitcoin can never provide real-time, trustless proof-of-publishing. Instead there can only be 0-confirmation, which relies on trust of miners. Systems that can have failure modes eventually experience Murphy’s law from time-to-time. The argument in Bitcoin is always that miners have too much invested in hardware (that can not be repurposed) thus will not attack the network (while shorting it). Yet Craig then criticizes SegWit claiming we can not trust miners. Even if that argument is deemed acceptable, I think we will find applications of blockchains where the lack of real-time consensus on ordering of transactions will be an unacceptable flaw that excludes popular use cases and/or makes coding blockchain apps too brittle and difficult.

I will excerpt from a Pastebin I wrote on July 14:

For example, Craig is correct on this statement, "Risk is not perfect, every system is probabilistic". Even my consensus algorithm blockchain design has that characteristic.

Craig makes a very important point that propagation is 2.3 seconds to 98% of the hashrate. So presuming the network is made only of nodes with powerful hardware, then validation can also be fast perhaps a couple of seconds. So he presumes that nodes with lower hashrate are not losing more than a few seconds out of the average 10 mins to mine a new block, thus he thinks non-uniform hashrate is not a problem. He must set the minimum transaction fee high enough that those who only win a block every 1000 blocks they validate have insignificant costs, unless he wants mining to be highly centralized. And that is one of the Achilles heels that makes my design superior. In my design, the transaction fees can go much lower. There is no way users will want to pay $0.005 every time they make a comment on a blog post for example. So his design for Bitcoin can not compete with the markets I am shooting for, unless he further centralizes it.

Also his analysis fails to refute the fact that only two fabs in the world make ASICs, thus the manufacturers can extract most of the profits and centralize mining. Also there is never a way to know how centralized the control of hashrate is. Mining nodes are a Sybil attack on identity.

Also it impossible for Craig to refute selfish mining in the 50% centralization case. The cartel could be delaying their blocks and taking more than their proportional hashrate share of the profits thus eventually controlling 100% of the network hashrate. All of this can be surreptitiously and can never be detected. So all of his data is useless for this.