Beyond the electricity-guzzling, ‘failed’ blockchain experiment
Mon 12 Sep 2016
A Chinese researcher has proposed a new cryptocurrency system which, unlike Bitcoin, does not use a blockchain, and suggests that Bitcoin is a ‘failed’ experiment which encourages excessive energy consumption in an ‘impossible’ pursuit of genuine decentralisation.
In the paper Noncentralized Cryptocurrency with No Blockchain, Qian Xiaochao suggests that a second stab at Bitcoin’s ambitions could be feasible in ‘the 22nd century when some truly revolutionary breakthrough of communication technique or computational theory may have occurred.’
Xiaochao’s central contention is that Bitcoin is not genuinely decentralised, but rather ‘non-centralised’, and the paper goes on to define the three possible states in which a cryptocurrency system can exist:
In a centralised state, one single entity (i.e. a bank) controls the system; in a decentralised state, the system relies on a single safety contingency, the private key; whereas a non-centralised system is neither centralised nor decentralised.
Xiaochao contends that Bitcoin is defined by the third state, not – as its supporters would like to believe – by the second.
And if Bitcoin really is ‘non-centralised’ rather than ‘decentralised’, the enormous computing and energy resources which Bitcoin mining devours are being expended in vain, since the system can neither scale nor preserve itself reliably enough.
Xiaochao does not claim that his own proposed replacement system is any more decentralised, but that at least it is less Quixotic and energy-draining. ‘For decentralization,’ he writes ‘the blockchain mechanism doesn’t work and for noncentralization the blockchain method is unnecessary.’ He goes on to describe Bitcoin’s core model as a ‘trivial and wasteful design’.
The impossible dream of blockchain currency
Though the paper does not refer to the more recent Ethereum heist, it cites the Mt.Gox hot wallet grab, which began in 2011, as a significant example of why Bitcoin can never scale or be viably decentralised.
The paper contends that the miners and mining pools which power the Bitcoin blockchain are a core centralising force in themselves, since the weaker miners are inevitably going to diminish due to low rewards; in fact, it characterises the process as Darwinian and reductionist, suffering from the same tension between competition and consolidation as any other free-market impetus.
Sybil attacks and re-centralisation
This equation between economic and computer systems is best exemplified, as the paper notes, in the Sybil attack* scenario, wherein large-scale P2P systems become ‘re-centralised’ by attackers which represent themselves to the system as varied and different entities.
‘According to [the previous] definition of decentralization and noncentralization,’ writes Xiaochao. ‘Bitcoin is obviously not decentralized as some fans claimed but running as a noncentralized system.’
Suggesting that genuine decentralisation is ‘too good to be true’ in the face of four major attack behaviours (malicious message constructs, malicious schedule attacks, Sybil attacks and the tendency towards centralisation), the paper proposes a blockchain-free alternative entitled ‘X-Coin’. And it supposes the presence of a ‘weak virtual general adversary’, with strong but not unlimited computing resources and reach.
51% is still the issue
The system replaces the mining process with normal data lists, all of which have a local and global timeout value. Assuming the central agency to be a bank, a user would set a maximum transaction fee, and a bank to whom the fee is acceptable would broadcast a list of the transactions to other banks.
In this system the bank is acting as an ‘agent’, sharing the transaction lists and generating transaction packages which get broadcast and granted by other agents. Once a bank collects more than 51% of grants, it compiles the transactions into a ‘Package-51’ and broadcasts this as a potential candidate balanceview. The total transaction fees are then divided between the agent, the deputy bank and the participating banks.
Since more than one Package-51 can be generated, the resultant balanceview can then be split as necessary, in a process comparable to a blockchain fork. The paper proposes a checkpoint mechanism with a confirmation delay of 6 blocks, guaranteeing the termination and consistency of the transactions to a high probability – without anyone needing custom hardware or deep pockets for electricity consumption.
A blockchain for your grandchildren?
Xiaochao presents a radically negative take on the blockchain, as a future technology which is attempting a doomed first iteration ahead of its time. And the paper does alight on the genuine concern that any P2P system that needs such exhausting computing and energy resources may either be a little too precipitous, or a perhaps less elegant piece of software than it might be (since any system would benefit from such massive resource overheads, and from a private ecosystem of bespoke hardware).
Which is why it will be interesting to see the full implementation of IBM’s appropriation of the blockchain, eventually.
*John R. Douceur’s Microsoft Research paper on this subject, now nearly 15 years old, contends ‘that it is practically impossible, in a distributed computing environment, for initially unknown remote computing elements to present convincingly distinct identities. With no logically central, trusted authority to vouch for a one-to-one correspondence between entity and identity, it is always possible for an unfamiliar entity to present more than one identity, except under conditions that are not practically realizable for large-scale distributed systems.’