A parliament without a parliamentarian

I mean to write a bit about the “DAO hack“, which provides a fascinating true-crime introduction to the fascinating (virtual) world of blockchains and cryptocurrency if you don’t already follow this stuff. But before saying what I think about that DAO-stardly DAO-eed, I think it might be useful to explain what I think “blockchains” are, in social rather than technical terms.

A blockchain is just a parliament without a parliamentarian.

Like a parliament, a blockchain is a means by which a group of people collaborates to produce an ordered list of accepted “resolutions” that is deemed authoritative and legitimate by some community it represents. That list of resolutions may cause the construction of alteration of some side-product. Laws passed by Congress create or alter the United States’ legal code. “Blocks” accepted by Bitcoin “miners” change account balances in an implicit ledger of who owns what Bitcoins. But the authoritative source is always an ordered list of accepted resolutions, from which the state of the side products may mechanically be derived.

A traditional parliament requires a lot of work and organization, and it doesn’t scale very well. It’s hard to coordinate getting even a few hundred legislators into a room to vote on a resolution. Keeping an orderly chamber requires organization and hierarchy. Typically there is some kind of leadership, there is a hierarchical structure and “rules of order”, whose purpose is to orchestrate the consideration of resolutions. A “parliamentarian” acts as a keeper and adjudicator of those rules, which are often obscure even to the membership of the parliament. The parliamentarian and the hierarchy her rules enshrine may not be neutral. Most obviously, the leadership of a deliberative assembly may be capable of preventing consideration of resolutions that the membership would pass if required to vote on the question. Anyone who follows real-world politics understands that it matters very much who “controls” a legislative house.

A blockchain represents a deliberative assembly that may be very large (thousands or millions of participants bound by computer networks), and that may be very open (anyone may participate, “legislators” may come and go at will). Traditional rules of order are not up to the task of managing this sort of assembly. Further, the inventors of blockchains did not approve of the traditional prerogatives of a parliament’s “leadership”. Among blockchain enthusiasts, preventing the consideration of a potentially acceptable resolution is usually referred to as a “censorship attack”.

So, instead of a “leadership” that orchestrates the consideration of resolutions, blockchains hold a kind of lottery among its legislators, who are called “miners”. Every few minutes, or even seconds, the winner of a new lottery is announced. The winner gets to submit resolutions for consideration by the parliament, and is financially rewarded if her “block” of resolutions is accepted by the majority. The proposal of resolutions is not restricted to miners. They may be submitted by, well, anyone at all. Miners check the resolutions and decide if they are likely to pass. They combine the ones that do seem likely to pass into a “block”, and hope to win the lottery. A miner may try to exclude resolutions that are likely to pass but that she herself disagrees with, but that sort of censorship is unlikely to have any effect, since she is unlikely to win the lottery in any particular round. Once a block of resolutions pass, each participant updates its own personal copy of the list of passed resolutions to include the new ones. Only participants with a fully up-to-date copy of the list may participate in the next lottery. Since winning the lottery and proposing a successful block is financially rewarded, while censoring proposals or ignoring blocks that the majority would accept is ineffectual, participants usually propose anything that they think would pass and go along with anything that has already passed. Blockchains reward consensus: It is lucrative to go along with most others would go along with. Understanding the will of the majority of their colleagues and bending to it is the job of each and every legislator.

In the most prominent, current blockchains, the norms about what sort of resolution is likely to pass are simple and widely shared. On the bitcoin blockchain, most resolutions amount to something like, “Unspent money belonging to User A in an amount of 3 BTC should be assigned to User B.” A miner checks that User A does in fact have 3 BTC unspent, and that the resolution is properly signed by User A. If so, this transaction is very likely to pass, as the core shared norm of the community of BTC miners is that people should be able to spend their own unspent money however they choose. Note, though, that this in only a norm. If more than half of the participants in the Bitcoin blockchain suddenly decided that User A was evil and should not, in fact, be permitted to spend her money, then lottery-winners would quickly learn to exclude her resolutions, and that would become a new, communally enforced norm. People in the Bitcoin community who consider a norm like this illegitimate would refer to it as “censorship” or a “51% attack”. But “51% attack” is just another way of describing “majority rules” when you don’t like the decision of the majority.

But majority of whom? Open blockchains, like Bitcoin and Ethereum, permit anyone who wishes to participate as a miner, without any kind of vetting or “voter ID”. That’s obviously a problem, because over the internet it is easy to pretend to be a hundred or a thousand or a million people if you want to (a tactic evocatively referred to as a “Sybil Attack” in the blockchain community). To prevent that, instead of “one person, one vote”, blockchains are something like “one GHz, one vote”. Ones weight in the parliament is determined by how much computing power one can bring to bear, and, it turns out, sometimes by the form of the computing power, as Bitcoin for example is best run by very specialized chips. The legitimacy of blockchains, as of more traditional parliaments, derives in part from notions of participation or at least representation, and also from expectations that they will honor and reinforce communal norms. The power of blockchains, as of more traditional parliaments, may depend to a certain degree on their continued legitimacy, but might also survive a loss of legitimacy by virtue of network effects. Congress itself produces nothing but a set of official minutes, but those minutes create important social facts because we each expect other people to take them seriously, so we ourselves take them seriously, so the contents of those minutes create important social consequences. The Bitcoin blockchain produces lists of who spent what to whom of an imaginary, artificial, funny money. But Bitcoin users have become willing to surrender objects of real value for appearing on lists of Bitcoin recipients, and as long as we expect that to be true, we must take the blockchain’s adjudication of who owns what seriously. A blockchain, like a parliament, is much more a social institution than a technological one, although very clever technology was necessary to design blockchain systems that could become socially credible. Like political systems, some mix of continued legitimacy and path-dependent coordination equilibria (“network effects”) determine how durably and powerfully blockchains will be able to shape social facts into the future. Continued legitimacy may depend on continued adherence to widely shared norms, on perceptions of fairness and representation, and on how effectively the blockchain’s decisions serve the actual interests of the community that relies upon it. Ethereum‘s is the most interesting and ambitious widely deployed open blockchain, a parliament whose job is to enforce the behavior of social institutions and financial contracts inscribed as directly consequential computer programs rather than in human habit or legal text. That was never going to be straightforward, and the DAO hack has beautifully surfaced some the difficulties and contradictions inherent to the enterprise. More on that soon, I hope.

FD: I am a critic, but also an enthusiastic participant in the blockchain hyposphere. I am financially and professionally invested in the Ethereum project in particular. I am also a DAO token holder, directly injured by the DAO hack. I’m more intrigued than angry about it, though. Nothing that has already been perfected is very interesting.

Update History:

  • 19-Jun-2016, 5:05 a.m. PDT: “to explain what I think ‘blockchains’ are”
  • 20-Jun-2016, 4:50 a.m. PDT: “some mix of continued legitimacy and path-dependent coordination equilibria (“network effects”) determines how”

6 Responses to “A parliament without a parliamentarian”

  1. shm writes:

    I was hoping you would give me another month or two before writing about smart contracts. This comment was supposed to have a link in it.

    The DAO hack reveals a weakness that underlies the programming language in which the contract was written, a fundamental weakness in its type system. In some circles this weakness has been understood for a long time, as well as its corollary: a universe of possibility where the theorem-proving capabilities of modern type systems will meet finance’s insatiable thirst for certainty. Somewhere, in a small team at a large bank, seduced out from the comfort of the academy and into the heart of the machine, a few individuals spend their days developing a new language that they hope will succeed Ethereum’s problematic lingua franca, the weakly-typed “solidity”.

    It’s probably no surprise then that no one ever asks where the name of the “fluidity” project comes from, just as no one asked where its motivating idea came from: that finance exists to create complexity, and therefore a mechanism to more responsibly support the advancement of financial complexity will always be able to find demand. I guess sometimes people make their own sense of things.

  2. Phil Koop writes:

    Is a blockchain “much more a social institution than a technological one?” Yes! Just as a parliament is much more a social institution than a legal one.

    Continued legitimacy may depend on continued adherence to widely shared norms, on perceptions of fairness and representation, and on how effectively the blockchain’s decisions serve the actual interests of the community that relies upon it.”

    Well said!

    Is a blockchain “just a parliament without a parliamentarians?” No. It is a parliament without elected parliamentarians. It is as though parliamentarians were selected according to who could lift the most weight, or run the fastest marathon, or something. But parliamentarians are inseparable from blockchains as they are from any other sort of parliament.

    Is a 51% attack “just another way of describing “majority rules” when you don’t like the decision of the majority?” No, exactly wrong. A 51% attack is another way of describing “minority rules” when the majority don’t like the decision of the minority. We saw that when GHash achieved 55% power for a period; people weren’t too upset because there were no identifiable bad consequences in the moment. Had there been, it would have been a “55% attack.” You have unhelpfully conflated majority in parliament with majority in the population. It is the former that validates blockchains but the latter that determines their value. 100% of the value of Bitcoin derives from its users, not its miners. Perversely, that is exactly the meaning of the “social institution” point!

    The nature of blockchains, considered as social institutions, is that technology will never be able to take a back seat. Just because someone once had a majority of Bitcoin hash power, and Bitcoin did not collapse at that time, does not mean that this is a viable social institution going forward. (In fact, Bitcoin is insecure once someone has > 25% hash power.) Without the normal social sanctions that operate in ordinary human interactions, blockchains are completely dependent on technical soundness for social stability. When soundness is replaced with complacency, the institution will be destroyed, as happened to The Dao.

  3. vlade writes:

    @shm – under Goedel theorem, no smart contract can EVER be bulet proof. Any smart contract is by definition a list of axioms (ultimately driven down to the turing machine). Given that you can reduce (map) any actual instance of a smart contract to an integer, GT applies (smart contracts defined by code can be either consistent, or complete in the provability sense, but not both). QED.

    Hence there will be ALWAYS exploits like these (i.e. not a hack, but finding an area where either consistency or completness fails). It’s fundamentally unfixable, it’s a feature.

    Human language is subject to GT too, but it doesn’t try to pretend it’s precise, or that the contracts in human language are precise and immutable (you could argue it claims completness, but not consistency). That’s what the courts are for, to decide what to do where the consistency check fails.

    Bitcoin, blockchain etc. are trying to impose cyber-reality on “real” reality. Guess who’s going to win, given the constraints of cyber-reality? Especially since, whatever you say, these “smart contracts” are in reality ruled by the “dumb contracts” known more commonly as civil and criminal law codes as applicable.

    That doesn’t mean there’s no use for technology like that – although I’m not sure what it would be right now.

  4. thaaanos writes:

    GT doesn’t rule out the existence of provable Theorems, hence it says nil about exploitability of smart contracts. It simply says that some theorems may never be proved or that some contracts may never end evaluating. Both attackers and normal users like their contracts to evaluate.

  5. jpt4 writes:


    Do note that Bitcoin’s Script [0] avoids the Goedelian trap by being strictly sub-Turing, and that total functions have been studied as an alternative foundation to partial functions for practical general purpose programming [1].

    [0] https://en.bitcoin.it/wiki/Script

    [1] https://en.wikipedia.org/wiki/Total_functional_programming

  6. Sam E writes:

    To me, the most salient disruptive potential of blockchain-enabled contracts stems from their circumvention of societal constraints: In society as we have known it, many kinds of contracts are unenforceable as a matter of societal policy, and large payments are generally traceable, again as a matter of societal policy. These policies embody generations of experience that blockchain enthusiasts discount.

    Enforceable contracts with criminal intent and untraceable payment are a threat that has not yet matured, and ransomware will not be the end of the story of blockchain-enabled extortion. Can we really judge the scope of criminal innovation in our brave new blockchain world? What realistic benefit could offset this unfathomable threat?

    We can opt out of blockchain-enabled confusion and fraud, but cannot opt out of blockchain-enabled crime.