topynate

I wrote this just over one year ago, but only made it public for a few days. Here it is. I still agree with perhaps 80% of the content. As for the style? Pfah.

Bitcoin Reifies

Blockchains as a natural phenomenon

Bitcoin is real. We can keep it, lose it, give it away or receive it — but we can’t counterfeit it. There is no such thing as a fake bitcoin, nor will there ever be, radical developments in cryptography or quantum computation notwithstanding. Yes, this makes it similar to gold or crude oil. That’s not important. Gold’s been around for a while already; it’s shiny, but then again, it’s rather dull.

What I want to argue is that owning and using Bitcoin is fundamentally different from owning and using fiat currencies, precious metals or any other asset. Bitcoin turns social relations into effectively innate objects and properties. These include “transferable property”, “trade” and “autonomy”.

Is property real? Proudhon called it theft, which, Marx pointed out, is circular: theft, by definition, entails taking of property. But later on in Proudhon’s What is Property?, he takes the time to offer ten proofs that property is impossible, i.e. not a Thing at all. Property, says Proudhon, is made real by the spectre of state violence, the bureaucratic workings of an unjust social order, if your eyes are glazing over I’m right there with you, this is radical stuff for 1840 but I was tired of it before I hit my twenties.

Proudhon may be hackneyed here in 1917+97, but he was on to something: property is a social reality, not a physical reality. Me and my Krugerrands are not connected by a fifth fundamental force; ownership particles do not pass invisibly from me to my coins. I own them in the sense that, firstly, I control access to them, and secondly, no-one is likely to challenge my statement “these are mine”.

Philosophically, we have here a set of objects that we assume to exist in their own right — in this case, people — and a relationship between them — the strong probability that they will not violate explicit and implicit claims of ‘property ownership’. When we take that relationship, and turn it into a new Thing — property — we reify property, which is to say, make it real. So long as we understand what we are doing, all is well; but if we start talking about fundamental or self-evident property rights, we risk committing the fallacy of reification — treating an idea as a physical entity. And if you do that, sooner or later someone who doesn’t care about the mystical ownership particles will upend your worldview and your dressing table while you’re out campaigning for Ron or Rand Paul.

Here’s the fun part: Property means something special when it refers to Bitcoin. “I own gold” has the standard social meaning. “I own Bitcoin” means the following:

I remember (i.e. there exists within my brain) a password (data that, with high probability, exists in no other brain or storage device on the planet) that decrypts a key (a larger piece of data unlikely to be replicated in the observable universe) which can sign a transaction that transfers 5 bitcoin (a complicated mathematical statement with precise definition).

An owner of Bitcoin no longer has the problem faced by the owner of any other transferable asset: ‘will my property rights be respected?’ Holders of Bitcoin own it in virtue of material facts independent of their social relations. The world may turn its back on Bitcoin, yea, its value may fall to fractions of a cent, but those who own it, will own it regardless.

One way of looking at this is that Bitcoin is a reification of property — and not in a good way. You see, there’s reification in the good ‘understand the universe by creating new concepts’ sense, reification in the incorrect ‘consider an idea to be a physical object’ sense, and reification in the evil sense of ‘make a soft human relationship into a hard thing expressed by trade in a market’. Amongst Marxists who think seriously about Bitcoin, this is the modal opinion, expressed with clarity by Michael Betancourt. Where’s the harm? Betancourt:

The Bitcoin proposal does more than eliminate human oversight and legal restrictions from the circulation of exchange values. It is an attempt to replace existing, formally organized, legal systems of protection for all parties in the transaction — the buyer, the seller, and the bank — with one based upon the assumption that the only parties to a transaction who need protection are sellers … The only parties who have significance in such a framework are the owners (sellers) whose decisions about how to limit access find literal form in the DRM itself … The full visibility of Bitcoin transactions enables the tracking of all informal association networks (all exchanges of value) involving any particular Bitcoin …

No social relations at the exchange level implies no enforcement of purchase agreements — but only for the buyer of goods. Reification of transactions and property as data on the blockchain creates informational Panopticism as a natural consequence (cf. Shoshana Zuboff ).

Let the objections stand, for the moment. The important thing about the Marxist view of reification is that it’s always fake. Reification never actually gets rid of the underlying social relation, any more than capitalism gets rid of labour. A bitcoin transaction, says Betancourt, still represents social property relations, but now, thanks to reification, it dehumanizes them.

Bitcoin reifies property

Bitcoin is not a Marxist reification. Bitcoin reifies in the rare sense of ‘ex nihilo, actually create a physical object’. Bitcoin reifies property. Property before bitcoin is an abstraction, a social relation treated provisionally as an object, but never attaining that status (Property is Impossible). Bitcoin quite literally makes property into something physical. Anything that can store a private key and keep it secret, and can use it to create and emit transactions, can own Bitcoin. The relation ‘X owns Bitcoin’ is spatially local and temporally persistent; in other words, it more closely resembles relations like ‘X is made of wood’ or ‘X weighs 20 kilograms’ than it does relations like ‘X is a dollar billionaire’. Property is possible — when property is Bitcoin.

And for all that, it’s still only really interesting to Libertarians, who I somehow made it this far without mentioning directly. At long last, property rights really are self-evident! Big smiles, everyone: retroactively correct is the best kind of correct. Just make sure to keep all your assets in Bitcoin.

But what if Bitcoin reified another abstraction we’ve been using these last few millennia?

Trade is a reification

Recall Betancourt’s criticism that by design, Bitcoin only protects sellers, which is to say, the recipients of Bitcoin transactions. Bitcoin was clearly designed to support protection methods for all involved parties, so let’s look at the underlying, correct assumption: that trade, which is just as much a social construct as property, depends on social enforcement to function. I give you something, you give me something. Either one of us has a variety of methods to cheat the other, so we rely on third parties to cut our risk — either literally present third parties, or the justice system, which is more or less a third party in every legal transaction.

We can do this with Bitcoin, of course — multisignature escrow, reputation networks, all kinds of technological infrastructure to support our social construct. That’s fantastically useful, and can certainly provide buyer protection, but it’s not different in kind from the social systems we were already using.

Bitcoin reifies trade in information

I have bitcoin. You have information. I want to pay you for it, we agree on a price, I send you the Bitcoin and you disappear. I can sue, if I know who you are, and I care to reveal what I was paying for.

There is a better way. The Bitcoin protocol is flexible enough to represent the concept of a trade of bitcoin for data. Recall that the most common kind of Bitcoin transaction is made to a pubkey hash. This means that:

To receive funds, a seller must make public a cryptographic hash of his public key;
To spend received funds, a seller must make a transaction revealing the public key hashing to the previously revealed hash, along with a signature verifying that the owner of the corresponding private key has authenticated the transaction.

By taking the public key, hashing it with a different hash function, and using that as the decryption key, we obtain a simple version of the scheme available under the moniker of PayPub, devised by Peter Todd and implemented by Amir Taaki. Their scheme allows the crowdsourcing of payment for public publication. A similar, but more mathematically complex scheme allows a buyer to pay for information, and be charged if and only if she receives it from the seller. Similarly the seller receives the funds if and only if she publishes the decryption keys for the information. Note that the information can be pre-encrypted with the buyer’s public key, thus keeping it secret from third parties.

What we have described is a reification of a trade. You sell me information, and either I receive it and you get paid, or I don’t get the information and you don’t get the bitcoin.

Who still pays for information in 2014?

Governments, actually. But that’s the wrong question. Who pays for information in 2080?

Bitcoin reifies autonomous software objects

An autonomous software object is:

running code, that
makes decisions about
actions it takes, with external consequences; and
is, to a greater or lesser extent, free of external control of its future behaviour.

One catchphrase for such things is DAO, standing for Decentralised Autonomous Organisation. Such entities run on the blockchain itself: the Ethereum project will issue a blockchain protocol with features designed to support DAOs. A DAO is an autonomous software object, but an autonomous software object is more general than a DAO. One problem with DAOs is that they have to be stupid, because each operation must be verified each time any computer processing the blockchain sees a DAO transaction, imposing a very large redundancy factor. However, DAOs are safe — as in, protected from the outside world. They cannot be destroyed without destroying the blockchain (hint: where we’re going, that doesn’t happen).

Here we reach the crux: off the blockchain, without Bitcoin, autonomous software objects only barely exist. We can treat a software object as autonomous, but it’s the treating that makes it so — it has no recognized property rights, so cannot itself take part in a market economy. A human proxy is required, who is de facto and de jure that software’s owner.

Not only can a autonomous software object have no bank account, it is too simple to reliably interact autonomously with a socially constructed monetary system. There is no way to mathematically verify a bank transfer, generate a new bank account, access balance, confirm receipt of funds, etc. While APIs are available, intelligent human supervision is essential to maintain functionality.

Compare a program controlling Bitcoin. A program running on a server suffers outages, restarts etc. As more copies are run on more servers, in different locations, the reliability of the program increases. The different copies of the program can track each other, and make requests to other agents, for instance to load a new instance onto a new server. When such a program can turn its requests into payments for ‘information’ — which is to say, payments for data that itself has executable or financial meaning — then it becomes the master of its fate. Such a program can pay for its own hosting, manage key security, and adapt to change; in short, be a full participant in a networked market economy. Moreover, it can do this while having no high level knowledge about the nature of its environment as a network of linked computing systems, managed by hypersocial primates. It just needs to pay for the facilitation of its goals.

One experimental cryptographic technology enables a ‘heavy’ process to generate output along with a short, easy to check proof that the output was calculated using a particular program. Such a scheme enables a DAO to outsource its computation to anywhere it likes, and the DAO, running blockchain-verified code, can guarantee that it will pay for such computation. The technique isn’t very practical, but the nice thing about algorithms is that they only get better over time, not worse.

Can the Bitcoin protocol support this? Perhaps. In any case, autonomous software objects do not have to rely on the blockchain for their own computation. As such, Bitcoin reifies autonomous software objects — or it will.

Bitcoin doesn’t reify anonymity

The recipients and senders of each transaction are visible. Even the ones that are hidden through a coin tumbler, certainly aren’t hidden in a way that can be turned into a mathematical theorem. This matters more than is commonly thought. A physical Bitcoin transaction is strictly superior to a socially constructed one; those who benefit from the gap in value between those two parallel transactions have an incentive to turn the former back into the latter. Deanonymising a transaction does just that: if your government knows who you are, it can tell you what to do. As the advantages of Bitcoin over other value exchange systems become apparent, the motivation to turn it into a ‘normal’, socially-determined system will increase. Higher level cryptographic protocols like CoinJoin are a useful stop-gap, until anonymous transactions make it into Bitcoin as a protocol upgrade. Unlike extended DAO support, this is a do-or-die feature for Bitcoin — and the window of opportunity to introduce a change of this magnitude will not remain open forever.

Endgame

One possible future for this world is that of increasing machine intelligence, eventually reaching and then surpassing that of humans. Some, or all, of this intelligence could start out human in origin. This could happen at anytime between now and forever, but once it starts, it’s likely to happen very quickly indeed.

What would happen to the economy in this scenario is something of an underexplored topic. Of course, machine intelligence could be inimical to human life, which would render the economy moot, or a Friendly AI could make market economies obsolete.

If markets do survive, then they will be peopled with a lot more inhabitants than exist right now. That’s because 1) the inhabitants won’t all be people, and 2) you can have a lot more intelligent beings when you run them on dedicated hardware. If physical, planet-based life survives, it will be in the minority.

What will the inhabitants of Matrioshka-Sol trade? I really have no idea, but it won’t be physical. They’re all software objects, even the human ones. To the extent that they need a raw material, it’ll be energy. Matter — any matter — will be a very, very distant second.

What will they trade with? As in, what will be the unit of account? Well, these are independent agents, or independent coalitions of agents. They will have to negotiate settlement of contract, just like we do. They can either rely on currency backed by interlocking social arrangements, which presupposes an enforcement method, a more or less common civilization, and that no sub-human level intelligences remain in full commerce with the markets, or they can use a cryptographic currency — likely one coded right into the computational substrate. And we all know how protocols have a way of sticking around…

Particulars

Author Archives: topynate

Relocation

Minimal solution

Numberwang

Note on the block size limit

Bitcoin reifies – an old post