Decentralization comes with People!
Every so often in crypto, another data point emerges in the wild that supports or disproves a previous theory or fundamental school of thought. The recent fiasco with Ethereum and its crown jewel proof of concept project, The DAO, was such a data point that made me want to revisit some past debates about decentralization and its misconceptions. The fact that Ethereum was supposed to be decentralized (some argue more than Bitcoin by measures of node operation cost), yet, how the community could be considering supporting a hard fork to break the coin fungibility of its system, in the name of ‘justice’ and making victims whole, stands in the face of everything a good monetary system should be.
The problem with decentralization is that it is waved around like a banner and war cry blindly by the rallying masses normally in a frenzy of hate and rhetoric. The energy and resources expended which can be harnessed and deployed against enemies of a any political movement, so long as the narrative can be phrased in a way that threatens “Decentralization”. In many ways this is how religions operate. If your god is nothing more than an ill defined concept, then mostly anything can be portrayed as blasphemy, and many wars in modern history have been fought over exactly such, essentially political differences.
The problem with decentralization is its ambiguity and unquantifiable nature. It is very much like porn, you can’t describe it accurately, but you certainly know it when you see it. Some recent attempts at quantifying decentralization such as CONOP (cost-of-node-option, or cost of running a full node) only further serve to mislead the public. The problem is that decentralization is ‘lack of centralized power’, but at what point does a power cross the line from being somewhat decentralized, to somewhat centralized? There is no formal definition of the point when a system ceases to be decentralized and becomes centralized. The only thing we know is this, a system is most certainly centralized if it is controlled by 1 person, and a system is maximally decentralized, if it is controlled in part by every human being alive. Not very useful working definitions, to be sure. Ignoring this problem is the first mistake many in the industry make. Trying to replace this innate way of gauging decentralization with a nicer, more tangible, manageable one is the second mistake they make. The third, more detrimental mistake is to then use these proxy measures to optimize the Bitcoin protocol in order to maximize them.
The cost of running a full-node is a nice convenient proxy that some smart people have proposed. I have several big problems with using this measure. While the cost of running a full node is one variable that drives the decentralization topology of a network, it is certainly not the only one. It only addresses the economic barrier to entry for transaction validating participants in the system. While important, it ignores a much larger demographic of control, that which is the users or simple holders of Bitcoin. Designing a system just to minimize CONOP is as folly as making economic policy based solely on maximizing GDP. While a country with high GDP sounds good on paper, it could be the most wasteful, environmentally damaging, and destructive in terms of human lives sacrificed, yet it is still seen to be ‘successful’. One needs to look no further than the economy of the USA, which is based largely off of military development, and the waging of wars to see how maximizing GDP can’t be the simple solution to a better planet. GDP says nothing about the general happiness and well-being of the population. Happiness, like decentralization, is something that we can all agree that every human wants, but it is a hard thing to measure.
The biggest problem with CONOP is that it treats everyone as a potential operator of a ‘full-node’. A full-node is a part of the network providing a bitcoin service which keeps track of the entire blockchain, and helps cooperatively verify and relay transactions on the network. These are important parts of the ecosystem, but they are by no means the only group that have a stake in the system.
If you subscribe to the theory that every ‘stakeholder’ of the USD system is a holder of US dollars in any form, then you will also see that every holder of Bitcoin is in part, a stakeholder*. These are the common users of the system, the holders of the currency, the people who spend or accept Bitcoin as part of their ‘real’ businesses. These are the people who are affected by the success or failure of the system, and thus they have built-in incentives for acting in its best interests. This is a generalization, of course there are outliers and malicious actors as well, but by far and large, holders of a currency want to see it continue to be useful. This power that they wield, while I would not go as far as to formalize it into the consensus model itself (due to malicious actors), is useful as a deterrent from government persecution. You cannot ban liquor if everyone in the country is drinking it. Well, I suppose technically you can, but you will have riots in the streets. And these days, thanks to the internet, riots and police action, or other such blatant trampling of civil rights are much harder to silence and censor from the public eye. Political unrest is a lot harder to stamp out compared to the 1920s and the years of prohibition. The governments of the world are powerful, and the only thing they fear is threat of rebellion. Civil unrest leads to police action and brutality, which leads to martyrs, which leads to open rebellion, which eventually, as history teaches us, leads to the crumbling of regimes. This, for the most part, is why having a decentralized system of money is important. This, is the peoples hedge against incompetent or abusive governments. This is our ‘air bag’ to cushion the blow of an economic engine run amok whether by malice or by accident.
So back to decentralization. If we can’t rely on proxy measures like CONOP to measure decentralization as it doesn’t give us a total picture, just as GDP doesn’t give us the big picture of a countries prosperity, what can we use? In the past I have proposed a more focused measure, CCR, or Cost of Corruption. Indeed the amount of money it takes to bribe enough actors in the network for your own personal will to be enforced *IS* exactly the attack we wish to defend against, in the pursuit of decentralization. But that is almost as hard to measure. How do you know how much someone is willing to take a bribe for, without trying to bribe them? This time around, I would like to introduce another concept. Not a measure, but a thought experiment. Instead of taking the top down approach as with CCR, defining the goal and wondering how we get there, this time let’s consider some ground up situations and whether or not they would seem to achieve our goal.
Let’s start with a thought experiment:
Which scenario would be the most decentralized?
- 1,000,000 people running full nodes in their own homes, with 1,000,000 independent people using bitcoin (about a small city worth of people)
- 10,000 full nodes running independently, but with 100,000,000 people using bitcoin (about the size of a G20 country)
- 10 full nodes running, everyone on the planet using bitcoin
Considering this problem will reveal to you what decentralization really means. Advocates of CONOP and most core devs would say the 1st scenario is best. More full nodes the better. But as we have seen with the recent Ethereum problems with the DAO and the subsequent immature way they are handling the proposal of a hard fork — which clearly is breaking the fundamental fungibility of their bearer token — , just because they have more people running full nodes independently, it doesn’t matter much to decentralization if they all belong to the same demographic or political affiliation! If you have 1,000,000 16-year old geeks contributing to your network, it is arguably less decentralized than 1000 random people on the planet earth selected from all walks of life, from different backgrounds and different cultural traditions. CONOP does not measure this. It is this diversity in demographic of your network participants that is an ** factor in determining decentralization.
I would thus argue that case #3 is the most decentralized of the choices (ignoring the technical issues surrounding 10 nodes serving everyone, it is a thought experiment after all). It is because in each of these scenarios, if I were to imagine a (or all) world government(s) becoming hostile towards Bitcoin, which situation would Bitcoin be completely to state sanctioned persecution? Answer: case #3. Sure, those 10 node operators would be quickly threatened, closed down and jailed, but the important thing is that if everyone on the planet was using (or holding) Bitcoin then other nodes would pop up immediately, spontaneously. Not only that, it likely wouldn’t ever come to this point, as NO government in the world is stupid enough to pass a law that would immediately spark a rebellion. Of course, case #3 is an extreme scenario cooked up for a thought experiment, the ideal maximally decentralized configuration would likely be somewhere closer to case #2. The point again, is that decentralization, like porn, is something that we can only know for sure once it is observed, or the network is threatened. At which point, unlike what many proponents who believe that would be too late, I believe if enough people are using Bitcoin, then the network will find a way to spin up enough honest nodes to fight any degree of persecution. The key is having that user base who will demand decentralization and independence, and gladly pay for it.
This is why, while I respect those who believe that enhancing privacy or anonymizing Bitcoin is the best defence against the dark forces of the Men in Black, I believe that is inferior to the more historically proven way to get a maximally decentralized system — to get as large and as diverse a user base as possible. Once that is done, the system won’t need caretakers to protect it. The masses of people will protect it themselves.
*Note, I don’t advocate using holdings as a PoS consensus model
Originally published at https://www.wallstreettechnologist.com on July 11, 2016.
