Decentralization and its Discontents by Sarah Friend (Devcon4)

besides being a software engineer I'm also a co-organizer of a conference called our networks on all aspects of the distributed web this presentation is going to be a little more philosophical than some of the other more maybe hands-on talks it's also a shortened version of a longer presentation so I've tried to keep in all of the most interesting content but I will link to the longer one on the last slide so this is a totally not comprehensive overview of the decentralized space and though this presentation will be hopefully very relevant to blockchains and this is an aetherium conference I think it's good to remember that this movement spans other tech stacks too so we have you know the network layer things like community mesh networks protocol there we have block chains but also things like IP FS application layer this is what we might fit most daps and then we also have this social or governance layer where we might put things like loomio or most dowse this is the agenda for our talk and like I said this presentation is cut down so instead we're going to focus primarily on the first question on the list and we can leave the final two maybe for discussion during the breakout or on conference section so what is decentralization are we sure that we know you might also call this new version decentralization without so many discontents so yeah it turns out to be a more complicated question than you might think one that has many unsatisfying explanations when we look at the word we don't get a lot of help all this word means is not centralization and I'm not the first person to point this out there's a name for this type of definition and it's an ant anemic definition but the problem with an ant anemic definition is you know we have this thing that we can call centralization but what are all these other things that aren't the same thing and also aren't centralization what are the names for all these different not centralized ations the origin of the terms in the mid-1800s Alex de Tocqueville would write that the French Revolution began with a push towards decentralization that became in the end an extension of centralization which I think is possibly still interesting to keep in mind as we confront for example blockchain maximalism it also reminded me of this Hannah Arendt passage which is the most radical revolutionary will become a conservative the day after the Revolution and indeed the word decentralization is still used more often in a political or organizational context than a technical one after I came up with the title for this talk I found out it's also a book one with the subtitle an essay on class political agency and national perspective in Indonesian politics the Google definitions of the word are the transfer of authority from central to local government and the movement of departments of a large organization away from a single administrative centre to other locations so these are both political and organizational definitions and it's possibly ironic to be part of a community that focuses so much on decentralization but struggles so much with governance and also worth noting off the bat that in a political and organizational context the concept of decentralization has been a component of many different political movements some very at odds with one another so for example there's a strong free-market rationale for decentralization this is from an essay on information markets the sentence before it cites Adam Smith and the one after it cites Friedrich von Hayek and the argument here is you know we have all these rational actors independently moving throughout society without any kind of central coordinating but there's also an entire movement called decentralized socialism and this is a passage from an essay talking about how you know Mao Mao Zedong and the Cultural Revolution was a reaction to the centralization of Soviet Russia so really a dizzying range of thinkers have called for some kind of decentralization and of course that's what makes the concept so interesting kind of like universal cick income it has this fascinating cross partisan support which gives it this kind of gravitational pull it's like a dense messy knot that's fascinating to untangle so we may not all want decentralization for the same reason and I can't take credit for this caption but I also don't want to Doc's its creator I'll let you all interpret this phenomenon which is certainly the case that we have competing motivations you might say it's a success or weakness of the decentralization movement you might say we've decentralized our motivations alright so it may be difficult to define decentralization but can you measure it this is one way people often like to it's called the Lorenz curve or Gini coefficient and it's often used to measure the decentralization of wealth so there are a few components of this graph and one of them is this straight line that we call the line of equality and this is the curve that we would get if the distribution of wealth in a system were totally uniform so if 10% of the population had ten percent of the wealth fifty percent had 50 percent etc etc the Lorenz curve is what we get when we plot the actuals and the Gini coefficient is the difference between these two lines this is from an excellent recent article called quantifying decentralization by Balaji Srinivasan and he took the top 100 cryptocurrencies and he plotted them on the Lorenz curve and you will see that they are very far away from the line of perfect equality and again from quantified decentralization Srinivasan describes how a decentralized system is composed of a set of decentralized sub systems so we have you know miners we have the developer community we have the concept of geographical decentralization or governmental decentralization and I think maybe it's helpful here to borrow the term intersectional and to think about intersectional decentralization if we imagine decentralization as like this Cartesian plane of intersecting decentralization x' will find that we can't get the complete picture without considering all of them that none of them really make sense to discuss in the absence of others in the blockchain world we talk about the scalability trilemma so this is from the etherion wiki and you may be familiar with it when designing a blockchain these are areas you have to make trade-offs within it's sort of like a cap theorem for block chains and notably one of them is decentralization with its own very specific definition and here it is that each node on the network should not have to have you know too much compute resources available to it maybe a laptop scalability is one and here it's defined as the network being able to process more transactions than any individual node and security which is the amount of compute resources an attacker needs to have to successfully attack the network so of course decentralization is on this list and ok way to think of that is as a stat that we could lower in order to improve these other qualities so when we talk about scalability in the blockchain world it's almost a catechism to compare it to the transactions per second if something like Visa visa process is 24,000 transactions per second another way to understand that statement is that visa has dropped its decentralization stat to zero and max out its scalability um so in conclusion here are some questions in rapid-fire do we want decentralization I think this deserves to be an open question or at least one we don't forget that it's possible to ask and if we accept that there's no canonical decentralization or no perfect decentralization then which one do we want are there limits to decentralization and if so what are they right now we talk a lot about decentralizing the network and application layer but how often do we consider the centralization of the manufacturing pipeline or of supply chains how far down the proverbial stack do we care about decentralization and finally a recursive question can we decentralize the building of decentralized networks I see specialization and centralization is tightly connected concepts and a technical expert is often a central point of failure for a decentralized network and if pursuit of power tends to centralize the system how can we prevent this from happening to our networks over time what underlines all of this that there are many possible decentralization and perhaps no perfect one but all of them may prove meaningless or become just ideological background noise if not carefully and thoughtfully examined so thank you my name is Sara again that's my handle and this will YouTube link to the longer version of this presentation thank you all so much