Degrees of Decentralization: August 6, 2017 Snippets
As always, thanks for reading. Want Snippets delivered to your inbox, a whole day earlier? Subscribe here.
Today’s theme: Is being “decentralized” an all-or-nothing proposition? Can a system be “somewhat decentralized”? How much is enough?
Let’s recap the last few weeks:
We’ve talked about the idea of truth and skin in the game- what is the truth, and who gets to decide? How do we agree on what is true? What kind of skin in the game should be required in order to participate? Is decentralized consensus essential to a healthy functioning society in the future?
We then talked about the differences between proof of work and proof of stake as two candidates for decentralized consensus protocols — ways for us all to agree on what is true without the trust required in a centralized system, with a formalized skin in the game mechanism.
And last week, we took a detour to look at the SEC’s recent ruling, where they explicitly stated that tokens whose value depended on the actions of a select management group — in other words, clearly not a decentralized system — will be treated as securities under US law. All of this brings us to a core question that we’re now ready to address. If we agree that for some applications, decentralized consensus is a desirable goal to pursue, then what is “sufficiently” decentralized? Can a system be kind of decentralized, but not really? Is it an all or nothing proposition?
Ethereum founder Vitalik Buterin approached this question earlier this year, in a piece that walks through several distinct flavours of decentralization — architectural decentralization (is the system made of a lot of independently contributing parts?), political decentralization (how many individuals or organizations are ultimately in charge) and logical decentralization (can the organization have multiple ‘states’ at once? If you cut the system in half, will both halves continue to operate independently?)
The meaning of decentralization | Vitalik Buterin
He then describes three attractive features of decentralization: fault tolerance (resilience to bugs), attack resistance (resilience to outside malfeasance) and collusion resistance (resilience to internal mischief). In his words:
“Fault tolerance — decentralized systems are less likely to fail accidentally because they rely on many separate components that are not likely [to fail all at once].
Attack resistance — decentralized systems are more expensive to attack and destroy or manipulate because they lack sensitive central points that can be attacked at much lower cost than the economic size of the surrounding system.
Collusion resistance — it is much harder for participants in decentralized systems to collude to act in ways that benefit them at the expense of other participants, whereas the leaderships of corporations and governments collude in ways that benefit themselves but harm less well-corrdinated citizens, customers, employees and the general public all the time.”
In each of these examples, “fully decentralized” is actually a lot harder to define than it initially seems. Fault tolerance is an easy example: an airplane might have four engines, so you might say that the job of keeping the plane in the air is decentralized. But what if all four engines are made in the same factory? What if they’re all consuming fuel from the same tank? What if they all run the same software, and that software has the same flaw? Same with attack resistance: if a foreign power tried to infiltrate and co-opt Bitcoin mining, although it’s true that Bitcoin miners are “decentralized”, a lot of them are pooled together — and almost all of those pools are in China. So how decentralized is it really? And finally, with collusion resistance, we really get to the heart of where skin in the game is important: as Vitalik puts it, it’s quite hard to even define collusion in the first place, beyond “coordination that we don’t like”. The SEC’s ruling last week leaned heavily on this point: do participants put their trust in a management team, and have limited voting rights? Traditional securities that meet these definitions are very much not decentralized, so we can understand why SEC oversight is necessary. But how much decentralization is sufficient? Is this something we can measure?
Balaji Srinivasan and Leland Lee of 21 took a stab at it last week, articulating a new approach to quantifying decentralization in blockchain networks:
Quantifying decentralization | Balaji Srinivasan & Leland Lee
They use the simple and intuitive concept of the Lorenz Curve, known to most along with its cousin the Gini Coefficient as an indicator of income inequality. As it turns out, “inequality” and “centralization” (in our context) are effectively the same thing: a group in which one participant controls a disproportionate amount of the power is completely “unequal”, but could also be thought of as simply “centralized”. Srinivasan and Lee use this methodology to quantify the overall cryptocurrency community, and Bitcoin and Ethereum specifically. They break public blockchains down into six subsystems that can each be described in this way: Miners (what % of mining rewards go to the top % of miners?), Clients (what % of Bitcoin users use the top clients, like Bitcoin Core and Bitcoin Unlimited?), Developers (what % of commits to the codebase come from the top % of miners?), Exchanges (what % of transactions happen on the top % of exchanges?), Nodes (what % of nodes are in the top % of countries?) and finally, Token ownership (what % of tokens are owned by the top % of people?).
Although it’s an early attempt, it’s a very useful one: we finally have a way to put a number on the amount of centralization (or the amount of inequality, if you prefer!) inherent to many different aspects of a blockchain ecosystem. We can argue about the relative weights and merits of each category, or whether or not any other categories should be included, but it’s a good start. Importantly, it lets us describe some number, which they call the “Nakamoto Coefficient”, which represents the smallest number of units of “something”, whatever that might be, that represent the easiest path towards compromising the network through successful attack or collusion. Again, we have to remember — we’re talking about blockchains and cryptocurrency today, but it may well be the foundation for a lot of our infrastructure and societal security tomorrow. Next week we’ll talk further about this idea of attack versus defence, paths towards being compromised, and what “deep security” might mean for critical networks and infrastructure — like our electrical grid, voting process, and more — in the future.
In with the new:
The future and the process | Andy Weissman
Cryptocurrency’s Netscape moment | Elad Gil
Why is ARKit better than the alternatives? | Matt Miesnieks
We survived spreadsheets; we’ll survive AI | Greg Ip, WSJ
The culture desk:
My buddy, Sam Shepard | Patti Smith, in The New Yorker
Aziz Ansari on quitting the Internet, loneliness, and Master of None | Mark Anthony Green, GQ
Inside Patreon, the economic engine of Internet culture | Adi Robertson, The Verge
How Spotify is ending the Loudness Wars | Ian Shepherd
Inefficiency Week at The Ringer: a look at what gets lost in the chase for efficiency
Allen Iverson, iPhones and Inbox Zero: what we lose when we obsess over efficiency | Ryan O’Hanlon
The Three True Outcomes and the end of inefficient baseball | Michael Baumann
Is our on-demand culture really making lives easier? | Molly McHugh
Other reading from around the Internet:
Migrating Premise to Google Cloud | Giri Anantharaman
Voronoi diagram of people enjoying a park | Jason Kottke
Evolutionary public health: a new series | The Lancet
Biological Markets: the secret economic lives of animals | Ben Crair, Bloomberg
The founder’s null hypothesis | Tom Tunguz
Safeguarding electronic communications: understanding the evolving duties | Law Technology Today
A small defence of the Twitter thread: it’s like a poem | Dieter Bohn, The Verge
In this week’s news and notes from the Social Capital family:
We’re delighted to share news that Hustle has raised their Series A, led by Social Capital, to further their mission to help organizations of all kinds become meaningfully and personally connected to their communities through personal conversations that scale.
Announcing Hustle’s Series A | Roddy Lindsay
Hustle scores $8M to kill telemarketing with personalized texts | Josh Constine, Techcrunch
A bit like CRM for SMS, but a lot more than that: Hustle is foremost a tool to help organizations treat groups like communities. From grassroots political activism to community fundraisers to corporate sales initiatives and everything in between, Hustle is a way for one-on-one text conversations to scale up to thousands or even millions at a time without losing their personal, authentic purpose. With valuable lessons learned from political campaigns, including the Bernie 2016 campaign where Hustle played a critical role, the team is now ready to bring that same organizing power to organizations of all shapes and sizes around the world. As Roddy Lindsay, co-founder and CEO, put it in Hustle’s Series A announcement: "We all share a vision of Hustle as a category-defining communications platform for which positive social impact and business success go hand in hand, built by a team representative of the humanity we serve." This vision is reflected in Hustle’s core values, which they live by every day:
Authenticity: open communication and a culture of inclusion that allows all people to be their authentic selves.
Impact: accountability to measured progress and meaningful impact.
Respect: treating users, partners and teammates with respect, approaching everything with genuine empathy.
Empowerment: empowering teammates to achieve excellence, collaborate towards shared goals, and build shared trust.
We’re really lucky and fortunate to be able to partner with a team with such an important mission, strong values, and impressive execution. In an age where everyone has a megaphone and noise is drowning out signal progressively more each day, it’s crucial for tools to Hustle to exist. But it’s even more important for the team that builds them to share a strong set of values that serves the greater cause of advancing the arrow of human progress. If you want to join the Hustle team, they’re hiring for a number of positions: software engineering in infrastructure, data, platform and product, communication design, and several sales positions across San Francisco, New York and Washington DC. And if you or an organization you know might want to use Hustle, you can request a demo here and a Hustle team member will get in touch.
Have a great week,
Alex & the team at Social Capital
