It was years ago, already impatient for Byzantium and Constantinople, that I first read an article on Algorand by Silvio Micali. The article gave a hint, a brief indication, of a simpler way forward for PoS than bonds and slashing conditions. As a self-proclaimed intuitive I knew immediately that this was the path forward for Proof of Stake. Silvio's credentials, the constellations, one just knows! The road was clearly articulated - for me albeit in the subtext haha, I'm no cryptographer - but this will do it I thought, this is it. Alogrand will be the true PoS. I was quite excited.
Soon Algorand fell silent and my flighty attention flittered elsewhere. From graphene nanomaterials - still impatient here, wen production ready already! - to graphene based blockchains and Dan Larimer. Everyone has their Utopian checklist and now graphene was pencilled nominally in two categories of mine. Dan writes so well and one senses that he has the clearest of minds, that he has balanced all of the competing concerns, social, technical and economic, that he has arrived at the most efficient, and yes perhaps expedient, configuration possible. There will trade-offs.
If you have no idea what I am talking about but still want to read on watch the YouTube video below. Silvio Micali does a fantastic job of explaining it all including those trade-offs that I am trying to respond to here. I better watch it again myself :)
CS;DW ( Can't be Stuffed; Didn't Watch)
Alogrand Pure PoS: cryptographic magic selects random stake weighted committees at every round. Selection is achieved locally without requiring any negotiation with the network, amazing, and selected nodes get to produce exactly one message for the next block. These are somehow assembled and propagated via relay nodes with finality being achieved as soon as the block is confirmed, that being two seconds or so. Algorand does not plan on Turing complete smart contracts but will offer a simpler instruction set with complex computation off chain, proofs on chain. Probably fancy proofs. Proofs validating that only the expected and untampered code was executed.Just speculating there.
DPoS ala EOSIO: 21 block producers, aka super nodes, are elected by stake to produce blocks in a round. Elections are continuous. Super nodes means fast and beefy. Smart contracts execute on web assembly, with all of the expressiveness, proven libraries and ever expanding development tools that come with it. And of course they are blazingly fast and faster on eosvm. Half second blocks with two second finality on the way. WASM as a platform is only getting bigger and better, and it's not just for the blockchain so there are synergies there for adoption too. Dapps do not need their own infrastructure to execute, rather they run on the ddos resistant battle hardened infrastructure provided by the network, those super nodes.
Micali in his critique of DPoS claims that Eosio has sacrificed decentralisation in the blockchain trilemma, given it up on a technicality. And here Algorand steps in with a new technology, a Godel prize grade technology, wresting from cryptographic theory equations that can marshall an iron clad consensus mechanism. Amazing stuff, magic. I am a fan. In awe and a fan. But not so fast, not so fast.
When weighing the relative merits of PoS versus DPoS it all comes down to this: algorithmically chosen committees versus human chosen block producers. That's the crux of it. Putting the finer and less material details to one side. It's Humans versus Machines.
With the launch of the Algorand network, and in the face of Weiss ratings downgrades, collusion blog posts, etcetera, etcetera I have been revisiting the whole PoS versus DPoS thing, trying to reconcile these two approaches to consensus for once and for all. And I believe I have found an answer ;) A grand unified consensus theory no less! A theory intimated in various channels and chat rooms, in the many texts and tweets that I have been reading of late.
Social structure can only be found by first bringing human beings into mutual relationship and interaction, so that they find social structure in their interconnection. Rudolf Stener Conscious Society Lecture 5 Page 75
Consensus is a social construct. It comes from what we do together in the world. It is external, experiential. Theory is internal. Theory is what we do for and how we orient ourselves. In the same way that language does not arise if we are alone, the basis for consensus cannot arise without interaction. Society is not a theory. Society only exists in practice. After we build it. Together. This difficult truth is already well known to many DPoS communities. We know that the game theory, that rationality even, does not limit the ingenuity or confoundedness with which the game is played. Tweaks to incentive structures cannot be tested. They need to play out in the wild on particular chains, within particular communities. Culture is strange like that. It's phenomenological. Culture resists an empirical approach. Social interaction is influenced by things like belonging, shaming, emotions. Thinking that the mechanics of an algorithm or the edicts of a political theory will establish healthy communities does not make sense. It's a confusion. Some point to the history of communism as an example of the failure to build a society from theory. It is only through our interactions, through our lived experience, that we can bootstrap our communities into something which can sustain ourselves.
If this is true then it follows that an algorithmic approach to consensus, PoW and PoS, cannot be about establishing communities. If not for communities then what for? What use the blockchain machine?
RADICAL TRANSPARENCY VS RADICAL PRIVACY
We have been discussing radical transparency in Eos circles. Real, visceral transparency. Risky transparency. Activist transparency.
If you want to change the world you must be prepared to take a beating. Anonymous speech without consequences to speakers censors the truth though confusion and fake news. There are consequences to speaking the truth and freedom demands we face them. #voice Dan Larimer Reading that I have to mention Julian Assange. Well you just have to.
Some posit a radically transparent future of smaller, more manageable sized communities in which its members can be known. DACs and DAOs. Visualising Eosio as a circle, as the round, the round of producers, imagine the globe covered in circles. Nested, overlapping. Interconnect these circles. Within radical transparency. Without radical privacy.
Smaller communities need chains that can scale down in size, in infrastructure, less nodes. Chains that are protected by their members rather than by stake and value. Adaptable. Trusted chains on trust networks. Identity can be decentralised too, there are alternatives to KYC, they are coming. After all one chain is centralised too no?
Unless that chain is for trustlessness. And for radical privacy trustlessness is exactly what we need. Lock step. Machines. To that end I imagine a single Algorand network spanning the globe. The many connected by the one. Single is better too because the more value it holds the more secure it is for what else can secure a network where there is no trust? Here I imagine a series of random lights flashing across a glob of circles, carrying the messages between them. Can you see it? Crytographic mathematical privacy. One large network. Circulating impenetrable messages that can penetrate only through the membranes of their destined cell walls.
Man cannot live by trustlessness alone.
This is a rewrite of an earlier attempted article on PoS and DPoS. What do you reckon? It's a learning procedure really lol. Correct me in the comments.
I wanted to talk about things like Algorand's ability to select random juries, other stuff too. Couldn't manage it! I will definitely read up on it more though.
Anyways I think UOS and Steemit are perfect since they are fundamentally social platforms, it is all about social consensus! DPoS chains really need to cover that dimension natively IMO. Eos' main problem at the moment seems to be that the community is fractured across social media platforms and languages with no way to easily iterate on social consensus. Hopefully voice will help sort it.