5 ms·

Yeah, so if you look at some commentary around that paper there is a variant of the problem with "approximate agreement" that can solve for example the 3 person

Yeah, so if you look at some commentary around that paper there is a variant of the problem with "approximate agreement" that can solve for example the 3 person variant. This fits with the theory (really just a hypothesis at this point but actively worked on) that I have w.r.t. datalisp. Edit: see for example slide 30 of this slideshow: https://www.cs.cmu.edu/~15712/lectures/15-bft.pdf https://www.cs.cmu.edu/~15712/lectures/15-bft.pdf I'm not (yet) very good at probability theory so it's a bit slow going filling in the model I have in mind (that is; deciding the message format) but propnets are the correct framework to fit it into.

jude- 5y ago
If you read the Avalanche paper, you'll see early on that all of its claims to being BFT stem from an assumption about the distribution of message arrival rates in the system. That is a very generous assumption that can easily be wrong. Also the original BFT literature makes no assumptions about the arrival times or even reliability of the network -- BFT constraints apply even if the network is 100% reliable. It only gets worse from there; 2/3 honest votes is the best you can do under ideal circumstances.
- openfuture 5y ago
  I am in no way trying to defend avalanche. I have been speaking from the standpoint of datalisp. Again, the Byzantine generals problem of the paper is not an accurate description of what is happening in the datalisp network and does not apply except locally (in a 'neighbourhood' of the dispute) so it's not 2/3 of the whole network but rather 2/3 of the echo chamber that needs to agree... Anyway you seem to be eager to dismiss off-hand so I am losing interest in debating with you as-is.
  - jude- 5y ago
    So is global agreement on state a criterion for datalisp's correct operation, or not? If not, then why compare it to Avalanche at all?
    - openfuture 5y ago
      Could you please read our exchange from your first comment and tell me how you parse it? From what I can tell; I agree with your dismissal of avalanche and offer an intuition for what may be the problem by contrasting it to my work-in-progress idea, then we have a series of back-and-forths where I consistently mention datalisp but never once talk about avalanche.. Regarding global agreement of state, yes it is not a criterion for datalisp. Datalisp approaches the network like a sheaf (locally consistent but globally not necessarily so) and uses the properties of propagator networks (which are fully CP, therefore they converge on a fixpoint) to reason about convergent consensus that executes lazily. The casual dismissal of datalisp (as it stands) is that the message format for communicating your beliefs w.r.t. probability of signal-to-noise is not fully worked out yet and all we have is a series of heuristical arguments. However, datalisp is a data-intechange format first and foremost and optimized for building authenticated datastructures. It is not specialized for cryptocurrencies per-se (since the focus is on second order byzantine fault tolerance).
      - jude- 5y ago
        This would be going a lot more smoothly if you could articulate which distributed systems properties, exactly, you intend datalisp to provide. So far, you're contradicted yourself at least once here: > I believe the problem in the first place is forcing the network to come to global consensus. In my project the idea is to form "lazy consensus" (i.e. just enough consensus for everyone to have a consistent outlook on the world). So, which is it? Not coming to global consensus and "just enough consensus for everyone to have a consistent outlook on the world" seem mutually exclusive. Don't be afraid to lay down a bold system description, and don't hold back on the specifics. Not trying to brag, but I have a PhD in distributed computing, I work on a blockchain at my day-job (https://github.com/blockstack/stacks-blockchain https://github.com/blockstack/stacks-blockchain), and I chair its governance process. I'm pretty sure I'll be able to understand a rigorous system design. But, I don't have very much patience for weasel-wording, pussyfooting, or hand-waving.
        openfuture 5y ago

jude- 5y ago

If you read the Avalanche paper, you'll see early on that all of its claims to being BFT stem from an assumption about the distribution of message arrival rates in the system. That is a very generous assumption that can easily be wrong. Also the original BFT literature makes no assumptions about the arrival times or even reliability of the network -- BFT constraints apply even if the network is 100% reliable. It only gets worse from there; 2/3 honest votes is the best you can do under ideal circumstances.

openfuture 5y ago
I am in no way trying to defend avalanche. I have been speaking from the standpoint of datalisp. Again, the Byzantine generals problem of the paper is not an accurate description of what is happening in the datalisp network and does not apply except locally (in a 'neighbourhood' of the dispute) so it's not 2/3 of the whole network but rather 2/3 of the echo chamber that needs to agree... Anyway you seem to be eager to dismiss off-hand so I am losing interest in debating with you as-is.
- jude- 5y ago
  So is global agreement on state a criterion for datalisp's correct operation, or not? If not, then why compare it to Avalanche at all?
  - openfuture 5y ago
    Could you please read our exchange from your first comment and tell me how you parse it? From what I can tell; I agree with your dismissal of avalanche and offer an intuition for what may be the problem by contrasting it to my work-in-progress idea, then we have a series of back-and-forths where I consistently mention datalisp but never once talk about avalanche.. Regarding global agreement of state, yes it is not a criterion for datalisp. Datalisp approaches the network like a sheaf (locally consistent but globally not necessarily so) and uses the properties of propagator networks (which are fully CP, therefore they converge on a fixpoint) to reason about convergent consensus that executes lazily. The casual dismissal of datalisp (as it stands) is that the message format for communicating your beliefs w.r.t. probability of signal-to-noise is not fully worked out yet and all we have is a series of heuristical arguments. However, datalisp is a data-intechange format first and foremost and optimized for building authenticated datastructures. It is not specialized for cryptocurrencies per-se (since the focus is on second order byzantine fault tolerance).
    - jude- 5y ago
      This would be going a lot more smoothly if you could articulate which distributed systems properties, exactly, you intend datalisp to provide. So far, you're contradicted yourself at least once here: > I believe the problem in the first place is forcing the network to come to global consensus. In my project the idea is to form "lazy consensus" (i.e. just enough consensus for everyone to have a consistent outlook on the world). So, which is it? Not coming to global consensus and "just enough consensus for everyone to have a consistent outlook on the world" seem mutually exclusive. Don't be afraid to lay down a bold system description, and don't hold back on the specifics. Not trying to brag, but I have a PhD in distributed computing, I work on a blockchain at my day-job (https://github.com/blockstack/stacks-blockchain https://github.com/blockstack/stacks-blockchain), and I chair its governance process. I'm pretty sure I'll be able to understand a rigorous system design. But, I don't have very much patience for weasel-wording, pussyfooting, or hand-waving.
      - openfuture 5y ago