https://rhonix.io logo
Join Discord
Powered by
# technical-talk
  • p

    prathyush

    04/18/2023, 7:39 PM
    Monero is strictly better than bitcoin.
  • p

    prathyush

    04/18/2023, 7:40 PM
    The best thing that happened is a ton of money went into research.
  • i

    ikky

    04/18/2023, 7:40 PM
    For me, allowing entry to a network is potentially choosing who has the right to participate based on criteria that are not always objective. Not to mention the censorship risks.
  • i

    ikky

    04/18/2023, 7:42 PM
    Otherwise no interest in recreating copies of Amazon web services or vpn providers.
  • p

    prathyush

    04/18/2023, 7:42 PM
    True. But it is also true that there is a limit to what can be allowed to happen.
  • p

    prathyush

    04/18/2023, 7:44 PM
    Democracy to me is a happy bargain XD.
  • p

    prathyush

    04/18/2023, 7:46 PM
    I also think permissioned blockchains would coexist with permissionless block chains. But the bulk of the activity will be in permissioned blockchain.
  • p

    prathyush

    04/18/2023, 7:46 PM
    Monero or some variant of it will not die.
  • p

    prathyush

    04/18/2023, 7:49 PM
    And the obvious pragmatic point i.e the overwhelming cost of Proof of work.
  • p

    prathyush

    04/18/2023, 7:50 PM
    ZK technology will be at the heart of privacy preservation.
  • i

    ikky

    04/18/2023, 8:12 PM
    I agree with ZK. But to come back to TOR, is it relevant for the sake of relevance to pay for a service that is supposed to help freedom of expression and communication in non-liberal countries?
  • p

    prathyush

    04/19/2023, 4:19 PM
    It's not just non liberal countries. I would want a good solution to anonymity. Tracking every movement of the individual is a huge problem. Most major internet companies are basically spyware behind a garb of legitimacy.
  • p

    prathyush

    04/22/2023, 7:12 AM
    https://twitter.com/mathildepapillo/status/1649433082750185472?s=20
  • p

    prathyush

    04/22/2023, 8:15 AM
    also new research on subject. https://twitter.com/HajijMustafa/status/1649343920613244929?s=20
  • p

    prathyush

    04/22/2023, 8:22 AM
    I believe that simplical complexes will revolutionize the field of AI architectures.
  • p

    prathyush

    04/22/2023, 8:26 AM
    The other very interesting Research direction is neural turing machines. https://arxiv.org/abs/1410.5401
  • t

    Tobi

    04/22/2023, 1:05 PM
    Yes, greg was also talking alot about simplical complexes.
  • p

    prathyush

    04/22/2023, 2:49 PM
    This is an highly speculative idea, but I believe AI architectures will do away with layers eventually. The AI model will transform a simplical complex (or variant) in multiple steps. You can develop algorithms that are agnostic to data shape. Second independent idea, but also related idea is network weights will be stored as content addressable memory/associative memory(or variant). I suspect this will allow for much more generality. Because you are doing are working one order higher in a sense.
  • p

    prathyush

    04/22/2023, 5:22 PM
    Living without images

    https://www.youtube.com/watch?v=N9Z7vkuqdYQ▾

  • p

    prathyush

    04/23/2023, 7:37 PM
    I think looking at unforgables and scoping sematics from a fresh perspective would be useful.
  • i

    ikky

    04/25/2023, 6:16 PM
    I found that https://nymtech.net/ @prathyush
  • p

    prathyush

    04/26/2023, 4:11 AM
    I haven't gone into it, but looks promising in comparison to the other 2 networks.
  • f

    fabcotech

    04/26/2023, 3:37 PM
    I think what nym want to build is almost impossible
  • f

    fabcotech

    04/26/2023, 3:38 PM
    Prooving that you have transmitted a given packet, and get tokens/income out of it. You'd have to make packet transfer of TCP/IP verifiable by other peers, that's hard. And the act of verifying should take less energy/work than the act of transferring the packets.
  • f

    fabcotech

    04/26/2023, 3:39 PM
    Maybe an income system with tokens do not fit mixnets or tor-like networks.
  • p

    prathyush

    04/27/2023, 2:58 AM
    > You'd have to make packet transfer of TCP/IP verifiable by other peers, that's hard. Yes even I realized it, It's probably impossible, because both the sender or receiver can lie. I think you can only build a reputation based system for tor nodes.
  • p

    prathyush

    04/27/2023, 2:59 AM
    But I see no reason why a reputation based system cannot work.
  • p

    prathyush

    04/28/2023, 5:54 PM
    https://probablydance.com/2018/06/16/fibonacci-hashing-the-optimization-that-the-world-forgot-or-a-better-alternative-to-integer-modulo/
  • p

    prathyush

    04/28/2023, 5:54 PM
    Neat 🙂
  • c

    CoolDrT_gb

    05/01/2023, 9:59 PM
    I like this but, not sure where it sits in the performance side!