october 2nd, 2022

whitepaper circle #1: dag bft

• learning method: going through the slides, making sure i understand everything, and rewording into my own words so i can read over it in the future and explain to myself

• byzantine fault tolerance:

  • fault resilience of decentralized stuff to bad actors and machine failures

    • how to have n participants w/ f bad actors agree on infinitely growing actions

    • honest is tracked as n >= 3f+1

• proof of work consensus

  • what is a nonce number that blockchain miners are solving for. When the solution is found, the blockchain miner that solves it is given the block reward.

  • miners prove they executed transactions by guessing a nonce

  • cons of pow

    • high energy consuption

    • high barrier of entry

      • capital hardware costs for profitable mining

    • pros of pow

      • provable security

      • adjustable network difficulty

      • neutral compared to pos

    • proof of stake

      • validators randomly chosen to create blocks, confirm which ones are correct, validating the transactions

      • stake a large amount of tokens to prevent bad activity

      • cons: relies n inherent value of token instead of computational power)

      • pros: less computational power, lower barriers of entry

    • definitions ~ finality

      • finality is the affirmation that all well-formed blocks will not be revoked once committed to the blockchain.

      • (eg. if a malicious actor is able to accumulate 51% of mining power, they can conduct a double spend attack).

    • block finality

      • achieved at checkpoints needing 2.3 of validators

      • examples of bad activity

        • double signing blocks (bad for network security because its when a validating entity (private key) submits two signed messages for the same block.

          • probabilistic finality:

            • over time

            • eth

          • absolute finality

            • finalized once included in a block and added to a chain

            • tendermint needs 2/3 of validators to prevote

            • economic finality: block reordering or transactions not being final = economic penalty !!

DAG BFT consensus:

• consensus logic is SEPARATE from network comms

• no blocks … bundles of transactions

• in round r, each validator sends all other validators a message w/ vertex

• validator replies w signature if it ha