5 ms·

Facebook's blockchain will most likely be a consortium of corporate validators. In this case you could say that the blockchain is "just" a normal distributed ap

• wz1000 7y ago
  Every git repository counts as a blockchain by that definition.
  • pauletienney 7y ago
    except for "append-only" and "verifiable records"
    • wz1000 7y ago
      Git is "append-only". Just like a normal blockchain, you can't modify an older commit without rewriting all the commits since then. Git also supports signed commits.
      • mbrock 7y ago
        So Git could be a somewhat reasonable backend for a cryptocurrency... if only its hash function weren't broken.
        • frutiger 7y ago
          Not quite since git purposefully excludes the computationally hard proof of work requirement. In git you use other external signals to determine which fork to select in the case of diverging histories. With blockchain, all participants are untrusted, so you just go with the longest chain by convention, and rely on the fact that it’s extremely difficult to build such a chain.
          • mbrock 7y ago
            Yeah, you would need to overlay Git with a client that selects a repository’s master branch based on a PoW scheme included in the commits.
  • mbrock 7y ago
    Indeed you could even think of them as embodying the "proof of work" principle in that you naturally look for the branch with the most work done to find the true Schelling point fork for your own contributions. (By the way, I often see these two points in the same thread: "blockchains should be replaced with centralized SQL databases" and "Git repositories are just like blockchains" which makes me wonder if Git repositories also ought to be replaced with centralized SQL databases.)
    • bigiain 7y ago
      Aka Perforce...
    • rubber_duck 7y ago
      Well git model has a lot of issues with scaling repo size - git extensions like GVFS and LFS are used to remedy the problem but they break the decentralized part so you have this hybrid solutions.
    • wz1000 7y ago
      > blockchains should be replaced with centralized SQL databases Blockchains whose tokens are subject to the authority of a central issuer anyway should be replaced with centralized SQL databases. Bitcoin/namecoin/etc. can't be replaced with centralized databases. FBCoin can, as can any blockchain that is supposed to track meatspace resources(since physical resources are subject to the authority of the local government/men with guns anyway). > Git repositories are just like blockchains Nope, I don't consider this to be true. Blockchains are supposed to have decentralized consensus and allow arbitrary nodes to propose changes.
      • mbrock 7y ago
        The purpose of using a blockchain scheme to issue a centralized liability is probably to increase trust in the system due to the public verifiability of issuance and transfers.
        • wz1000 7y ago
  • bigiain 7y ago
    Yes. Which is oe of the reasons my eyes roll so hard when "business types" and "idea guys" decide they need to explain "blockchain" to me, but have never heard of Merkel Trees...
    • tromp 7y ago
      Merkle Trees, named after Ralph Merkle https://en.wikipedia.org/wiki/Merkle_tree https://en.wikipedia.org/wiki/Merkle_tree
  • momokoko 7y ago
    Merkle Tree.
  • Shorel 7y ago
    Only if a javascript file counts as an entry in a ledger.
  • stale2002 7y ago
    Yes! They are! Git repos are absolutely Blockchains! People don't understand how simple the concept of Blockchain is.
• drexlspivey 7y ago
  Your definition is missing the "block" from blockchain and it's also missing a concensus protocol.
  • mbrock 7y ago
    Well obviously a distributed database needs a consensus protocol.