Bitcoin bothers me a bit :) The architecture inherently favors centralization. Mr. Assange, who is a suspect character in my book, appears to be pushing for adopting this mechanism for archival processes, possibly knowing full well that this, in the long term, engenders 'custodians of truth'. An extant, lightweight, and fully validated approach to do precisely that, maintain inviolable records of morphology of digital artifacts, is ignored. (I doubt he is ignorant of GIT..)
I don't like that mining has become more centralized, but it's understanding since the reward for mining is so big that has incentivated rich people have invest in it. The advantage is that it means the network has more hashing power as result which means it's hard to subvert. Despite this, the blockchain is still preferable to the regular banking system in my view. I never heard of GIT as an alternative to keep an un-hackable decentralized ledger of records over time. I'd love to learn how it works in comparison to bitcoin if you could explain. The link as gibberish to me.
The Blockchain uses cryptography and its decentralized nature means that one would have to not only crack the crypto but corrupt all copies of the blockchain to successfully change its past records. How does GIT prevent past records from being modified?
Well, Git uses SHA-1. It is a cryptographic hash, but of course vulnerable to collisions and already a bit deprecated. But a Git like system with a more robust hash (SHA-3, if you will) addresses that. (That paper that was gibberish explains it.) [p.s. Same with git re. 'copies'. If someone somehow manages to hack the SHA-1 sig for a commit on github for repo x, all the extant clones of the repo still maintain the original records and can challenge it.]
If one were to assert control over bitcoin, it would probably start with miners. Hopefully, new developments will make mining more resistant to centralization, rather than less. Unfortunately the Litecoin and dogecoin approach doesn't seem to have worked, as scrypt ASICs have since been developed and deployed. That said, my understanding is that Assange is talking about the decentralized ledger. Of course, if you control the mining, you control the ledger... A decentralized git repo would do the trick. Looks like someone is working on something called gitchain.
k, just watched the vid for Gitchain. This scheme is addressing availability and per Brewer's CAP theorem, full availability and consistency* in a distributed system is impossible. [Shades of momentum/position here, btw .. ;)] Even (or specially) in a P2P system, CAP holds. Partitions are unavoidable. One must either pick full consistency (but sacrifice availability), or, insist on full availability but sacrifice consistency (which in case of a Git like system, means you may not have access to the latest commits.) Vector Clocks or BC type systems, carry the entire state and thus face the same issues that keeps insects down to a certain size. It can not scale indefinitely.
(Thanks for the gitchain link.) | decentralized ledger I'm not sure I quite understand. If the aim is to have a system where there is an immutable record of changes to a (set of) digital objects, then any content addressable based system, e.g. Mercurial, Git, CamlStore, would do the job. Taking Git as a most familiar system, the DAG of Git affords both 'attribution' and 'timeline' datum. Can the proverbial Uncle Joe or NYT go and change commit records in Github? Let's discuss "decentralized". (I'm on and off the net as you know but will be following up! :)]
No, but someone could at Github could, correct? Isn't the requisite for an immutable record the distribution of the consensus on the validity of the record? My interpretation of the decentralized ledger, is the distribution of not only the data on the ledger, but also the distribution of the consensus mechanism required for changes of the ledger.Can the proverbial Uncle Joe or NYT go and change commit records in Github?
The plan is not only for Gitchain to run across myriad machines, but for it to store some Git metadata–user names and permissions, for example–in a cryptographically verifiable format similar to the bitcoin blockchain. This would allow software developers to prove that their code hasn’t been tampered with–just one example of how the project could be expanded in new ways.
Github personnel can truncate records (as in remove certain commits past time t. They can not go and cherry pick a prior commit's content.). Having mirrors that periodically pull from Github would address that trivially. The consensus bit goes back to my earlier comment re Vector Clocks and the CAP matter posted today. Perfect consensus in the material world is impossible.