@janPiteJanseke Thanks -- sorry to be late returning to mastodon, so late that the link 404s now. Do you know of a new link?
"The old promise of the internet — niche communities, human connection, people exchanging ideas, maybe even paying each other for the work they’d made — never really lost its appeal, but this year it came back with a miniature vengeance."
Tiny, weird online communities made a comeback in 2017 - Tiny, weird online communities made a comeback in 2017 - https://www.theverge.com/2017/12/28/16795090/internet-community-2017-post-mortem-tumblr-amino-drip-tinyletter
@taoeffect Good, this is progress from "ordered events"! It would help to state things more carefully up front so we don't have to toot back and forth to narrow false universals or over-general claims into more useful particular ones.
Git and blockchains use specialized Merkle trees or DAGs, but git does not enforce consensus on meaning of contents -- forking/diverging is easy by design.
Ok, let's say Merkle-ization is the common better hammer. Should sync use it? Maybe, but we need specifics.
@taoeffect Blockchains are more than "secure logs" unless you make that definition an axiom instead of a proposition with a proof.
So, what's your proof? It will hinge on defining "secure" in terms of threats and assets.
@taoeffect A blockchain is not merely an ordered list of events. Right?
@taoeffect Cats are mammals but not all mammals are cats. Blockchains are logs but not all logs are blockchains. Crazy but true!
Blockchains have specific costs as well as benefits in the face of distinct requirements and threats. Scenarios that do not face those threats, or have cheaper ways of getting the needed benefits (and not the unnecessary ones) can and do avoid blockchains.
Again, I doubt much two-phase commit to centralized dbs for OLTP will be replaced by blockchain protocols, ever.
@taoeffect Again, not the case with my examples.
Again, you made a sweeping statement: "If your app shares data over the Internet, <blockchain>." No qualification about complexity or number of users.
@taoeffect Scope creep (or jump) again: my examples were not complicated database-ish ones.
If blockchain is the hammer for all nails and pays its way, we should see it take over for centralized databases in all OLTP, a Darwinian sweep of the genome if you will. I doubt that will happen. If I'm right, please don't say it is because I didn't clap my hands three times :-P.
In my experience with programming and distributed systems, there is never one hammer for all nails.
@taoeffect 3 is absurd scope creep and you should retract it. I'm ignoring it for now.
1 and 2 have standard answers (see https://github.com/brave/sync/wiki/Design). If there is a conflict, something has to resolve it. In a client-encrypted system, that something is the client software, without or (if necessary) with interaction & help from the user.
Blockchains for consensus are fantastic, but any conflict among clients breaks consensus and requires client-specific resolution algorithm. More in later replies.
@taoeffect Kind of circular arguing there :-P. Why blockchain? You have 512 chars to argue for the claim you asserted.
@taoeffect Why blockchain at all, if client-secret-key encryption for sync among paired devices?
Very interesting usage (for good!) of the Intel SGX by Whisper Systems: https://signal.org/blog/private-contact-discovery/
Whoever perpetrated 9/11 made all US citizens "political". To swallow the official story and demand allegiance while saying "no politics" engages in politics.
Read the NIST report about WTC7. NIST won't release its model or "data" behind conclusion that fire took down steel framing at free fall. This is enough to cry foul & demand a real investigation.
The way 9/11 and then Amerithrax (US-based whether Ivins or someone else) were used to justify invading Iraq stinks.
Someone lied us into war.
Server run by the main developers of the project It is not focused on any particular niche interest - everyone is welcome as long as you follow our code of conduct!