Even the easy part needs chain of custody for signatures! Public key signatures produced today will probably not be trusted in twenty years. Single-source-of-truth needs something like chain-of-hand-offs, which is more complicated…
You want semi-cooperative model in the sense of not doing anything extra on handovers between different mirroring services.
The problem is still predicting what setup will get well-organised hand-overs on service shutdown (as one doesn’t want to rely on a single project being long-living, that is Internet Archive model which will hopefully work out but has specific risks). And I think mirroring DreamWidth is a lot of traffic, so you want something more fragmented, which might be hard to check for long-term vitality in advance…
I have to agree that even my easy case doesn’t seem to be as easy as I hoped for, I didn’t consider pubkey security for the far future. So we have to crack the hard case first then, but for high traffic services it should work with federated/distributed infrastructure in theory?
«Both copies are equally valid» is still the easy case…
I am not sure full-hard-case will be ever cracked (I could just as easily imagine a federated setup where there is a sea of claims to be relevant to some «location» however this is defined, and then there are different views, and maybe sometimes you get this https://en.wikipedia.org/wiki/Ring_species situation of the «same» chat has fully disjoint views and smooth transitions between them, depending on which moderators you trust more)
I think that with long term enough thinking at the middle scale what needs to be searched for is some kind of commitment to handovers as a praxis… and I honestly have no clue how to find out whether any small community in this direction already exists!
Even the easy part needs chain of custody for signatures! Public key signatures produced today will probably not be trusted in twenty years. Single-source-of-truth needs something like chain-of-hand-offs, which is more complicated…
You want semi-cooperative model in the sense of not doing anything extra on handovers between different mirroring services.
The problem is still predicting what setup will get well-organised hand-overs on service shutdown (as one doesn’t want to rely on a single project being long-living, that is Internet Archive model which will hopefully work out but has specific risks). And I think mirroring DreamWidth is a lot of traffic, so you want something more fragmented, which might be hard to check for long-term vitality in advance…
score 1 · canonical
I have to agree that even my easy case doesn’t seem to be as easy as I hoped for, I didn’t consider pubkey security for the far future. So we have to crack the hard case first then, but for high traffic services it should work with federated/distributed infrastructure in theory?
score 1 · canonical
«Both copies are equally valid» is still the easy case…
I am not sure full-hard-case will be ever cracked (I could just as easily imagine a federated setup where there is a sea of claims to be relevant to some «location» however this is defined, and then there are different views, and maybe sometimes you get this https://en.wikipedia.org/wiki/Ring_species situation of the «same» chat has fully disjoint views and smooth transitions between them, depending on which moderators you trust more)
I think that with long term enough thinking at the middle scale what needs to be searched for is some kind of commitment to handovers as a praxis… and I honestly have no clue how to find out whether any small community in this direction already exists!
score 1 · canonical