And for data loggers not connected to the internet this helps. It would be just as good to put it in a central repo if there was an internet connection available. But if you're logging a refrigerator temp on a container traveling between countries there isn't going to be a cheap option for internet.
Every git commit contains a hash of all the previous commits, and is itself referred to by a hash. So if you change an old commit, all commits descending from that get a new hash, and so a new identity.
If you publish the latest hash in the newspaper, you've locked down all the previous commits also.
You are correct.
We understand that... GP even said:
>> I get the value of having a chain of "blocks" of data
How does a consensus algorithm help when the nodes are not connected to the p2p network?
Speaking of trusted parties, (multiple) of those could be set up using a chain of trust in every network segment that do as much cross-signing/-verification as deemed necessary. Once it goes online again, the performed operations could go on-chain.
Why isn't putting that data (or a hash of it) in a publicly available git repository just as good? You can even publish the newest branch hash in the newspaper everyday if you want.