It's not a redefinition, it's a definition that has been around since the 1990's [0], the only requirement for a blockchain is A) directed linked list (block points to previous block) and B) cryptographic security (use hashes as block index)
Bitcoin is merely the first application to use Blockchain in a decentralized proof-of-work manner.
It is a blockchain is you don't merge since you have a commit pointing at a previous commit, (block points to block), merges point to two parents but Ethereum has that almost with uncle blocks.
The commit log of a repo is almost certainly a blockchain, which is why I mentioned "to some extend"
Think of a blockchain as only one strand (say, the leftmost strand, from a leaf all the way to the root) of a Merkle tree. Think of a git repository as the whole tree. So no, they are not equivalent.
If they were there would never be a way to have more than one branch depend on a previous one.
As mentioned, the git commit log, not the entire repo, which is more organized like a merkel tree, though that is only really the storage.
The commit log is a strict DAG, if you don't have any merges, it is a strict blockchain, otherwise it gets fuzzy but is close enough.
I never said it's equivalent but that is what a blockchain is; "A blockchain [...] is a continuously growing list of records [...] which are linked and secured using cryptography."[0]
This doesn't mention "Can't be a merkle tree" for one, but also doesn't mention "can't consist of branches which merge again" or "requires proof of work" or "must use signatures" (cryptography refers to the hashes normally but can refer to anything else too)
Git also acts like a blockchain, if you change the past (edit a past commit) you need to update all following commits for the new hash and their children hashes and anyone who has the repository must do a force pull, similarly to how Bitcoin rejects fork chains with less proof-of-work.
It might be worth mentioning that Bitcoin also uses MerkleTrees for storage as they are very good for quickly comparing sets.
A blockchain does not necessarily process financial transactions at all and I don't see why it must be a requirement.
The common consensus for the term "blockchain" is not based on the 1991 paper, but on Bitcoin's use of similar technology to make a distributed ledger. The term itself is relatively recent, and was not used until well after Satoshi's paper. For example, Harvard Business Review describes it this way:
"At its most basic, blockchain is a vast, global distributed ledger or database running on millions of devices and open to anyone" [1]
That incorporates at least two concepts that are typically associated with the term "blockchain" that go beyond the basic cryptographic data structure: they are distributed, and they are open. Lots of technologies use a similar data structure; git uses a similar data structure. But when a silicon valley startup comes to you and pitches you on their "hot new blockchain tech" they probably don't mean git.
>but on Bitcoin's use of similar technology to make a distributed ledger.
That doesn't really change a thing. Bitcoin merely uses a blockchain to ensure consensus, if all nodes are trusted then you don't need proof of work and you can switch to simply linking the list (see Git or Ethereum Testnets), and the contents of the blockchain are mostly irrelevant too (you can put anything on there, transactions are merely one type of thing you put in there)
At it's most basic, a blockchain is just a directed graph of nodes where a cryptographic method links two nodes together in a child-parent relationship. Everything else is just added fluff for the specific usecase.
That's definitely the definition some companies are operating under.. but really?
WALs and journals have existed for ages and both fit the definition of a directed list of blocks.
Hell, ext2 and FAT fit that definition (inodes forming a directed list of block-references).
I guess we all store our data on something that's technically a blockchain by your definition.