[yellowapple]

> Or just skip the extra complexity and have it all handled by a central authority?

Then every agency would have to trust some central authority. Maybe doable for interstate queries (with the associated federal bloat, but that's hardly anything new), but once you go international that becomes a lot harder.

> Also, who would run the blockchain?

Who runs Bitcoin? Ethereum? Cardano? Dogecoin?

> If they are the only ones who can issue tokens and they have to trust each other anyway, why does it need to be a blockchain anyway instead of some shared database?

The amount of trust required for "I'm reasonably sure this agency issued this driver's license" is far lower than the trust required for "I'm okay with this agency having control over my agency's records".

> Or if you want to keep it distributed, a peer-to-peer distributed hash table or other peer-to-peer database of ID tokens that only the DMV's have write access to.

A.k.a. a blockchain. You don't even need to restrict write access; DMVs can readily ignore transactions by non-DMV entities.

> If all the states share a blockchain, then they could just as easily share a database and web interface that you can look up.

That ain't anywhere near as easy, both politically and technically. This is already something they struggle do to, and distributed ledgers make that struggle entirely unnecessary.

> if there's a centrally mandated blockchain they should be using

There wouldn't need to be a mandate. States are perfectly capable of making agreements with one another, and could opt into a public blockchain at their leisure. CCW reciprocity is a good example of this; states have on their own come to recognize concealed carry permits issued by other states, and a blockchain would be a natural fit for recording and authenticating said permits under that dynamic.

Even assuming the need for a mandate, it's a hell of a lot easier for the federal government to say "hey, all states need to use this public blockchain per this standard" than for the federal government to roll out a whole central database that's scalable enough to handle queries nationwide and say "hey, all states need to use this central database".

> The person checking (bar staff in your example) still need a website or app that they can input the data to derive the hash (as not to leak personal data) and check the hash against the blockchain, this could just as easily work against a central or state database

Yes, and that would - again - require either:

1. a central authority that each state's agency trusts to be the record of truth, or:

2. 50 separate databases with 50 separate APIs.

A public blockchain absolves the need for either; agencies can maintain their licensing data autonomously (and, for that matter, data pertaining to those licenses; e.g. recording traffic citations, additional endorsements, etc., even for licenses issued by other agencies) without needing to deal with all that infrastructure.

[stale2002]

> Then every agency would have to trust some central authority.

They already have to do that! The central authority has to issue the licenses in the first place. Unless you are suggesting some absurd, decentralized driver's license issuing? Even that is begging the question though, as there has to be some centralized process to decide who is allowed to issue driver's licenses.

> A.k.a. a blockchain. You don't even need to restrict write access; DMVs can readily ignore transactions by non-DMV entities.

Alright, you are really stretching the definition of blockchain. Like, lets say someone just mirrors the database. So there is a web interface, and someone else copied the whole thing down.

Does that now count as a "blockchain" lol? Having more that 1 copies, of a mysql database, is a blockchain now?

[yellowapple]

> The central authority has to issue the licenses in the first place.

The US federal government is not at all involved in issuing state drivers' licenses, last I checked. Nor is some international agency at all involved in issuing drivers' licenses within any given country. That's what I'm getting at with "central authority" - specifically, a central authority external to the issuer and/or validator.

> Like, lets say someone just mirrors the database. So there is a web interface, and someone else copied the whole thing down.

Is that person going to continually copy it down? What about data from other agencies? How many people are going to be using your web interface?

What I'm getting at there is that this all seems like an awful lot of work compared to just, you know, including some data in a transaction on any ol' existing blockchain (even Bitcoin or Dogecoin or what have you is good enough for this, let alone something that actually supports NFTs like Ethereum or Cardano) and using any ol' blockchain client/explorer to lookup whatever transaction got printed on the ID card.

[stale2002]

> Is that person going to continually copy it down? What about data from other agencies? How many people are going to be using your web interface?

> What I'm getting at there is that this all seems like an awful lot of work compared

They are literally the same thing.

The definition of what a "Blockchain" is has stretched so much, that having 1 mirror of a database would now fit people's definition of what a Blockchain is.

Once we get rid of proof of work, and transactions, the only thing we are left with is "a database which has mirrors".

[yellowapple]

> Once we get rid of proof of work, and transactions

Neither of which I suggested getting rid of above (though to be clear, proof of stake and proof of burn both exist). Transactions are especially handy when you move beyond merely recording the drivers' licenses themselves and actually start recording things like, say, traffic tickets or new endorsements or what have you - and that handiness is significantly improved further when agencies are sharing a single public blockchain instead of trying to maintain umpteen different internal databases explicitly tracking each other.

Even taking your assumption that a blockchain and a database mirror are "literally the same thing", you're still missing the key difference: that the former gives you the benefits of the latter (or at least the ones relevant to this context) with significantly less effort. It's similar in comparison to, say, using IPFS v. mirroring an FTP server; yes, you can create a bunch of FTP mirrors, and reinvent a poorly-specified ad-hoc substitute for half of IPFS, but it'd be a heck of a lot easier and more robust to... just use IPFS.