But in a crypto-ledger system, if a computer in the UK embassy in Kenya is infected, you will still be issuing fake passports. This is effectively moving the problem away from a central server to all of the clients, and I am not sure this improves security.

I would assume that cryptographic signatures are a better protection against tempering.

It doesn't move the problem, it prevents it in certain scenarios or makes it more difficult. The scenario you describe is already a valid method of attack.

A simple git repository with signed tags (assuming the key is well protected) would provide those features.

Decentralized consensus does not solve data immutability. The append-only feature in Blockchain does that. So, you only need an append-only DB, without any form of decentralization, to do the same job.

The article is about blockchain technology (the title is misleading). So the append-only feature is implied, at least I hope so.

Guess the question is why you want a network of nodes, all of which may append a fake record, if you only want no one to change the data.

No can append a fake record without the matching private key needed to sign it.

Yes, the point is that decentralization consensus doesn't help.

Now you're moving the goal posts. It does help because it adding or removing from the chain will be noticed.