[lxgr]

> In Germany it is. Nobody else but government offices (federal, state, county, municipal) is allowed to use them.

That still does not make it "government information". It's a primary key that (currently) may not be used by the industry. But by itself, it does not identify anything or anyone. In that sense, it's even less sensitive than a name; realistically though, if it's widely stored next to the associated name anyway, it's effectively the same as a name in terms of sensitivity.

A credit score is effectively a database shared across the financial industry. As such, it needs some sort of primary key. That can be either be something globally unique, like an SSN or equivalent, or a wonky composite primary key (first_name, last_name, date_of_birth, last_known_address) which will cause lots of false positives and false negatives:

What if you change your name? What if you move? What if there's somebody with the same name born on the same day in the same city? What if the spelling of your city has changed between your birth and your time of requesting a loan/credit card?

You may object to the idea of credit scoring in general, but being ok with credit scoring, yet objecting to the usage of a sane primary key to do it, makes no sense to me.

[usr1106]

If a reliable primary key exists, it can easily be misused for many purposes: On the more benign side replace the Android Advertisement ID. On the malign side registering political opinions of citizens.

I don't want to live in such society. Yes, bad things can and have be done before. But making them simpler, cheaper, and more scalable needs to be avoided.

[lxgr]

Fuzzy primary keys might have been a deterrent in the past century, but I seriously doubt that they'd stop anyone today from creating detailed user profiles. Not having reliable primary keys is a technical non-solution for a regulatory problem.

The much more effective solution here is to regulate businesses in when they can request/use somebody's primary key and/or other PII, and to simply not allow it in any case where a pseudonymous identifier or partial information (e.g. only somebody's approximate age rather than their full date of birth) would do just as well.

[usr1106]

I live in a country where a unique identify number exists (social security number). People that understand something about information security would know that knowledge of a primary key is not authentication. It has not deterred the government or the courts to accept that knowledge of the social security number makes any contract valid. Example: If someone takes a loan with your number, you pay it back. You could argue that's not the fault of they key, that's the fault of the government and the courts. I have seen so much stupidity here that I am convinced that the traditional West German standpoint that a unique identifier violates human dignity makes sense. That Germany forgets their history is a pity.

[lxgr]

So your actual objection to SSN-like numbers isn’t that they’re bad for privacy but rather that they’re a poor bearer token authentication mechanism? I think nobody was ever arguing that.

And Germans arguably aren’t “forgetting their history”, they are just regulating to achieve desired outcomes (no government and corporate privacy invasion; strong authentication where necessary), not mechanisms (no unique identifiers).

Times and technology change, so why uphold an old (interpretation of) law that is neither necessary nor sufficient to achieve the desired outcome in the present day?