Badly designed from the get-go... each server should only serve multiples of different primes...
Have you ever wanted your own integer, one that is uniquely yours and can't be taken away?
Oh sure, you could generate a random integer with your device now, but no one will respect that it's yours until you publish it somewhere - somewhere that everyone can agree on.
Brooklyn Integers!According to their FAQ: Yeah kind of, but also not.
Its literally a number server. It gives you an available number from 0 to infinity and you can use that number as global identifier for something/anything.
If someone tries to register that number for something else, the database will say no.
Initially yes, serving numbers from 0 to Infinity, but people on the West Coast thought that it was a bit mean that the East Coast was dominating the numbers game, so they opened up their own database in San Francisco:
Mission Integers!Do they share keys, so that one knows what nunbers the other doesn't? How can they ensure that the same number isn't registered twice?
Infinity is big, so one server issues only even numbers, and the other issues only odd numbers, ensuring they'll never give the same number out without having to communicate. "Spooooky action at a distance!" (not really)
People on the otherside of the pond thought that the ownership of integers was a bit too US-centric, so someone then invented:
London Integers!How can the London server ensure it never gives out a number that the US servers have? They own the Odd and Even numbers for christsake!
Well, with a bit of aggressive negotiating it was agreed that the US server serving Odd numbers wouldn't give out multiples of 9.
Three number servers exist to give out unique integers to anyone who asks, all without checking each other:
1. Serves Even Numbers
2. Serves Odd Numbers (not multiples of 9]
3. Serves Multiples of 9
Badly designed from the get-go... each server should only serve multiples of different primes...
Just in case this is not a joke comment: If X is your prime base, and Y is the multiple, and Y is also prime, then you can't guarantee that one server won't have Y as it's prime base and X as the multiple. Not unless each server implements an `is_prime(N)` function for each new key, which brings us into the realm of quantum computing...
Umm, if X is prime and Y is a multiple of X, then Y is *not* prime, by definition...
Oh wait, I believe I understand you better now!
Example: First server serves multiples of 2, second server serves multiples of 3, they'll both try to serve 6... so really to make this foolproof, each server should only serve powers of distinct primes... that way there won't be collisions...
Have you ever wanted your own integer, one that is uniquely yours and can't be taken away?
Well, there's an easy way to do that! Just generate a random 256-bit number, and derive an ed25519 public key. That public key is your number. Anyone can verify that it's yours (since only you have the secret key with which you can sign messages), but as long as the secret number stays secret no one can take it away!
Very true, and you'd need to do 2¹³⁰ (~ 10³⁹) operations to even get within a fraction of a fraction of a chance of having a collision[1]!
But the point of this integer server isn't secrecy, it's annotation -- primarily providing a unique integer to a resource that can be uniquely shared over many different projects.... which is something you could also do with a random hash....
Hmm, you've raised a good point and now I feel stupid ;-)
1: https://crypto.stackexchange.com/questions/39641/what-are-the-odds-of-collisions-for-a-hash-function-with-256-bit-output
I mean, if you don't need secrecy you can just use a random 130-bit number and get equivalent collision resistance. That's basically what UUIDs are. Of course doing that misses the point of the number servers entirely (which I suspect is mostly for the fun of it).
That said there are some real world problems that involve getting everyone to agree on a number (most notably addressing). I mean, that's basically all IANA does.
And it's a tough problem!