So these dudes both "saw" FB and cryptocurrencies.
Hate as much as you want on FB and cryptocurrencies, it's still quite a feat to have foreseen both.
Regarding storing parts of the secret here and there: it's basically and "m out of n" scheme AFAICT. There are many variations of this but the overall idea is that you can afford to lose (n - m) parts and yet you'll be able to recover the secret. And if a thief were to steal parts, he'd need m parts to be able to recover the secret.
The difference between a Shamir scheme and a true multi-signature scheme is that the former requires combining the m-of-n pieces to reveal the single private key to sign a transaction, which is a huge vulnerability and single point of failure.
Bitcoin script allows the m-of-n signatures to remain geographically dispersed, each signing the transaction with only their own key, so no single party ever needs to possess the full private key.
In a exaggerated 100-out-of-1000 scheme, even if you steal 99 out of the 100 required shares, you still have zero information. You are better off trying to brute force the value from scratch than trying to use the 99 shares you have.
It's also very simple to implement, making it my favorite algorithm.
[1] https://en.wikipedia.org/wiki/Information-theoretic_security
The Winklevosses came up with an elaborate system to store and secure their own private keys. They cut up printouts of their private keys into pieces and then distributed them in envelopes to safe deposit boxes around the country, so if one envelope were stolen the thief would not have the entire key.
https://www.nytimes.com/2017/12/19/technology/bitcoin-winkle...