On: Planning for My Kidnapping

[yosamino]

I maintain an encrypted vault-file containing some instructions and passwords accessible at a url that is printed on a piece of paper. Along with it is a "share" of a password which I split up using SSS[0]. I laminate those along with a piece of dark cardboard (to make it more difficult to peek) and instructions.

They are distributed among friends and family to be cut open in the event of my death.

It's based on the trust that not more than 5 of those 10 people will conspire against me while I am still alive, but also at least 5 of them will be able to find one another, and unlock the password should I pass.

... good reminder, it's due for an update

[0] https://en.wikipedia.org/wiki/Shamir%27s_Secret_Sharing

[war1025 → yosamino]

I have heard of Shamir's Secret Sharing algorithm a few times in the past six months or so.

Today I finally decided to read the Wikipedia page and figure out how it works.

Really neat. Uses the fact that you need `k` points to uniquely define a `k-1` degree polynomial.

So if you want the secret to be recoverable by any `k` pieces, you just need to define a `k-1` degree polynomial and generate a series of points along that curve, with the secret at a known location on the curve

Basically it sets your secret to the polynomial constant term (i.e. the y value for x=0), generates random coefficients for the rest of the polynomial factors, and then computes however many points you requested. The shares are just encoded (x,y) pairs.

There is slightly more complicated math it to better obscure the secret, but in essence that's all it is. I thought that was really neat.

[ZackJ37 → yosamino]

This is exactly something I intend to add to my document storage platform... I have some notes about it here: https://www.fideliusvaults.com/roadmap/. There's a balance of security and convenience that I've been calling the 'paranoia scale', and this sits on the far end towards 'least convenient, most secure'.