[Cryptography] Cubbit

[Matt Palmer]

On Fri, Jun 05, 2020 at 09:56:56PM -0400, Jerry Leichter wrote:
> > If you want the software-only version there's things like Storj, where your
> > data is stored on the spare hard drive space of random people who have signed
> > up to it, a.k.a. Airbnb for data.  So bits of your data will be on a gaming
> > machine in Kazakhstan, bits on a cash register in France, and bits on a Kodi
> > box in Peru.  What could possibly go wrong?
>
> The description Cubbit provides says that they distribute 36 copies in a
> Reed-Solomon code that allows recovery from any 24 copies.  But one way or
> another ...  36 copies requires that, somewhere, there be 35 times the
> space of the original copy to provide the redundancy.

I think they might have described it poorly.  There wouldn't be 36 full
"copies" of the data floating around if they're using Reed-Solomon; it would
be 36 chunks of data, each much smaller than the original block of data,
which, if you mash any 24 of them together, you can get back said original
block of data.  If I'm remembering my RS theory correctly, the overhead of
any (n, k) RS code is `k / n`, so in this case there's a 50% overhead in
storage, (but with N+12 redundancy, effectively).

I still wouldn't want to trust it with my data, because I don't like the
idea of the wrong twelve randos turning off their Cubbits causing all my
data to disappear, but the idea, in principle, isn't *quite* as bad as it
might have originally sounded.

- Matt