Re: [Gzz] hemppah's research problems document

[b . fallenstein]

> > To solve that problem, we need to do tree hashing (meaning that we need
> a
> > new kind of block id).
> > 
> > -b. 
> > 
> 
> So, this *might* be an issue ?

Hm, you can look at it like this: It is an issue, but probably the solution
is really straight-forward. I suggest a structure similar to bitzi's
urn:bitprint: 
(IIRC).

> I went and searched some basic information about Tiger Tree Hashing (which
> is
> tree based hashing algrorithm). I found this from slashdot.org:
[snip]

I suggest you first read about Merkle hash trees-- a good crypto book should
have that, I think; then about THEX, which is a specification for 
creating Merkle hash trees (pretty straight-forward); and then about Tiger,
which is a message digest algorithm like SHA-1. Tigertree hashing then is 
simply THEX using the Tiger message digest.

The important theoretical part is Merkle hash trees; Tigertree is simply
that with a few options selected (digest=Tiger, block size=1KB and so on). 
That's important for practice so that different implementations will be able
to interoperate (if they used different digest algorithms or block sizes,
they 
wouldn't be).

Of course, if you know all about tree hashing already, there isn't much left
to do.

- Benja