[Remops] Why is 4096 bit Mixmaster consider 'experimental'?

[lists at notatla.org.uk]

> > The 4096-bit mixmaster remailers run today by other people are
> > production if the operators consider them so.  But bear in mind
> > there may be few users as yet.

> Great to hear! So let me ask you this: would you be opposed to a small
> team using your code as the base for a new mixmaster project? We don't
> want to step on any toes here so we're being very cautious.

"My" 4096-bit mixmaster is the previous mixmaster slightly extended;
literally by days of work and hundreds of lines of code.

It keeps all the 3DES stuff and does AES on top and there are
conditional sections like
    if (1024==klen) {
      do-old-thing;
    } else {
      do-new-thing;
    }
for deciding whether to consume 1 block of 512 bytes or 2 blocks of them.

While my toes would be unharmed I think starting from that base would be
a poor engineering choice.  The mixminion project with python looks
more promising.