[E-Lang] latency compensation

[Norman Hardy]

[This is Norm's summary of the proposed use of hash chaining to simplify 
Pluribus.  This summary is based on a phone conversation, and I believe was 
written without reading the notes I recently sent to e-lang.  I am posting 
this (with permission) as it says many things more clearly than I did.  
--MarkM]


Let me replay the ideas that you mentioned a couple of days ago.

I think that authority to an object within a vat, from outside the vat, 
consists of knowledge of the public key of the vat and the swiss number that 
goes with the object.

The question that you raised is whether an outsider can be trusted to choose 
the swiss number in order know it while the message to create that object 
travels to the vat.

One obstacle to this is that the outsider may choose a number of an extant 
object, or a deceased object. The vat can guard against the first but it is 
a bit of extra work. It is more expensive to guard against the second for it 
requires the vat to remember all past numbers. It must be guarded against 
lest the new object be mistaken for an old object. The outsider may 
legitimately know the deceased object's number but should not thereby be able 
to steal the identity of the deceased object and there by learn secrets 
directed to it from outside the vat.

I suspect that I didn't say anything new to you above.

You went on the mention the arcHash idea. I suppose it goes as follows. The 
outsider chooses a random number N and sends it along with the creation 
request. The hash of N becomes the swiss number and this seems to prevent 
the outsider from colliding with past objects not of his creation. The 
outsider knows the new swiss number early and the vat knows it soon enough.

I don't recall further extensions of the idea.