On Mon, Jun 20, 2016 at 04:21:39PM +0000, zaki--- via bitcoin-dev wrote:
> Hi Peter,
> 
> I didn't entirely understand the process of transaction linearization.
> 
> What I see is a potential process where when the miner assembles the block,
> he strips all but one sigscript per tx. The selection of which  sigscript
> is retained is determined by the random oracle.  Is this is primary benefit
> you are suggesting?
> 
> It appears to me that blocks still need to contain a list of full TX Input
> and Tx Outputs with your approach. Some of the description seems to
> indicate that there are opportunities to elide further data but it's
> unclear to me how.

I think you've misunderstood what I'm proposing. The state machine approach I
described doesn't necessarily require blocks or even miners to exist at all.
Rather, it assumes that a single-use seal primitive is available, and a random
beacon primitive for tx linearization, and then builds a system on top of those
primitives. Transaction data - the proofs that certain states have been reached
in the system - does not need to be broadcast publicly; if Alice wants to
convince Bob that she has given him money, the only person who needs that
transaction (and transactions prior to it in the tx history) is Bob.

So as to your question about miners assembling blocks, and what blocks contain:
there doesn't need to be blocks at all! Transaction history linearization is
something your wallet would do for you.

-- 
https://petertodd.org 'peter'[:-1]@petertodd.org