On Sat, Sep 09, 2017 at 11:11:57PM +0200, Jorge Timón wrote:
> Tier Nolan, right, a new tx version would be required.
> 
> I have to look deeper into the CT as sf proposal.
> 
> What futures upgrades could this conflict with it's precisely the
> question here. So that vague statement without providing any example
> it's not very valuable.

So with Confidential Transactions, the only thing that's changed relative to a
normal Bitcoin transaction is that fact that the sum of input values is >= the
sum of output values is proven via a CT proof, rather than revealing the actual
sums. Other than that, CT transactions don't need to be any different from
regular transactions.

For CT to be a softfork, we have to ensure that each CT transaction's sum of
inputs and outputs is valid. An obvious way to do this is to have a pool of
"shielded" outputs, whose total value is the sum of all CT-protected outputs.
Outputs in this pool would appear to be anyone-can-spend outputs to pre-CT
nodes.

This gives us three main cases:

1) Spending unshielded outputs to CT-shielded outputs

Since the CT-shielded output's value is unknown, we can simply set their value
to zero. Secondly, we will add the newly CT-shielded value to the pool with an
additional output whose value is the sum of all newly created CT-shielded
outputs.


2) Spending CT-shielded outputs to unshielded outputs

Here one or more CT-shielded outputs will be spent. Since their value is zero,
we make up the difference by spending one or more outputs from the CT pool,
with the change - if any - assigned to a CT-pool output.


3) Spending CT-shielded outputs to CT-shielded outputs

Since both the inputs and outputs are zero-valued, to pre-CT nodes the
transaction is perfectly valid: the sum of coins spent is 0 BTC, and the sum of
coins created is also 0 BTC. We do have the problem of paying miners fees, but
that could be done with an additional CT output that the miner can spend, a
child-pays-for-parent transaction, or something else entirely that I haven't
thought of.


> Although TXO commitments are interesting, I don't think they make UTXO
> growth a "non-issue" and I also don't think they justify not doing
> this.
> 
> Yeah, the costs for spammers are very small and doesn't really improve
> things all that much, as acknowledged in the initial post.

Suppose zero-valued outputs are prohibited. In case #3 above, if there are more
outputs than inputs, we need to add an additional input from the CT-shielded
pool to make up the difference, and an additional change output back to the
CT-shielded pool.

If shielded-to-shielded transactions are common, these extra outputs could
consume a significant % of the total blockchain space - that's a significant
cost. Meanwhile the benefit is so small it's essentially theoretical: an
additional satoshi per output is an almost trivial cost to an attacker.

Quite simply, I just don't think the cost-benefit tradeoff of what you're
proposing makes sense.

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