On Sunday, July 14, 2013 7:42:06 PM Pieter Wuille wrote:
> On Sun, Jul 14, 2013 at 07:33:06PM +0000, Luke-Jr wrote:
> > > The issue is that unless there is a cost to mining a *invalid* block
> > > the merge mined coin has little protection from miners who mine invalid
> > > blocks, either maliciously or through negligence. If the coin isn't
> > > worth much, either because it's market value is low or the worth is
> > > negative to the malicious miner, your theories of value have nothing
> > > to do with the issue.
> > 
> > Invalid blocks are rejected by validating clients in all circumstances.
> 
> I don't think that's what John means.
> 
> If you have hash power for the parent chain, mining invalid blocks for the
> merge-mined chain costs you nothing. Yes, they will be invalid, but you've
> lost nothing.

Nor gained anything. So the "lesser" chain maybe can't trust SPV.
But trusting SPV was already a bad idea anyway.

Note that the parent chain is not in any privileged position here either: a 
merged-mined chain could provide the value to the miner he is interested in, 
while he sees nothing of the parent chain. In short, merged mining is pretty 
much unavoidable in any case.

> The basic assumption underlying mining security is that it is more
> profitable to collaborate with mining a chain (and profit from the block
> payout) than to attack it. In the case of merged mining, this assumption
> is not valid.

The basic assumption of SPV is that more people will be assisting rather than 
making invalid blocks. That motive doesn't necessarily need to be economic, 
nor do proper validating clients rely on it. The only real assumption behind 
mining is that the majority will not be aiming to reverse transactions with 
valid blocks.

P.S. How about a Zerocoin with no-reward/PoSacrifice merged mining as well as 
(rewarded) Prime POW; maybe with no subsidy halving, to try a new economic 
idea as well ;)