Everybody hits this conceptual problem eventually: new, dilutive issues.
I don't think it's a conceptual problem. Dilution of an existing color is more like a convenience function, it isn't fundamental.
Association between color and issuing address is also merely a convenience. It isn't fundamental. On blockchain level, we trace transaction outputs. Addresses are nearly irrelevant.
So it's more like my question was: do you REALLY need that convenience? Or maybe we can simply use most simple and restrictive approach?
"One color = one transaction output" kinda makes sense.
The thing is, they are unavoidable. It is a variant of the Sybil attack. An issuer may create new shares at any time, simply by representing themselves under a new identity. An issuer may also issue more shares/bonds than there are assets to back their bond (i.e. a mining company selling 1TH worth of bonds, when they only have 500GH).
These are different things. On stock/bond markets shares represent ownership of securities. I.e. contact which is associated with color will say that shareholders which can be identified through blockchain tracing have certain rights. Once you own shares you can demand contract to be enforced via legal mechanism, at least in theory.
I.e. bond holders can sue issuer if he didn't back his bonds with assets as promised, and perhaps he will be personally liable if contract says that.
Electronic signatures are recognized by law enforcement, and I'm sure there is a way to tie identity to contract. So as long as holder identification mechanism works correctly, colored coin mechanism is about as good as having a signed paper.
However if you allow dilution of existing colors, it might break identification mechanism. E.g. contract says there are 1000 bonds, but 1,000,000 are issued. Obviously this is wrong, but how do we find which of these 1,000,000 bonds are valid? If there is no good way to do it, I'd say it is a failure of colored coins as of an identification mechanism.
Don't forget that issuer can always claim that his private key was stolen and that he's not responsible for over-issuance. Again, I don't want theft, or theft-used-as-an-excuse, to be able to break holder identification via colored coins.
If we use the most restricted model, it provides a certain guarantee: if you currently hold 1 bond out of 1000 issued, you can always prove that you own 1 bond of 1000 issued, no matter what.
Without rating and third party auditing and real-world identity checking (of the issuer), voting to authorize a new issue might simply provide an illusion of safety behind which dishonest folks hide.
Rating, third party audit and real-world identity check are orthogonal to this: they are necessary to for safety of investment, but even if issuer is honest there is always a room for software failure or theft which can wrack a havoc through dilution.
The question to ask oneself is: what would a rational (if amoral) economic actor do? What are the economic incentives that drive the issuer, and the holders? And what can software do to change or enhance those incentives?
Satoshi spent a long time designing the economic incentives surrounding bitcoin. For example, he worked hard to make block header hashing a static affair, not directly related to transaction or block byte count. Increasing the "work" as block size increased would encourage miners to make their blocks as small as possible, he realized.
Satoshi also made it so that only 21 million of coins will be issued, and it isn't a matter of incentive but a matter of consensus.
Even if all miners will refuse to cut block awards client software simply won't accept their blocks. So expansion of Bitcoin emission is only possible with approval of considerable number of
users, not
miners. Does it remind you anything?
I believe that implicit-consensus-based approach is inherently bitcoinesque, for better or worse.
