non-repudiation works just as well with signatures
How would you prevent double-spending?[1] http://blog.koehntopp.de/uploads/chaum_fiat_naor_ecash.pdf
Yeah, a central ledger works great if you want none of the benefits that Bitcoin offers.
>Essentially, anything which is inherently centralized has no business being on a blockchain. This is really anything backed by a physical asset
Bitcoin has proven that decentralized assets have huge advantage. Your starting assumption is wrong; wealth is not inherently centralized, and assets are not inherently centralized.
In this case, there is no point in representing the asset in a blockchain. The blockchain is probably the least inefficient datastructure we have seen in modern times. A centralised asset is much more efficiently represented by a centralised ledger. You can still use triple-ledger accounting.
That's true. But gold is a physical asset; shares in a company are not. There is no technical challenge to distributing abstract assets (like shares) over a decentralized system, and there are many advantages to doing so.
This sounds a bit like saying that Bitcoin is inherently centralized because someone can physically hold a gun to your head and compel you to turn over all your coins. I think there are still benefits to a blockchain approach, even if the represented asset is "centralized."
With a signed token (optionally blinded), you get non-repudiability, but if the central exchange is shut down, you don't have safe transfers (it becomes impossible to prevent double spending, so you probably do a single payment to a trustworthy entity and get new tokens from elsewhere). Does nothing for redemption risk.
With blockchains, you have non-repudiability (although less strong than the signed tokens, due to 51% double spend attack being easier than defeating RSA 4096 or whatever), and continue with transferability even in the face of issuer failure, but still have redemption risk (if it's a "backed" currency).
Arguably sidechains give you non-repudiability, eternal transferability, and no redemption risk (at least back to bitcoin). However, it only makes sense for things which are anchored in the root blockchain (i.e. ultimately bitcoin derivatives of some kind); it doesn't make sense for gold bars.
IBM shares are essentially centrally controlled by IBM.
Gold (or other physical commodity) backed shares can be centrally issued by each holder of gold; essentially, each bar or vault would have its own issue. This is just negotiable warehouse receipts.
The only classes where decentralized issues make sense are "virtual" currencies like bitcoin, or equities/debts of virtual corporations. There probably isn't a market for more than one virtual currency in the long run, except maybe for sidechains (or exchangeable alternatives) due to purely technical things, or compliance, like zerocash vs. bitcoin.
Why does that mean we have to trade IBM shares with a centralized system? There are huge advantages to having decentralized trading systems.
I still don't think the benefits of this make up for the costs of the blockchain for most uses.
So they can send the dividends to the addresses that own shares. In fact, that should be even cheaper than using a bank or any other middleman-requiring system. Another advantage of decentralization.
I am a bit confused. My impression so far was that something like this would be commonly seen as a decentralized scenario?
I would consider a promise from pg or sama to pay 1oz of gold upon presentation to be pretty low on counterparty risk; I'd consider the same from a random person online who claimed to have gold and be willing to redeem it upon presentation of one of his tokens as a much higher counterparty risk. The market would price these separate issues, and then you could build a basket of gold issues into something meta (maybe with a blockchain, or maybe with smart contracts, to reduce the meta counterparty risky).
Blockchains have a few serious disadvantages over other protocols (chaumian blinding, or just signed receipts, or a central ledger) -- inefficiency, lack of privacy (zerocoin addresses this, but at the cost of incredibly advanced mathematics), and the 51% attack.
Essentially, anything which is inherently centralized has no business being on a blockchain. This is really anything backed by a physical asset -- the ideas to do colored coins which were gold warehouse receipts were insane/inane. Most of the sidechain proposals I've seen are equally absurd -- the idea of equities issued in a blockchain where the equities are essentially tied to a single real-world issuer is an unnecessary step. You could just have the entity issue digital shares using a centralized protocol; non-repudiation works just as well with signatures. (you might use a distributed system for a virtual corporation of some kind, but that's an exception).
Bitcoin itself makes sense (it's a decentralized store of value), and some of the smart contracts stuff might make sense (but the smart contracts cease being self-enforcing, and thus smart, as soon as a lawyer or government gets involved...), but most of the sidechain proposals involve trying to recreate assets from off the blockchain in a less efficient, less secure way, just to prop up the value of bitcoin itself.
There's no reason to use a smart contract to "prevent issue of too many shares" -- you can just have the market price shares based on what the market believes is the outstanding treasury and float. People can build systems to provide better controls there, which might in some cases involve the blockchain, but it's not essential -- we can price assets using a market with imperfect information already.
A system with ~infinity of distinct issues, traded in realtime on market(s), with perhaps something like bitcoin as a unit of account, makes a lot more sense to me than trying to have one global currency (bitcoin) and then doing things like sidechains (which don't let you escape the limitations of the blockchain protocol).