Topic: Refocused Invictus P2P Exchange Discussion (Read 2591 times)

Sorry, I just noticed your post.

BitShares are NOT shares in a company backed by real people. 
BitShares grows the money supply through a combination of paying miners and stock-splits which appear like dividends.
BitShares pay real dividends via half of transaction fees.

BitAPPL is NOT issued by Apple Computer but is a speculative derivative based upon the relative value of BitShares and APPL stock.  Dividends paid on BitAPPL are not sourced from Apple Computer but from the BitShares held as collateral at the creation of the BitAPPL.   The dividend payments show up as BitShares and not USD or any other currency. 

I actually came up with a system for a decentralized exchange several months ago and have been looking for people interested in making one.
I have a lot of problems with this whitepaper, I think I got about about 2/3rds of the way through it before I realized it's not really decentralized at all. I mean it's no more decentralized than any other Forex, and simply making it peer-to-peer doesn't change the fact that fees, incentives and the behaviour of escrows are regulated by the developers rather than the government. It stands to reason that a free market solution would have the users of the exchange create their own fees, incentives etc.

Rather than detailing every problem I have with this I think it would be easier to explain the solution I came up with and hopefully the flaws will become more apparent.

One of the biggest flaws in the way exchanges are run today is that most assets are priced by currency. For example, if I want to sell my gold to buy palladium or company shares or a bond etc, I'd probably have to sell it for USD first and then by my desired asset. And the seller will be similarly restricted in that he can only sell his asset for USD or a similar currency, even if he wants to sell it for gold. A free market system allows a buyer and seller to transact any asset pair without going through some kind of "world reserve currency". This means anything from gold/USD to bandwith/driving lessons.

Secondly, there seems to be a lot of discussion on how to determine trust of a 3rd party in the absence of government enforcement, as if that were something we relied on in the first place. Now I didn't wait for the government to approve of MtGox before I made an account there, and I don't expect Paypal to tell me which ebay sellers are the 'bad eggs', I go to the crowd for that. Reviews, ratings, experience etc. That's something software does very well. As a buyer, the statistics I'm interested in from the seller are things like the number of successful transactions, the number of disputes, whether their products are exactly as described and delivered on time etc. That kind of information can be associated with the users digital identity, be it some kind of public key or a self signed certificate etc.

While there may be philosophical arguments as to what ownership is or isn't, it doesn't really matter until there is a dispute over a given asset. The best way to foster effective dispute resolution period, is to open it up to the market. Just as Bitcoin created a market for ASIC miners, which would otherwise be a fairly useless piece of hardware, a free market exchange would allow users to choose their own service. Escrows and dispute resolution organisations of all different shapes and sizes would spring up to fill various niches. For example, if I trade exotic financial derivatives with other savvy traders who realize we are playing a zero sum game on the market, it stands to reason that I would choose a savvy dispute resolution organisation with a proven track record in exotic financial derivatives. The seller and I would determine our own conditions and agree to be bound by the judgement of the DRO rather than relying on government regulation, which at the moment is as effective as a wet stick being beaten by the carrot it used to wield. DROs can offer varying degrees of anonymity and ways of ensuring transactions run smoothly. And DROs will fail or survive depending on whether the market uses them, just like any other business. If you don't trust the DRO's the seller uses then don't trade with him. Simple as that.

Now on to trading fees. There are many organisations which generate revenue by allowing traders to buy and sell assets which they hold. Take MtGox for example, they have the bank accounts and the software which allow us to trade and receive currencies and we are grateful for it. This is simply a different kind of market order and it rests on the distinction between the owner and the possessor of the asset. And every one of us should have the ability to do the same thing, assuming the market trusts us of course. Imagine you ran a business storing vintage wine and generated revenue from the trading fees of speculators, it's really no different from an exchange selling coffee, cacao, sugar etc. The only way we know those assets exist is because some trusted party says they do and probably has insurance policies to account for fires, natural disasters, human error etc. An example implementation of this would require the buyer to have a funded account with a 3rd party to pay the fees and approve the transaction.

Now asset distinctions, this is actually a tougher nut to crack in the sense that some assets are regular - one US Dollar behaves exactly like every other US Dollar - and some are irregular - one second hand Playstation may be wildly different from another second hand Playstation. That's something I don't really have a good solution for. However, we can categorize all assets into 4 main distinctions: Commodities, Currencies, Services and Contracts. When you do that, a lot of the confusion regarding various financial instruments and legal distinctions dissipate into the ether. For example, bonds, short selling, leverage, company shares, hedging, derivatives etc, are all contracts between 2 or more parties. A contract is essentially a set of if/else statements determined by the contract itself, statutes and legal tradition. A free market solution would not restrict people from using leverage or complicated contracts with more pages than the dictionary but simply give them the responsibility of clearly identifying liability and ensuring enforcement.

Now I could go on all day about the implications of this kind of system and what it would mean in the long term but these are the core features which would make it work. I completely endorse the idea of a standard protocol for communicating with the network but I'm pretty confused about some of the other ideas mentioned in the white paper.

Bitshare - polymorphic digital asset - mined like bitcoins, produces dividends.
My understanding of a share, is a contract identifying a percentage of ownership of the thing which it represents, created by the people who manage the thing which it represents and who also pay dividends from the revenue generated by it.
Mining on the other hand is a design feature which incentivizes miners to make bitcoins valuable in the absence of a bitcoin economy.
Now if you want to sell ownership of your business/project to raise capital for it, you are already incentivised to make that thing valuable and you don't need any other obstacle in your way to make your investment obvious.
I mean if the purpose of mining is to distribute the creation of something so that anyone can become an entrepreneur in that asset, but your system requires late entrants of the mining game to pay dividends to early adopters then there are no incentives for late adopters.
And why would you mine your own shares? You already have your company and can sell it as you please.
As much as I like the word polymorphic, I just don't know what this thing is or why I as a user of your system would be interested.
So is it decentralized or are you forced as a miner to pay dividends to other holders of bitshares?
I mean if I find gold in a field somewhere I don't immediately give 5% of it to other people who already have gold.

What are bitshares for?
Can you give me an example or thought experiment to explain the advantages of it, why would I use it?

I think we agree mostly. Isn't money a kind of contract as well? You trust the central bank / the government not to print too much of the currency. Bitcoin changes this structure. You don't have to trust the central bank, you trust the network to be stable/safe/whatever. Mike Hearn, in his talk at the Bitcon 2012, points out this was Satoshi's idea from the beginning. How could two parties interact, without direct mediation? So then the question is: how can two parties exchange property without intermediation? The two parties need the same "concept" of property. I think we can have types of interactions as well, through online plattforms. Imagine I can pull and fork a contract like a github-repo, then base software agents on an online contract.


You will have to define "decentralized market". Actually on current centralized markets you have quite a lot of intermediation, through brokers. If a broker / agent has a some freedom in negotiation we can call this discretion. Say like a trusted lawyer acting on one's behalf. The more you trust the agent the most freedom you will grant.

I think you should think a bit about how dispute resolution works in the absence of law.

The arbiter
a) earns money by behaving honestly
b) can cheat to earn a short term profit, but at the cost of sacrificing future revenue from a.
c) must assume the role voluntarily

If ther arbiter is unable to earn sufficient monopoly rents, then the equilibrium decision is always cheat.
Accordingly, you have to have price gouging for the market to work honestly.
In economics this is called incentive compatibility.
You can make him pay a good faith deposit,  sure, but this doesn't help you create an efficient market. Now you need to pay him even larger rents, or you will run up against (c).
In economics this is called individual rationality.

Implication: decentralized markets based on arbitration don't scale. Period. There is an inescapable  monopoly vs. Honesty trade off. Once you reach a certain volume it is always more efficient to use formal legal enforcement under a State.

Analogy: the blockchain is based on automated formal enforcement of rules, not arbitration. If it had used arbitration, it would have failed just like every type of cryptomoney that preceded it. Automated Formal enforcement is the correct technological direction to go in (think skynet). If you are going with arbitration, then you will not be competitive with existing platforms. I.e. what's the point of all this?

bytemaster. What is a fair price? Itsn't that for the market decide? I don't understand why you want to socialize profits. I'm a bit confused I must say. A market should be a place where everyone pretty much can do what they want, and the collective decides on certain boundary conditions. And the prime reason why people buy and sell on markets is to make money.

If you have only call auctions, say every hour, then the price will move only once an hour. In futures markets there are sometimes limit periods, which means the price gets locked. Say for instance it's 2008 and Lehman just went bankrupt. You hold 5 billion $ worth of stock for clients. You decide to sell 10% of all stock and go into cash. Well, now the market is limit down and you can't get out. Then in history often, and still sometimes in undeveloped markets, the stock market will close. This is not a position you want to be in. My boss was in such a position in 2008, and that is a time you depend on the market working. In other words, the efficiency of markets depends on quick execution. That in turn depends on computer trading. Of course one can always go back to pen & paper. Again, there is this propaganda that computer trading is bad. But that bitcoiners buy into it, I find quite amazing, because there you have actual currency which depends on computers, which is much more radical. So it just make no sense at all. Perhaps it would make sense to impose time limits, but they should be nanoseconds, not hours. But why not have an arms race for compute power to make the system more efficient? Sounds like bitcoin to me.

A good market will have call auctions, based on volatility interrupts. And also call auctions by beginning and end of trade. Again, this is how markets today work. You guys should study a bit more how stock markets work. Here is DB's market model, which is IMO the best in the world. http://xetra.com/xetra/dispatch/en/kir/navigation/xetra/300_trading_clearing/100_trading_platforms/100_xetra/200_market_model It's only 40 pages to read.

But if the market is a 'block chain' and the profits are paid as dividends, doesn't this make the 'currency' or 'shares' in the block chain more valuable while still satisfying all customer appraisals of fair price?  It isn't like the 'trading house' is owned by a central cabal, in reality it is owned by the very people holding the currency and thus providing demand.   Those that consume bandwidth and blockchain resources pay those that hold the shares.   More fees means less need for inflation to fund mining.     

So while it may seem 'unfair' for the trading house to profit, that conception doesn't apply to decentralized solutions. 

No no no!

What you are describing is what many of my players wanted trading-house software for, true. They wanted to buy the cheapest stuff and sell it to the highest-paying buyer, the good old buy low sell high.

Max profits for the "market" aka trading-house, the robber-barons offering the "service" to the lower classes.

The normal markets we see around as bitcoin exchanges and so on is the other way around. The highest price offered gets first dibs on the lowest price someone is willing to accept, and gets it at the price the seller is willing to accept. The gap or spread is between the highest offer and the lowest ask.

The posted version of batches seems to involve finding a price at which the maximum volume can be exchanged without anyone paying higher than they are willing to nor accepting a price lower than they are willing to but with them maybe paying less than the highest they were willing to and being paid more than the lowest they were willing to.

When I came up with batches I had not been thinking of that, merely of doing the normal clearing we are used to at current exchanges, but doing it in batches.

-MarkM-

I have actually been thinking about how to make these batches work.   Bottom line, everyone always gets the price they bid and any change is paid as dividends to the network (fees for running the market).

So if you place a Bid for $4.50 for 1 BTS then it will be executed at that price against someone who is asking less than $4.50... say $4.45... the person making the ask gets their price $4.45 and the delta .05 is paid as fees.  Neither party has any room to 'complain' and it no longer matters 'when' your order gets processed... you made a value estimate and were 'committed to it for 5 minutes.   Because finding blocks is 'random' and so is propagation delay on the network, there is really no way to 'observe' and then 'bid'.  If you observe 2 minutes of broadcasts.... place an ask for the 'highest bid'... you only have a 50% chance of getting included in the next block because miners only run the 'market making algorithm' once per minute and each hash has an equal probability of closing the block and thus your attempt to be 'last to bid' caused you to be 'first to bid' for the next round.

Bottom line, bid the price you think is fair, because that is the price you will get.   What is fair will change over time, but you are 'locked in' from the moment you click send unless you bid too low or asked too high.    Can someone cheat this?   I don't see how unless you take very broad definition of cheating.  But for me, it is good enough that everyone is always guaranteed to get what they asked for or get nothing at all.  There are no games where you bid  $1 and hope the market will clear at $.95. 

Take a look at the thread https://bitcointalksearch.org/topic/what-can-bitcoin-learn-from-high-frequency-trading-regarding-the-block-size-288298

In there it is argued that actually batches would be more efficient markets, that in effect high frequency trading is a bunch of parasites that are not actually useful either to actual investors nor to actual companies seeking investment.

Batches fit nicely with periodic publication of blocks, since each block could represent the clearing of one batch.

-MarkM-

Say you have an orderbook which updates every 1ms, with 100 orders/sec. Then you have a second orderbook with an update of 100ms. The second OB tracks the first. Anyone trading in the second OB will basically lose money if he is a market taker (i.e. crosses the spread). An example: a couple years ago a bank made a market of options on a future. They were not very sophisticated and quoted a price with a 15 second delay, while the underlying had a new quote every second. A trader noticed this and scalped the bank for a couple of hundred thousand $. Then they saw somebody made a lot of profit against them, and they stopped trading with him. This kind of thing goes on in financial markets all the time on different time scales. For a liquidity taker (the buy side) trading in a such a swap market is a losing proposition. It mostly leads to transfer of wealth to the liquidity provider. If the taker only transacts once a month, he doesn't care he had 1% of fees in terms of a spread. If he transacts 100 / day, he will accumulate losses quickly and go out of business. You need market makers for a market to function, to minimize the costs. And they have different incentives. In many markets market makers get money from the exchange to provide quotes.

A very interesting use case for me would be exotic derivatives, which can be transacted with between countries, i.e. a market where transactions are very difficult and don't very often, or transactions in countries which don't even have stock markets. Say establishing a kind of a stock market in rural village in Africa. I believe something like this will happen. I certainly do hope we will have an alternative system in place in case the current one breaks down.

Could you please substantiate 'not very well in the short term'.   

Dividends earned here are not from the underlying asset, but from the exchange itself.   So dividends paid have nothing to do with the dividends paid by a real company on the actual stock. 

[minor edit]

I find it certainly valuable work in this direction. But you're talking about Swaps or a betting market, not an asset exchange. Somebody with a background in finance will not understand this.

Quite generally you can't base an economy on swaps. Say a company wants to raise money, i.e. sell stock on the market. That's what stock markets are there for in the first place. With swaps you don't real price discovery, but the price tracks the real market, but not very well in the short term. You also don't earn any dividends.

In my opinion, which differs quite a bit from the proposal, there might be some breakthroughs possible by having new types of contracts and negotiations. How this might work I don't know, but the problems are all matter of the law. In terms of actually building an alternative financial system, I think that's quite far away. You really want an exchange to be efficient, which is not possible with a blockchain. IMO you need other types of intermediation, which allow millions of tx's / hour or second or microsecond. Or you could have new types of auction mechanisms.

Where we agree is that looking at cross-exchange issues is important. For example, say it would be almost free to operate an exchange, and all exchanges are interconnected through APIs (and market models). Then the system will be populated by arbitrageurs. This is how the FX market works today. The futures and stock market has a very different structure. FX is OTC but still very performant. Nobody owns the FX market - it is a interbank network.

You do not have transferrable ownership of the underlying good, you have transferrable ownership on a bet backed by a crytpo-currency about which way the value of the asset will move relative to the crypto-currency.  This bet is entirely within the realm of the crypto-currency and all profits and losses occur in the crypto-currency.

I don't understand how you could have transferable owernship in such a network, because owernship is a legal term. If I hold one stock of AAPL corp this is a legal concept subject to US law. If I hold a stock of BMW in Germany, this is subject to German law. Similarly ETFs and Futures have a lot necessarily regulation. We don't have a legal layer for this (yet), so I can't even transfer the most simple asset from one country to another without a lot of hassle. So whatever these kinds of concepts are, they have little to do with shares as they currently exist. Markets and property without laws are not very useful. Bitcoin and Cryptocurrencies are really the exception, because they are an entire new construction, but they don't standard for anything else. Money is a medium and assets are not money. In order for something like this to work I think one first has to have the legal constructions in place, which I believe will take decades. Hopefully the economic system will function much better by then.

BitShares is a concept, a generalization of an idea.   There will be many 'BitShares' chains and unlike Bitcoin, we will actively seek and promote new 'alt-chains' each with their own derivatives. 

To support user generated assets, dynamically (originally the goal), I will have to change the features that exist on the main chain... which we are tentatively dubbing 'CashShares'.    Other chains will include 'CryptoShares' for most alt-coins.

We have started the design for general purpose prediction markets but will probably have the following differences:

1) long positions will not pay dividends, though short positions would still incur opportunity cost as their dividends are forfeited.
2) Require people to update their positions faster (say 3 months rather than 1 year).

There are many other details, but I am still working on them.

Will users have a way to create new asset types like Loom or Ripple? This could be a lot of fun as a derivatives market, seeing as even BitUSD and BitGold will be derivatives.

BitShares is a distributed autonomous corporation (DAC) where the 'shares' of the corporation are mined into existence and pay dividends based upon transaction fees.  These shares can be held as collateral for USD loans (BitUSD) from the DAC and other people can buy these collateralized debts of the DAC and earn 'interest' paid from the dividends due the collateral. 

Therefore, BitUSD acts as a way for two parties to speculate on the direction that the value of BitShares (aka the stock of the DAC) will move relative to USD.   If the price of BitShares relative to USD goes up, the short position makes money, if the price goes down the short loses money.   Either way the holder of USD maintains the purchasing power of 1 USD worth of BitShares plus interest.

We are releasing a product code named Project Quixote that provides a decentralized 'id' system that maps a human readable account to a public key and will be a complete alternative to Open-ID and entirely eliminate the need to ever share a 'bitcoin address' again.   These IDs are also used as part of a communication system similar to BitMessage (without having to share 'bitmessage addresses'). 

The foundation for the messages and IDs will be released first and will help us harden our network and P2P infrastructure while we implement the blockchain logic separately.    THe gensis block for the ID system will be launched in October, the BitShares test network will launch in November, and the live network will be in feburary after we are sure that there are no glaring bugs / redesign necessary.

Project Quixote will also double as a Bitcoin wallet that eliminates the need for Bitcoin addresses.

So, BitShares is a sort of alt-currency / alt-chain? Is it mined like Bitcoin, or differently?

Can you explain how a BitUDS is represented on the BitShares protocol, and how is its value determined? In what way does Bitshares act as a prediction market?

Also, I don't understand the roadmap. If the genesis blocks launches in 2014, what are the components that are launched before that? What does it mean to launch them without a genesis block?

BitShares is a effectively a set of continuous prediction markets on the value of BitShares relative to USD, BTC, EUR, etc.   Prediction markets are very hard to manipulate and are very good at integrating information quickly as a result the prediction market will track the value of USD, BTC, etc in a decentralized block-chain based manner. 

BitShares is not pre-mined, so you can invest early by adopting it on launch and mining on day one.  Also, promoting the idea and helping us develop it will insure you the best return on investment.  Thus, early adopters will benefit like Bitcoin early adopters. 

We have plans to reward early adopters that provide funds prior to release, but we are not yet prepared to announce our plans.

Mastercoin is horribly flawed and carrying around the ball and chain of Bitcoin's block chain bloat and all of the risks / problems associated with bitcoin.   It has scalability issues and its Escrow system for creating the equivalent of BitUSD is broken (I won a 3BTC bounty for showing how and why, he has not yet published a solution that I am aware of).   

Development roadmap is to release a BitMessage / BitID system in October, a beta of BitShares by Thanskgiving, and launch the genesis block in early 2014. 

I started reading the whitepaper, but I find it rather hard to read and don't have the time to read it more thoroughly at the moment.

1. Is there a short summary available that explains how BitShares implements an IFMFS?
2. Is there an incentive for people to "invest" in BitShares? Is it more like Bitcoin, Mastercoin and Ripple, in rewarding early adopters (assuming the technology captures significant market share), or is it more like Colored Coins, which is technologically sound, but offers no such incentives to early adopters?
3. Can you compare and contrast it with Mastercoin?
4. Is there a development roadmap? What's the current stage of development?

I made a few minor comments on the doc.