Topic: Bit Share 2.0 - Proposal for Decentralized Bank and Exchange (new and improved) (Read 1722 times)

Yes that was version 1.0 until I realized that I was price-fixing both term and collateral requirements and my knowledge of economics tells me that any and all price fixing is BAD VOODO.  Thus I was forced to innovate a way to eliminate price fixing.


Please move all discussion on this topic to this thread: https://bitcointalksearch.org/topic/old-bitshare-economic-theory-10-btc-bounty-to-prove-me-wrong-paid-215488  where I am providing many more answers / details including the answer to this question.

So I figured that a new coin-type would be allowed once a certain BitShare value in open bids for that coin were published on the network.  These bids would be backed by BitShare (until they were canceled).  With a high enough value threshold there should be enough 'volume' to get a market started and thus the network would allow the creation of the new coin-type in response to the highest available bid and work its way down.  Publishing all of these 'bids' and keeping them open would require transaction fees, time, and locking up of real value and therefore could not be spammed.

Before anyone would 'bid' on a new currency type they would have to know what the market 'consensus' meaning of that peg was.  They would also have to find someone with Paper-USD that would be willing to trade for crypto-USD at or near the 'peg' or else there would be no profit in creating crypto-USD over simply holding BitShares.  



You cannot redeem his specific collateral, but you could convert into BitShares at the  new exchange rate.  After all, the guy you meet via local-share-coins may only have a positive crypto-USD balance from trading with others.      


You got it.  The price variation in crypto-USD should ultimately be less volatile than the price variation in Mt.Gox USD... see other thread for more details... but needless to say a dollar in your hand is worth less than a dollar on Mt. Gox when you want to BUY NOW and a dollar in Mt. Gox is worth less than a dollar in your hand when you 'want out now' and the government makes it hard!   Bitinstant's whole business model is to capitalize on this price difference between USD credits with different issuers.   Bitshares would ultimately decentralize the function performed by Bitinstant as everyone could play a role and thus drive down Bitinstant's margin / profit due to competition.  

All crypto-USD issuance is tied to an exchange on the market.  Thus it is not possible to issue at 'arbitrary rates' for which there is no depth in the market to support.  (more specifically, the issuance would directly fulfill a bid and adjust the price for the next issuance)

Once again, please move follow-on questions to the help wanted thread where people are trying to collect my bounty to prove my idea will not work and thus convince me to not invest $20,000 in making this happen.

Thanks for your answers!  With regards to the amounts, I thought to have read you need 2x the amount as collateral somewhere.  But maybe that was your earlier "1.0" proposal, I don't remember.  Doesn't matter anyway.

Ok, so the total amount of BitShares will be constant (like with Bitcoins), but the amount of crypto-USD issued/redeemed depends on the market rate.  Again, I have more questions:

1) What about bootstrapping the system?  If there are not yet any crypto-USDs in existence, no one can actually issue bids on the blockchain (or at least not expect them to be filled when trying to sell BitShares).  How do you get an exchange rate for creating the first crypto-USDs?  And in general, I get a somehow uncomfortable feeling thinking that the network protocol depends so much on the market; but I guess there's no way around that if we want to peg crypto-USD to real USDs, and I'm also not sure about concrete attack vectors that could pose for manipulation.

2) Let me think out alound.  I'm trying to understand how "real" USDs would come into the market, when people want to mainly buy in to BitShares.  Say, I have some real USD and want to buy my first BitShare.  Then I would have to look up someone (like on localbitcoins.com or bitcoin.de) converting my cash into crypto-USDs first.  So maybe someone who already has BitShares is willing to do that, and thus collateralises some of his BitShares for crypto-USD, which he then sells to me for real USD.  I could then use those to buy BitShares either on the market or maybe even redeem his collateral.  What's the real advantage of this as opposed to directly finding someone willing to sell BitShares for cash directly?  Ok, well, I can see that it it probably advantageous because crypto-USD will hopefully have a more stable price in USD than BitShares and so the meetup and associated delays are not to risky.  And when a pool of crypto-USD is present already, one could use that to speculate directly on the market (as on Mt. Gox).  Is that what you also think about?

3) Thinking about it, it seems as if creating crypto-USDs against a collateral would let you effectively sell BitShares "above market":  If you sell a large chunk of BitShares ordinarily, it would fill up bid orders one after the other lowering the price on the way.  However, if you simply create as much crypto-USD as you like, you will always get the currently "best" bid-price - even if you "sell" 1000 BitShares and the highest ordinary bid is just for one.  Thus, no one would actually fulfill bid-orders, would they?  But if there are bid-orders anyway, one could simply leave out the crypto-USD creation stuff and instead of creating them by the network in return for a collateral simply force anyone to just sell his/her BitShares.  Wouldn't that be much simpler?  Getting back to 1), the only problem would then be how to actually create crypto-USD when no ones / no bid orders are there at all.

Correct, the system could work with any hashing algorithm.  My reason for suggesting the change at this point is because upon creating a new blockchain it is the only opportunity to make major changes to the rules.

So here is my thinking:  RAM is already a dense, specialized, ASIC and it is just as expensive to put RAM on a ASIC chip (transistor count wise) as it is to put RAM in your computer.  RAM is already widely distributed and dual use.  Thus security provided by using a lot of RAM is actually superior to security based solely on computational power.  The RAM actually 'limits' the density of calculation power and therefore keeps the network far more decentralized than what we are already seeing with bitcoin.   Any government could buy up specialized ASICs and cause Bitcoin a lot of trouble.  Any government could *outlaw* bitcoin asic if they were desperate.  Such actions would not be possible with RAM.   Sure the government could start buying up all of the RAM and building massive datacenters... but the size of the RAM market is so much larger than the mining market that it would be much harder to control. 

Now clearly, those who have invested in SHA256() mining power would not be too happy to see an Alt chain take off that is incompatible with their investment.   But their investment is just capital that could be sold to other bitcoin users if they opt to jump ship. 

The exchange rate comes from bids placed in the blockchain where highest-bid must be filled first.  Before anyone would be willing to 'bid' they would have to have a market-consensus on what P or Q is attempting to peg to.   Thus if someone knows that every rational actor in the network is attempting to peg P to the value of a dollar they would bid on P proportional to the dollar value and sell P when the price was too high.   It is ultimately about 'consensus' and common understanding about the *intention* of a particular peg.  Anyone individual would not be able to change the 'obvious intention' by bidding more or asking less because others who shared the common intention would see it as a buying or selling opportunity.   

This is one place where I intuitively see how it works but am still working on how to explain it.  As an INTJ this is always my biggest challenge.

You are right, so using proof-of-burn would be an 'upgrade path' approach.   However the resistance would only be from those who don't see the potential for higher returns from BitShares.  Any bitcoin holder that realizes the extra value in BitShares first would have first-mover advantage and those left holding BitCoins may be upset, but because they haven't taken the time to see the value advantage they will have done nothing to support the new value / growth of BitShares.  The existing holders of bitcoins would have to judge that Btc is worth more than Bitshares because existing holders of Btc would be the ones buying them off of you.   In fact, the early adopters of BitShare who think the of BitShare's are too high relative to bitcoin would be the ones selling.   Therefore, the market dynamics already ensure that everyone is rewarded according to their own personal risk and value assessment.

With a proof-of-burn approach bitcoin holders could 'sit on the sidelines and watch' without putting any skin in the game.  Meanwhile those who did put skin in the BitShare game (by burning first) would see no extra gain compared to bitcoin holders.   Furthermore, it would be a 'one-way' conversion wich would be even more risky as the 'trade' couldn't be undone later.   You need the market to allow two-way conversion at the proper exchange rate between BTC and BitShares.

The Network / Blockchain holds the collateral by converting a Dividend-Paying BitShare into Dividend-Paying crypto-Fiat at the current exchange rate.   After the 'conversion' step your wallet would contain:

1 BitShare (held as collateral for $125 crypto-USD that no longer pays dividends)
125 Crypto-USD   - which is paying the cyrpto-USD dividend rate (in BitShares).

If you were the one and only conversion from BitShare to CryptoUSD then the total dividend rate you receive on your 125 Crypto-USD would equal the dividends paid on 1 BitShare.  However, this is a big market with lots of competion so there are other people who converted 1 BitShare to 120 Crypto-USD, 121 USD and 130 USD.   Thus the dividend rate (paid in BitShares) for people who hold crypto-USD would be

(120+121+130+125) USD pays dividends Equal to 4 BitShares.  Thus the effective dividend ratio is $124 pays equal to 1 BitShare.  As you can see the exchange rate between BitShares and issuing of Crypto-USD causes many small conversions at different prices over time to accumulate into an average Crypto-USD vs BitShare 'interest rate'... when you normalize 'value for value' there will be slight variations on the %Return yield for USD and the % Return yield for BitShares, but these variations cause profit opportunities for people to buy Crypto-USD and pay-off their colateral any time BitShare has a higher Dividend rate as % of value.  Any time Crypto-USD has a higher Dividend rate as % of Bitshare then people will be issuing new Crypto-USD at the current exchange rate and then sell it bring the Dividend rate on all currencies back to "Equal" as a percent of value.  

So, when the market adjusts you (and everyone else with BitShares locked up as collateral) have incentive to buy-back crypto-USD so you can free your collateral and capitalize on the higher interest rate paid on BitShares relative to crypto-USD.  



The block chain has support to make bids that other people can accept provided the transaction sends outputs to the proper addresses with the proper exchange rate.  The blockchain rules would require that the highest bid be used for all conversions AND that collateral could only be issued in response to a bid.  Therefore, you can only create crypto-USD in the process of actively responding to the highest bid for crypto-USD on the block-chain exchange.  Thus the creation of crypto-USD is competing against those who are simply selling existing crypto-USD and it is immediately affecting the price.

If the highest bid is 1 BitShare for 1 crypto-USD then you could issue 1 crypto-USD for 1 bitshare.  No need to have any additional collateral... (that would be price fixing) and the system is based upon using your collateral to back the crypto-USD but instead using the Dividends that your collateral pays to back the crypto-USD and therefore it is impossible to have insufficient backing regardless of exchange rate fluctuation.



They couldn't do this, they have to create new crypto-USD in response the current highest bid, so they would use 1 BitShare to issue 2 crypto-USD... (upon closer examination your ratios are correct, but you still have 2x the collateral when it should be 1 :1 at the current exchange rate)

If you have 1 crypto-USD issued that is locking up 1 of your BitShares then it would take 1 crypto-USD to free your BitShare.   If someone else has 4 crypto-USD locking up their 1 BitShare then they would need 4 bit shares to unlock it.    This is critical because it is what drives market forces to redeem Shares issued at over-priced rates.   In your example the dividends paid on 1 USD fell over time from being equal to 1 Bit Share to being almost 1/4 what is paid on 1 BitShare.  Thus you make money by buying 1 crypto-USD for .25 BitShares and redeeming your 1 crypto-USD and thus quadruple your dividend payout.  When you do this you would simultaneously be causing the yield on crypto-USD to fall causing the divided ratio paid to crytpo-USD to more accurately reflect the change in market price.


I would mark each collateral output in the block chain with the rate at which it was issued.  This is done because I need market forces to 'redeem' the crypto-USD that is issued at higher than current market price and to re-issue crypto-USD at the new market price.  

That said I need to validate the potential of allowing ANYONE to reverse a crypto-USD issuance at the current exchange rate....  assuming the economic incentives work out (and I think they do) this would create a more fluid market and allow more actors to participate in correcting price variations.  

So (thinking out loud here)... if ANYONE could to a reverse conversion at the current exchange rate then it would 'destroy' an equal amount of crypto-USD  AND free up a matching amount of BitShare at the same ratio... therefore the network would still know  TOTAL USD ISSUED and TOTAL BitShares backing it and thus could calculate the dividend payout rate to USD holders and come to the same number.   Therefore I believe that just like anyone can issue crypto-USD at the current exchange rate, anyone can also redeem crypto-USD for BitShares at the current 'ask' price and thus there is no need to tie a specific colateralization to a specific redemption.  THis is a beautiful and powerful insight that will greatly enhance the efficiency of the market.


Thanks for you interest and feedback, keep the questions coming!

That would depress the value of BTC, which could lead to a lot of resistance, perhaps in the form of a fork that did use proof of burn.

This would be the 'upgrade path', but in theory it would be just as easy to sell your BTC to buy BitShares and you would be converting value-for-value.   I have considered this, but it would also require having the client track 2 block chains for some time.

I presume that has to do with "how many crypto-XXX you get for a certain collateral", namely the exchange rate used for those "loan" transactions.  That's also why am in particular interested in how exactly those would work, and where the exchange rate the network charges comes from.

The change to the hashing algorithm is orthogonal to the banking and exchange functionality, isn't it?

I'm having a hard time seeing why the crypto-USD would track the USD while crypto-EUR would track the euro. From the perspective of the block chain, what difference is there? We could just as well call them P and Q, and what remaining link with fiat would there then be?

I'm delighted to see so many ideas about decentralised exchanges pop up.  And I believe that your idea of holding BitShares collateral for virtual USD (or EUR or whatever) coins makes sense.  However, I do not yet really understand how such a crypto-USD creation would work in detail.  In particular, who holds the collateral and creates the crypto-USD in exchange?

It must be the network, because if that would already be a kind of "currency exchange" between users, there would be no way to initially get crypto-USD created.  But if that is done by the network, how exactly do you decide on the exchange rate?

Consider for instance this scenario and please tell me where I did misunderstand your idea:

1) Assume current exchange rate (however that is determined) would be 1 $/BitShare.  So I could get 1 crypto-USD by sending 2 BitShares of mine to the network to hold as collateral.

2) Now the exchange rate changes to be, say, 2 $/BitShare.  Another user also creates 1 crypto-USD by sending now only 1 BitShare as collateral.

3) Again the exchange rate changes, now we have 4 $/BitShare.  We both want to redeem our crypto-USDs.  Would I now get back "my" 2 BitShares just for one crypto-USD while the other user would have to pay 1 crypto-USD for one BitShare, and the official exchange rate is even 4 $/BitShare?  Or would the network return as many BitShares in return for redeeming the crypto-USDs as the exchange rate is?

In the first case, you would have to mark each crypto-USD according to the collateral, so that each crypto-USD would be able to buy back a different amount of BitShares -- which does not make any sense, because they should all be alike.  In the second case, you would end up with a totally unpredicatable actual supply of BitShares not equal to the original 21 million.

So ... how exactly is that loaning and redeeming of BitShares and crypto-USDs meant to work?

I'd be in favour of a mechanism that uses BTC as BitShare, or if that fails to get consensus, to use proof-of-burn to let existing holders of BTC transition 1:1 to the new currency.

CryptoUSD are as fungible as Bitcoin and as divisible as bitcoin.  They are simply another output in the block chain that is annotated.

There is no requirement.  By default 10 minutes works but if there is a compelling reason to reduce it after learning from Bitcoin I would be open to it.


The Botnet issue is a concern but equally of concern is that bitcoin ASIC would take it over and offer an easy way for a government to drop a ton of money on ASIC.   I honestly prefer Botnets to centralized / specialized miners.  The solution is to harden networks against Botnets not allow BitShare to be 'centralized' by specialized mining companies.

I had considered just such an option.  If it were to be implemented it would have to be a limited time and would require all clients to subscribe to the bitcoin network to 'verify' the coin generation.   It would be just as easy for someone who owns bitcoins to sell them and buy BitShares.   They could even use the built-in BitShare exchange to do it.

First thing to realize is that *assuming* BitShares gain value of their own like LiteCoin did, then interest payments paid to BitShare holders will have value.   Someone who posts BitShare collateral would LOSE the interest payments to the holder of the crypto-USD.    As the exchange rate went up between BitShares and crypto-USD then they would want to redeem their collateral at the older exchange rate and therefore would want to buy back crypto-USD.  All crypto-USD is fungible.  The cheapest way to buy crypto-USD is with paper-USD assuming there is more demand to withdraw than to deposit.   The end result is that these factors conspire to ensure that crypto-USD is redeemable at near face value.

The interest payments are a critical part of making sure crypto-USD is redeemable because without them there is no difference between selling Bitshares for paper-USD or using Bitshares as colateral because the holder of the bitshare isn't giving anything up.   Furthermore, the interest payments on crypto-USD is what brings paper-USD into the market and thereby allows people to redeem crypto-USD.   If you were to remove the interest payments the whole system would fall apart.

Perhaps I shouldn't call them interest payments.  In reality they are dividends paid to the share-holders of the distributed bank.  The distributed bank earns money on transaction fees and the cost of those transactions are born by all of the users of the system (not just the miners).  The dividends are what motivate people to LEND VALUE to the network because they get paid to do so even if the Bitshares themselves do not appreciate in value.   They can lend value to the system without taking currency exchange risk. 

I believe in this idea enough that I am going to put $20,000 on the table and hire some consultants to help me build it out.   So I would like to know the reasons for your concerns about dividend payments.

Are cryptoUSD fungible or are they somehow tied to their BitShareCoin collateral?
Why mining? Also: Why mine with some algorithm that might be "safe" from GPUs (maybe) FPGAs/ASICs now but not in the future?
How fast does this chain need to be (block target 10 minutes like Bitcoin)?

I guess mining is used to distribute BitShares as evenly and randomly as possible, right? This will only give it away to Botnet owners in case this gets any traction!
You could for example make them redeemable for Bitcoin - freeze a coin provably on the blockchain in some way and you can redeem them as BitShares using the same private key. The other way round might be harder/impossible, as it might require changes in Bitcoin to "unfreeze" the coins (in case you don't send them to a 1BitcoinEater... address but freeze them with an elaborate transaction).
Handing them to Bitcoin miners instead of freezing/destroying them might give incentives for miners to get BitShares by using their own coin as fees and getting back both the Bitcoin AND the BitShares. On the other hand, maybe you like that idea.

I still don't like the idea of interest payments and USD (or BitShares that can be exchanged for cryptoUSD that can be exchanged for USD) being created out of nothing and then calling this risk free and decentralized.
Who should ever let me redeem my cryptoUSD as a bank transaction/cash and who initiates that transaction first?

I have been posting a lot about my ideas on creating a distributed bank/exchange and for the most part have failed to gain much discussion or enthusiasm.  If you read my prior ideas, please do not skip this post as it is a VAST improvement and brings everything together in a far simpler manner.

First I want to introduce the basic crypto-currency rules for Bit Shares:
1)  BitShares follow BTC distribution model (50 BTC, 25 BTC, ... etc) until 21Million are in existence.
2)  BitShares pay interest with 50% of the mining reward + fees, the other 50% goes to the miner.  Justification: The average interest rate in a free market equals average business profitability and mining is a business.  This logically deduced from economics and I can defend it if you want.
3)  New crypto-Fiat can be created by holding BitShares as collateral at the CURRENT built-in exchange rate.   Later crypto-Fiat can be destroyed by redeeming the collateral.  While BitShares are held as collateral their interest payments are paid to holders of crypto-Fiat.
4) The hash algorithm is designed to be more secure and memory intensive than LiteCoin (say 128MB, random-access, sha256 based)
5) To modify this from bitcoin would only require annotating every output with a currency unit + introducing a 'bid' transaction output to facilitate an exchange between crypto-currencies + finding a way to distribute interest efficiently (I have it mostly figured with the exception of blockchain splits/merges).

That is it, 100% of the rules for the new currency and otherwise identical to Bitcoin.

So how does this create a decentralized Bank?

The goal of the economics in this system is to force market competition to drive the spread between crypto-USD and paper-USD toward 0.  Note that it will never be 0 because the spread between paper-USD and Mt. Gox USD, and USD in your local bank or USD in a foreign bank is already non-0 due to wire-transfer costs, deposit fees, atm fees, time delays, and gas costs (assuming you go to a local branch in person).  So the true goal is not absolute parity but parity that is at or below the cost of depositing or withdrawing USD from the existing banking system.

When there are more depositors than withdrawers from the distributed bank then the price of crypto-USD will be more than face value.  Holders of BitShares could either sell or borrow against their BitShares to get crypto-USD and then sell crypto-USD for paper USD at above face-value for a profit.  If they borrow against their BitShares to create new crypto-USD it comes at the opportunity cost of interest payments they would have received on their BitShares which are now directed to crypto-USD holders.   

Later when market forces change and the price of crypto-USD falls below face value then it is profitable for everyone who has borrowed against their BitShares to buy crypto-USD and redeem their BitShares so they can start collecting more interest and make a profit from the transaction because they bought crypto-USD below face value. 

The result is that the spread between crypto-USD and real-USD will approach the actual transaction costs and no less.

Because crypto-USD also pays interest buying and holding crypto-USD can cover the entirety of the transaction fee.  Without the exchange-rate-risk, this also means that people seeking a return on their investment and trust that BitShare (like bitcoin) is unlikely to become completely worthless (due to its value being derived from the service it provides) can safely deposit USD and get a return on investment that is probably better than they get at their bank!   Note that the value of a crypto-USD is almost entirely decoupled from the value of a BitShare and would not be negatively effected even if BitShares dropped by 99% relative to USD (I can back this up).  The only thing that could impact the value of crypto-USD is a complete failure of the Bitcoin algorithm.   More likely is that BitShares will go up relative to USD and thus drive up interest rates up while increasing liquidity of crypto-USD.  In fact, it is more likely that your FDIC insured bank will fail and do a bail-in than this distributed bank would fail.

What am I missing and how could I improve on it?   I will actively answer any questions you may have.