Topic: *old* BitShare Economic Theory 10 BTC bounty to prove me wrong... paid. - page 2. (Read 10080 times)

See the following video explaining bitshares: http://www.youtube.com/watch?v=RPbSaznUbg8&feature=youtu.be

The purpose of having an anonymous peg was merely to simulate the 'invisible hand of the market' that was confusing the debate, the idea being that there is just 1 eOil company, but 1000's of them all competing to provide goods to the market.  The purpose of making it anonymous was to make sure there was no 'trust' or 'contract' between eOil and the users of crypto-Oil.   So, the biggest problem you have forced me to face (whether you realize it or not) is that my pricing mechanism either forces everything above parity until there is no longer a connection or to 0 with most of the dividends going to the attacker.   

Clearly my pricing mechanism is broken... but never fear bytemaster is here.. and he has a new pricing strategy that is far better!

The source of our problems is establishing the minting rate... right now it is entirely one sided with no market forces to push it to the proper rate.  This occurs because the exchange was based upon trading existing crypto-Gold for BS and attempting to 'tack' minting onto an entirely un-related exchange.   So, here is my new strategy for controlling mint rates:

1) It is entirely pointless to have an exchange between BS and crypto-Gold because we know the exchange rate 'exactly' based upon the dividends... thus my old exchange was 'pointless'.
2) What we really want is an exchange between 'minting' and 'melting', thus when you melt you reverse the process of minting by turning crypto-Gold into BS at some 'rate'.   

So instead you turn the market into an 'mint' vs 'melt' market...
    - when I melt...  I want the highest possible ratio that allows me to buy the most gold.
    - when I mint.... I want the lowest possible ratio that allows me to sell the crypto-Gold for gold.
    - the two market participants must agree to the mint/melt price.
   
* note * the minter is no longer given an option to cover at his mint price, thus there is no longer a 'short' position to be covered.

So how would this work?
     - Initially there is no crypto-Gold... thus, you can mint at any price you want you are creating a new currency
     - This creates new crypto-Gold which the holder does not want to melt for some time.
     - When there are no melters, you can only 'mint' at the same or higher ratio (no debasing)
     - Once there is an offer to melt, that is the highest (and lowest) you can mint at (melter/minter just exchange).
     - When there are no minters, just melters, then you can melt at the current rate.

Why would someone mint in the first place?
     - to create a new peg and establish a new crypto-Gold.  Those who have crypto-Gold know they paid 1 gold coin for it, and it was at parity, they are not going to take a loss if they can help it.
     - If the value of gold goes up it requires more BS to back it, thus you can always mint at a new higher rate.
     - If the value of gold goes down, then crypto-Gold will be 'over-backed'.  No one will want to 'mint' at the old backing... and any crypto-Gold holders will not be able to 'melt' over the mint price,
          both parties realize that the only price they can 'agree on' is the current market price of gold.   This would allow 'one in' and 'one out' of the crypto-gold market.

Why would someone melt from a high-yield (over-backed) crypto-Gold balance into BS?  
     After all they could just trade their crypto-Gold balance at full value without melting!  No one will melt unless they are forced to melt and thus we need to find a way to force melting.  I think the way you do that is to only allow transfer between addresses while in BS form.  Thus, you can only transfer value through BS, but you can store it as crypto-Gold.   

If we ignore the technical challenge of turning every gold wire-transfer into two market orders... this approach should work.

So, how would eGold operate? 
    - initially they would Mint as much crypto-Gold as they could sell for gold with a fee.... thus they have acquired a lot of gold.
    - if the price of gold goes up... egold would sell some of their gold for BS and offer to mint for anyone willing to melt.. there would be a spread (egolds profit)
    - if the price of gold goes down... the crypto-gold would now be backed by too much BS... so the holders are earning a lot of interest... this causes a rush into
      the crypto-gold market with everyone wanting to 'mint'... but they could only mint at the lowest melt price.. this forces them to mint high and for the melter to melt low so
they can agree on a price... thus parity will be established with supply and demand.


I am re-opening the bounty on this thread!

Thanks for the award!  The attack requires greater capital without dust so eliminating dust bid minting prevention is certainly a good first step.  I still think I would succeed in either decoupling crypto-Oil or driving eOil out of business so I'm going to offer my final thoughts for you to consider:


Exactly! -- or anyway I would prevent new people from trading it at parity.  What you may be missing (and your simulation game is certainly missing) is that a large portion of any market (and ESPECIALLY consumable markets) isn't made up of speculators -- its made up of people who NEED crypto-Oil for a definite purpose.  In the case of crypto-Oil, probably so they can send it to someone somewhere who can then use it to receive physical oil from the tanker to ensure his refinery keeps running.  Crypto-oil is certainly better the actually shipping the oil, and it is also better then BS or bitcoin because it theoretically tracks oil.  So the refinery can hold 6 months "reserve" in crypto-oil "safely" and draw on it every day.

So when eOil says they cannot fill the order because they do not have crypto-Oil to sell these people aren't going to say "oh ok I guess I'll buy banana futures".  They are going to go to a competitive instrument -- like a trust backed colored-coin running over the bitcoin network.

And if eOil is doing say 1 million a day in transactions that takes a lot of people.  Sales people.  IT, etc.  And financial commitments like advertisements.  You can't just turn the people off and then turn them back on when the attacker gets bored and gives up.  So the attack destroys eOil because they lose customers and must pay their employees to do nothing,  because eOil can't sell them crypto-oil currency if they can't get it from the market.  One simple way to get it is to offer to pay more for it of course...

let me put it in 2 lines:

tl;dr: Minting is a key feature of your system.  Therefore you must think that it has some important function.  Whatever that function may be, the attacker can use blocking bit-dust bids and fills to create long periods where that function is not available.  This cannot be a good thing. :-)

Not quite because cash is a revenue stream of 0...  I am trying to come up with an analogy.

When this project is completed, I will be one of the first to read the white paper.  Very interesting ideas.

Correct.


Just because it goes up due to momentary scarsity does not mean you broke the peg (remember my +/- 10% trading range).
The margin between crytpo-Oil and oil is what provide profit motives for people to bring goods into/out of crypto-Oil based upon supply and demand.

So all you did is partially fill one order ABOVE MARKET VALUE.  Because there exists bids above market value we can assume the following:

0)  While eOil has crypto-Oil they will sell at a profit.
1)  eOil will not buy nor issue ( you stopped nothing they wouldn't have done anyway )
2)  Everyone else has profit motive to sell and bring the price down.
3)  You have opportunity cost by not selling the full amount.

So the only way you can claim that you are a rational, profit-seeking, actor is if you have some way to profit from this which is greater than the opportunity cost.


NOTE:  Those who purchase crypto-Oil have only one requirment:  its value shall not fall below parity by more than a couple of percent.  Therefore, breaking the peg to
the up side actually means they profit.  You cannot hurt someone who owns crypto-Oil by pushing the price above parity and keeping it there.  You would be losing money
every day you held the peg due to transaction costs. 


So, lets look at what it would take to ACTUALLY pull off this attack:

a) you would need to purchase crypto-Oil above market rates and hold it at opportunity cost.
b) you could drive everyone to sell their crypto-Oil for BS at a profit as the value of oil falls.
c) you would discourage / prevent new people from trading their oil for crypto-Oil.
d) you would cause eGold to issue crypto-Oil 2 which was cheaper than crypto-Oil yet also would never fall below parity.
e) everyone with crypto-Oil would trade into crypto-Oil 2 at a profit.
f) market will continue as normal with no one taking any losses except the attacker.
g) the attacker would then have to sell crytpo-Oil at a potential loss, and then start buying crypto-Oil 2 above market rates...
h) eventually the attacker will be bankrupt while everyone else is receiving higher dividends thanks to all of the transaction fees
   the attacker is racking up to no real effect.


So you have failed to put eOil out of business or steal any money from anyone.  You did succeed in disrupting the market, forcing it over to another crypto-Oil 2 which would cause ineffeciencies and confusion.
Your attack would only work once you had cleared out all other sellers, and thus would be very innefective in any large market like gold, usd, etc.

Your attack takes advantage of the fact that you can issue 'dust' orders.
Your attack is not profitable, and is more along the lines of DOS.
Your attack would not be possible if you could not place 'dust' orders because you would run out of capital too quickly by filling orders above market rate.
Therefore, you would need rules like bitcoin has to prevent 'dust' transactions from propagating or being included.


Therefore if we had the following rules:

You can issue against multiple bids at once provided they are at the same price.
All bids must be a multiple of 2x the minimum transaction fee. This forces grouping of bids in the same 'range' to be the same effective price.
Define a minimum 'order amount' to prevent dust orders... this order amount could be defined by 'miners' and therefore subject to 'market forces'.

The block chain is clearly not an effecient datastructure for micro or high-speed transactions, so lets assume that there are a number of
'spam filters' that the clients could run to detect this kind of DOS attack.

Given that assumption we can conclude that when it comes to 'trading' the blockchain will only be used by those who can trade in multiples of $100 and anyone
wanting to trade in smaller lots would have to do so 'off chain' on a Open Transactions server.


Does your attack work if you don't have access to dust?


Considering you failed to drive eGold out of business or steal money from anyone *and* your attacker was not motivated by profit and took constant losses, technically I don't think you should win the 0.25 bounty.

That said, this was a good attack and provoked thought and rule changes so I will award you 0.25 because it was still valuable.  It will be paid in a few hours.

Not quite how the dividend payments would be implemented.  Clearly it would be unsustainable to create actual
transactions for all dividend payments due... that would be 1 transaction per output and the fractions would be
below the size of .0000000001 bitshare.

The miner would only ever start out with 12.5 Bitshares + 1/2 of the fees, the matching part that is paid as dividends is simply 'noted'.

Now when a user wants to collect the dividends on a particular output, they must create a new transaction that 'spends the dividends' from
an existing output.  The software will look at the 'coin-age' of that output, total up all dividends for all blocks solved since the block
that included that output and calculate a 'total dividend payment'.  This dividend payment could then be transferred to any other address.

You can only 'spend' the dividends from an output with over 120 confirmations.   Even though you can spend the 'balance' at any time, just like in bitcoin,
    you can still go back and collect your dividends later.

Now this is where it gets interesting, because all transactions have a fee, it is impossible to redeem dividends that are too small.  Therefore,
    if your turn-over on an output (say 1 BTC held for 1 block) would result in a potential dividend of   12.5 / 21,000,000 and thus be 'unspendable' until
    the value of a BS rose dramatically.

As a result if you want to collect dividends you want to concentrate your balances and hold them for a long time.

Some more design work may need to be done to effeciently distribute dividends.  Another alternative is to have dividends accumulate separately in every output and
then follow spends.  Dividends would have to be expressed as Average Coin Age + Amount because I cannot convert them to balances until they reach 120+ confirmations
or else a chain split would invalidate a ton of transactions. 

Clearly there is some R&D on how to effeciently distribute dividend payments, but I believe that is solveable as it is just math and book-keeping.



   Lets try to clean up your example:

   Assumptions:
   1 BitShare = 1 BTC
   1 BTC = $100
   X ==     I will assume a $10 dollar bill. 

   User wants to create a crypto-X share, they can only create a crypto-X share in response to the highest bid for crypto-X that has at least 6 confirmations... so
   presumably somewhere in the blockchain there exists an output of the form:

   Output:  [BID]
        10 BitShares  spendable by any transaction that pays 1 crypto-X to address A


   User then create a new 'issue' transaction based upon that Output:
   
   Inputs:
         [BID] 10 Bitshares  (from bid)
   Outputs:
         1 crypto-X payable to address A
         10 mortgaged BitShares paying dividends to all crypto-X shares (not just the 1 we created), redeemable with 1 crypto-X


   To doublecheck out math:
     Value In:   10 BitShares
     Value out:  1 crypto-X  worth 10 BitShares
                 10 Mortgaged BitShares redemable with 1 crypto X
                 
     No value created or destroyed.

   The person who made the Bid received 1 crypto-X, and paid 10 BS, the person who issued the currency received 10 mortgaged bitshares.


If 1 BS pays a dividend of .01 BS per block and one crypto-Gold pays .10 BS per block, which one is worth more to own?   Clearly crypto-Gold pays 10x as much as BS and is thus worth 10x as much despite all payments being made in BS.


 
  I think the disconnect is that you are comparing worth by looking at 'absolute' dividend rates instead of 'relative' dividend rates.  Given two revenue streams paid in 'undefined' good X, one pays 3x the other 1x, which revenue is worth more and by how much assuming x != 0?   The value of X is entirely factored out of the value comparison and thus irrelevnt. 


   
   The dividend rate paid to early adoptors will indeed be high, but don't forget there will be One Million BitShares issued to those who invest pre-launch and help make it a reality, so the initial dividend rate will be 25/1,000,000 every 10 minutes. Also, like any new business those who recognize the value and buy in first see gains.  Those who 'wait' until it is safer have opportunity costs associated with being less risk-adverse.



Ok, the usability by the average user is HUGE and is actually VERY simple.  They will have a user interface that displays the current 'price' of all crypto-X in terms of BitShares by pre-calculating   crypto-X dividend rate / BS Dividend Rate and therefore users never even have to 'see' the dividend rates, instead they see relative prices derived from those dividend rates.   Thus they can tell when crypto-Gold is over-valued compared to crypto-USD without ever having to think about BitShares or Dividends. 

The only people that really need to understand the dividend rates / etc are the early adopters / investors who care about where the value is coming from.  Once they understand the machanics they can 'trust it' and start trading based solely upon absolute prices drived from interest rate ratios.   Thus all of the complexity is only present *now* when I have to show everyone why prices track as they do. 

If others can sell to fill at a profit then so can I.  My job is done, I have caused the currency to rise in value due to scarcity.  At 10 minutes a pop, if I bid or fill 1 satoshi worth of crypto-Oil I can go on essentially forever without materially changing my position.

Switching is an interesting proposition.  But I think technically I win our .25BTC if my attack forces everyone to abandon the crypto-Oil currency and move to another one.  And I think I win in intention as well because a currency switch can't happen quickly because the backing BS is tied up.  Also, you are not considering the human factor -- there is tremendous market inefficiency and cost involved to get people to switch over.  No one would trust eOil or crypto-Oil again...

And people would be leery with good reason, because I can block minting on EVERY crypto-Oil currency issued after its creation is announced on the blockchain.  So eOil would be forced to create not just another currencies, but a whole bunch of different ones, the number depends entirely on how much eOil is able to mint during currency creation -- this is basically just how much $ it has on-hand every day due to new sales.  So a new currency is created every day... and the whole system has to trade between the N crypto-oils by looking at the BS backing each one.  In other words, they are trading BS in a crypto-oilN wrapper.

 

Even if you only sell the smallest dust amount (similar to bitcoin's rules...so not new)... you will not be able to prevent others (not eOil) from selling to fill at a profit.

Eventually you either own all crypto-Oil, sell all of your crypto-Oil, or everyone else switches to crypto-Oil 2 with no losses.

Let me try to summarize the primary flaw in your attack:  you cannot force eOil to buy, thus you cannot force them to take a market position that would cause them to loose money.
What you can do is buy up all crypto-Oil (presumably at above market rates that profit eOil or someone else).
Once you own it all you must sell to prevent someone else from minting.
You are not actually earning any more dividends due to bids on crypto-Oil going higher and higher.  The only time dividends change is as a result of minting or covering.. therefore the attacker has constant 'opportunity cost' (violating the profit seeking rule).
The only people who "suffer" from your attack are those who hold oil and want to buy dividend-paying crypto-Oil.   You will have set the cost of dividend-paying crypto-Oil so far above parity, that they will not buy from you.  You are the only one left owning any, everyone else has sold to make a profit.
This will create demand for a new crypto-Oil which could be started by anyone.  The new crypto-Oil2 would be issued by eOil at parity and offered for sale at 1% above market price.  All of your customers would 'flee' and purchase crypto-Oil2 and use that until you lowered your prices... to parity.  Thus it
cost you a lot of money, but otherwise didn't harm anyone else.


If eOil refuses to buy above parity (why should they?) then the only thing your attack can do is increase the value of crypto-Oil.. something everyone who wants to withdraw crypto-Oil cannot complain about!

I think your attack is invalid even without the new rule changes I just proposed:

If you are filling every order with your crypto-Oil, then eOil has no need to aquire any crypto-Oil to provide to the market, presumably you are able to satisfy demand and thus keeping the price pegged.

If I were eOil, I would just sit back and watch because you are keeping the price stable for them.
If you attempt to fill "just a little" that is also fine: eOil will make profit on what it fills until it runs out and you make profit on what you fill.

Once eOil runs out of crypto-Oil, they can place a bid to buy it at market price (1% below what you are selling for).  They would be buying for less than they sold so this would be sustainable, of course their
bid would not be accepted by anyone until *you* satisfy all demand above parity.   In the mean time other people (not the 'attacker' nor eOil)  who own crypto-Oil could also sell at a profit. If there is no-one else,
    then the joke is on the attacker who continues to own crypto-Oil which is paying dividends below what he could receive by accepting the highest bid and thus he maintains a constant 'opportunity cost' and thus is not seeking profit but an arbitrary increase in the value of crypto-Oil.

From the market's perspecitve, having crypto-Oil valued so high above market price will bring in other sellers until there were no more holders of crypto-Oil.

You, the attacker, will have 'cornered the market' for crypto-Oil but to maintain your control you would have to 'refuse to sell' despite bids above market value... thus you have an opportunity cost (are losing potential dividends),
   that you could be making by converting your crypto-Oil to BitShares.   You will continue to lose money until the market price of crypto-Oil falls to the point where eOil can now buy profitablly (at parity).

If you decide you just want to 'destroy' crypto-Oil and never sell, 'just to spite' them.   Then those looking for crypto-Oil will just issue crypto-Oil2 which you will have none of and crypto-Oil2 will carry on following the price.   No holders of crypto-Oil lost money by your actions and you had a ton of opportunity cost.  eOil aslo did not lose money.

I *think* this is a variation of an attack I thought of in the shower this morning (posted and fixed (without rule change) in the new bounty thread)...  so let me first rephrase the attack how I understand it and then offer corrections.

The goal of the attack is to prevent eOil from issuing new crypto-Oil.  The result of not being able to issue new crypto-Oil is that crypto-Oil would diverge UP from the value of oil as more and more 'demand' for crypto-Oil must be met by the same 'supply' of existing crypto-Oil.

To execute this attack all that must occur is some trade of crypto-Oil for BitShares at the highest bid price to BUY crypto-Oil.
So the attacker who has some nominal amount of BS and crypto-Gold places an offer to buy and then accepts his own offer, or he must simply sell his crypto-Gold for the highest Bid.

If all he is doing is 'selling' crypto-Gold then he will eventually run out.  Because he would be selling at the highest-bid and taking them out, he would be increasingly selling below market value (at a loss).

If he is attempting to 'sell to himself' and thus never run out, then there are some other challenges he would face.

How would the market participants respond:
1) the attack wouldn't be 'free' due to transaction fees (2 of them, bid + acceptance)  (this isn't a rule change, it works just like bitcoin)
2) the issuer could offer a higher transaction fee to make their issuance take priority  (miners would ignore the sale with 0 fee in favor of the issuer with some fee)
3) the other holders of crypto-Gold could sell into the attack and thus claim the attackers above-market bid before the attacker could.
         * they could get priority by offering slightly higher fees than the attacker (made possible by above market bid)
         * if the attacker compensated by increasing his fees... it would make the attack cost more and thus unsustainable.

If the attacker wasn't placing bids, just 'filling' existing bids to buy cyrpto-Oil by providing crypto-Oil, then eventually he would runout of crypto-Oil.


All of that said, there could be some clarifications / tweeks to the rules that may make this attack 'clearly' non-profitable.

1) To issue a new higher bid, it must have a value greater than the old-highest-bid by 2x the transaction fee included.   I don't like this rule because 2x could be seen as 'arbitrary',
   but considering the value of 'x' is decided by the market it isn't as bad as it could be.  Like Bitcoin, there would have to be a min trx fee that most miners require.

     - net result is that the attacker exposes themself to a loss of 2x the transaction fee
     - the attacker creates a profit opportunity for someone else to make profit by accepting his above-market bid (selling him crypto-Oil) with fees up to 1.99 times the transaction fee used to create the bid.
     - therefore, the attacker would require 3x the normal transaction fee to achieve this attack:   1x  for creating the bid and 2x for accepting his own bid without the risk of losing to someone else.

2) I suppose the attacker does not need to issue a higher bid, they could simply match the current high bid.
     - this could be mitigated by processing bids of equal price in a first-come-first-serve basis which means he would have to use his crypto-gold to fill everyone elses order first.
     - presumably the highest un-accepted bid is also 'below market rates' and thus he would be selling his crypto-Gold at a loss.


3) I suppose he could 'fill' .00001 of the highest bid and thus tie up the network legitmately for a long time with only a small amount of crypto-Gold.
     - this could be mitigated by requiring all fill order amounts to be some multiple of the transaction fee.  It wouldn't make sence to pay a $1 trx fee to buy $1.01 of crypto-Gold unless you were
          attempting to 'attack' as cheaply as possible.
     - this multiple could probably be on the order of 10 x... though I hate to introduce random 'magic numbers' to the rules.  But I suppose this is no different than
       what bitcoin must do to prevent spam and dust in Bitcoin.

Ok, suppose I am a major holder of crypto-oil.  And suppose crypto-oil's demand is high because it has tremendous utility.  

Normal case:
Essentially, people are giving eOil (and others) money or oil contracts and eOil is giving them crypto-oil.  Eventually, eOil is going to run out of crypto-oil and will have to buy at market.  If nobody is willling to sell at that price, eOil's bid is minted so crypto-oil remains pegged to oil.  This makes sense because eOil's backing increased.

Attack:
So during every block I fill the minimum crypto-oil denomination, stopping the minting process.  Now the only way eOil is going to get crypto-oil currency is by bidding high enough to encourage existing holders (me primarily, but other holders can of course profit from my action) to sell their shares.  I have decoupled crypto-oil from oil and once eOil bids high enough I'll sell my crypto-oil to them at great profit (5 to 10% for a few days work), buying it back at a lower price when crypto-oil re-couples to oil.

If eOil refuses to buy my high sell offer, they cannot acquire crypto-oil and so must stop doing business.

Are bitcoins IOUs as well?   Would a dividend paying Bitcoin be an IOU?   Do Bitcoins have no value?   Do they only have value because of a pump & dump scheme?   Who 'says' bitcoins have value?  

If bitcoins have value then BitShares could develop the same value, (proof: Litecoin), if enough people thought they had added benefits of Bitcoin.   Even without the support of crypto-Gold, some people may prefer the dividend paying model to the existing model.

If my currency allows the creation of an anonymous 'peg' then it clearly has much more value.  

Ok, if you attempted to use a SSD for mining you would quickly wear it out, they have read/write limits and every hash attempt would completely change every sector.

Sure, big data centers can have huge amounts of ram.  They cannot purchase that ram much cheaper than anyone else can.   What is the total amount of RAM in the largest data center?   Compare that to the total ram in ALL home PCs. 

Now do the same comparison with mining SHA256 ASICs...

lol... ok... Your whole system is IOU's... You are not giving actual gold to people for actual gold (which would be stupid and redundant)... it is all virtual-debt... If there is no broker, there is no market... nothing to crash/alter... just private sales, which have "No records of value"... What are you going to just "say" it is worth x, with no proof that you sold any for x value? You need a broker to have a "value of transactions", which only a broker displays. Seriously, you know your stuff man... where do you think the charts you see are coming from? Brokered sales, recorded and displayed in a chart.

You win...

You develop your coin, then we will see who actually knows what the hell they are actually talking about.

You are the idiot who suggested RAM as a "limit". Not sure what computers you use, but SSDs are RAM, and multi-striped, are faster than most sticks of RAM, and since RAM is cheap, cheaper than an ASIC, or CPU, or GPU... That would hardly be a "limiting factor", as you stated. Yes, on a home-computer it might... but not on any server-computers with multiple TB of RAM installed. (And ASIC's have access to RAM, by the computer they are hooked-up to. They don't need to work with RAM directly, because they don't process anything in RAM.)

Have fun playing, you C++ master! lol.

P.S. My virtual-ram is in RAM. So, yes, I KNOW the magnitude of the two... it is YOU who doesn't understand the limits you think you are imposing. Thus, not good enough to program what you think you want to program.

You clearly don't understand the order of magnitude difference between RAM and virtual-RAM in terms of performance.   
You also have no understanding of Bitcoin because every one of your arguments could be levied against Bitcoin + Mt.Gox.