Bitcoin Forum

Bitcoin => Bitcoin Discussion => Topic started by: chaord on January 25, 2011, 06:10:39 AM



Title: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 06:10:39 AM
So I have been thinking about this for a while.  Bitcoin is definitely the best and arguably most successful implementation of a decentralized digital commodity that any of us have ever seen.  However, even Bitcoin is not perfect (I wish it were).  Therefore, I'm starting this thread for us to brainstorm and determine exactly what characteristics an ideal decentralized digital commodity would have.  

Since ultimately money is just a tool, I think it would be helpful to design the perfect tool, even if it can not be fully implemented yet.  My goal here is to create a quasi-scoring mechanism that we can use to rate the "idealness" of bitcoin and any of its (future) competitors.  Here's what I have so far:

Please check the wiki https://en.bitcoin.it/wiki/Ideal_Properties_of_Digital_Commodities for the most updated list (and explanations).

I'm not an expert, and this is not exhaustive (yet).  So please comment below (or update the wiki).

I know it might seem like some of these things are mutually exclusive, but time and time again humans have made the "impossible" possible, so feel free to think hypothetically here.

One would think that if we could design (or improve) a commodity like bitcoin such that it would receive all A's, people would adopt it on a massive scale.

Anxious to see what you guys think.



Title: Re: properties of an ideal digital money/commodity
Post by: FreeMoney on January 25, 2011, 06:31:57 AM
Bitcoins are very recognizable. It's a fact about the world that they are currently unrecognized.

Other than that I disagree about auditability. If you want to open your records how is that hard? Oh, I guess you can't prove that you don't have another address with more money than you are claiming, is that all you mean?

Anonymity doesn't seem to warrant an F at all, if you have a way to earn anonymously, you can spend anonmyously pretty easily. And even if people knew you had the coins at some point it's easy to flip them out for different ones at mtgox or something. Then they would need mtgox help to get you, but you could also do it three other places.

What would you rate PP on anonymity? F- I guess, are they really that similar?


Title: Re: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 07:09:06 AM
Bitcoins are very recognizable. It's a fact about the world that they are currently unrecognized.
Agree.  I was more referring to the fact that most mainstream people can hardly wrap their head around paypal, much less bitcoin.  Yet paypal is far more recognizable as a products/brand.  I changed it to a C-.

Quote
Other than that I disagree about auditability. If you want to open your records how is that hard? Oh, I guess you can't prove that you don't have another address with more money than you are claiming, is that all you mean?
By auditability, I meant it would be nice to do instant "approval" for transactions.  Eg, if you are supposed to pay me 30 BTC from an account you specify, it would be nice if I could check that account and make sure that you have at least 30 BTC in it, ideally without revealing your full balance.  Just thinking out loud here.  Maybe we should get rid of this property requirement all together?

Quote
Anonymity doesn't seem to warrant an F at all, if you have a way to earn anonymously, you can spend anonmyously pretty easily. And even if people knew you had the coins at some point it's easy to flip them out for different ones at mtgox or something. Then they would need mtgox help to get you, but you could also do it three other places.
I was probably too harsh here.  However, currently with bitcoin your transactions are easily traceable unless you choose to obsfucate them.  I'd like something the other way around:  untraceable, unless you explicitly choose to divulge transaction info (much like physical cash is today).  I changed this to a C

Quote
What would you rate PP on anonymity? F- I guess, are they really that similar?
Good point.  PP would definitely have been the F, not bitcoin.  Changed bitcoin to a C, since it takes some serious effort to remain anonymous.


Title: Re: properties of an ideal digital money/commodity
Post by: FreeMoney on January 25, 2011, 07:45:26 AM
Agree.  I was more referring to the fact that most mainstream people can hardly wrap their head around paypal, much less bitcoin.  Yet paypal is far more recognizable as a products/brand.  I changed it to a C-.
 
By auditability, I meant it would be nice to do instant "approval" for transactions.  Eg, if you are supposed to pay me 30 BTC from an account you specify, it would be nice if I could check that account and make sure that you have at least 30 BTC in it, ideally without revealing your full balance.  Just thinking out loud here.  Maybe we should get rid of this property requirement all together?
 
I was probably too harsh here.  However, currently with bitcoin your transactions are easily traceable unless you choose to obsfucate them.  I'd like something the other way around:  untraceable, unless you explicitly choose to divulge transaction info (much like physical cash is today).  I changed this to a C
 
Good point.  PP would definitely have been the F, not bitcoin.  Changed bitcoin to a C, since it takes some serious effort to remain anonymous.

That's all reasonable.

I think it's important to keep in mind that recognizability, auditability, anonymity, and theft are all improvable by building support systems or even just education.

The really essential monetary properties are nearly all aces. I agree that divisibility isn't perfect. Optimal I guess would be an arbitrary precision real number so you could send 1 trillionth of a coin or 1/3.


Title: Re: properties of an ideal digital money/commodity
Post by: theymos on January 25, 2011, 08:13:28 AM
Under what circumstances would you give Bitcoin an "A" rating in "theft"? It already seems better than anything else.

Bitcoin's anonymity will be much better once most everyone is using e-wallet services. The pooling required for true anonymity is then taken care of automatically, without any explicit mixing. Selective auditing will then be possible, too.

The incentive structure for miners and other network participants is an important factor in Bitcoin's long-term viability, though analyzing this is difficult. I've been thinking that the halving of block rewards might have been a mistake: not because there is any risk of problems due to deflation, but because distributing an "inflation tax" to miners increases system security to a degree that I'm not sure transaction fees can match.


Title: Re: properties of an ideal digital money/commodity
Post by: mpkomara on January 25, 2011, 08:29:39 AM
Recognizability is an A.  I think FreeMoney and I disagree with how you've defined "recognizable".  If I send you something, the network either recognizes it as a bitcoin or rejects it with near 100% accuracy.   To define recognizability, as you have, as something that is widely accepted is to say that a currency should be recognized if it is recognizable.  I think you either drop the category completely or you lump the better definition with unspoofability.


Title: Re: properties of an ideal digital money/commodity
Post by: caveden on January 25, 2011, 08:55:20 AM
Another ideal is free transactions. Bitcoin might have them now, while inflation rate is high, but eventually they might get expensive, as so much computational work has to be done in order to generate.


Title: Re: properties of an ideal digital money/commodity
Post by: caveden on January 25, 2011, 09:00:28 AM
The incentive structure for miners and other network participants is an important factor in Bitcoin's long-term viability, though analyzing this is difficult. I've been thinking that the halving of block rewards might have been a mistake: not because there is any risk of problems due to deflation, but because distributing an "inflation tax" to miners increases system security to a degree that I'm not sure transaction fees can match.

I agree with you. That's why I think it would be a good idea to try to create an artificial scarcity for transactions, while we still can: https://www.bitcoin.org/smf/index.php?topic=1865.0


Title: Re: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 03:44:23 PM
Recognizability is an A.  I think FreeMoney and I disagree with how you've defined "recognizable".  If I send you something, the network either recognizes it as a bitcoin or rejects it with near 100% accuracy.   To define recognizability, as you have, as something that is widely accepted is to say that a currency should be recognized if it is recognizable.  I think you either drop the category completely or you lump the better definition with unspoofability.
Agreed. 


Title: Re: properties of an ideal digital money/commodity
Post by: ptd on January 25, 2011, 03:52:23 PM
Someone should put this table on the wiki.


Title: Re: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 03:55:19 PM
Under what circumstances would you give Bitcoin an "A" rating in "theft"? It already seems better than anything else.
I got rid of the "theft" category entirely and split it between Storage and Unspoofable

Quote
Bitcoin's anonymity will be much better once most everyone is using e-wallet services. The pooling required for true anonymity is then taken care of automatically, without any explicit mixing. Selective auditing will then be possible, too.
Anonymity in the current bitcoin implementation requires trust of some third-party e-wallet service with lots of volume.  I think an ideal system can do better.

Quote
The incentive structure for miners and other network participants is an important factor in Bitcoin's long-term viability, though analyzing this is difficult. I've been thinking that the halving of block rewards might have been a mistake: not because there is any risk of problems due to deflation, but because distributing an "inflation tax" to miners increases system security to a degree that I'm not sure transaction fees can match.
I'm not quite sure I understand this.  Could you please elaborate?


Title: Re: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 04:15:56 PM
Someone should put this table on the wiki.
Done.  https://en.bitcoin.it/wiki/Ideal_Properties


Title: Re: properties of an ideal digital money/commodity
Post by: Hal on January 25, 2011, 06:27:07 PM
I'd add speed, you shouldn't have to wait too long for a transaction to go through. Bitcoin gets a B-, it's fast enough for a lot of things but too slow for some.

Scalability, could it be used for every transaction, everywhere in the world? In its current form I'd say no, but possibly with the addition of lightweight clients. So I'll say C, has potential but not there yet.


Title: Re: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 06:32:38 PM
I'd add speed, you shouldn't have to wait too long for a transaction to go through. Bitcoin gets a B-, it's fast enough for a lot of things but too slow for some.

Scalability, could it be used for every transaction, everywhere in the world? In its current form I'd say no, but possibly with the addition of lightweight clients. So I'll say C, has potential but not there yet.

Good points!  I'll add them.


Title: Re: properties of an ideal digital money/commodity
Post by: theymos on January 25, 2011, 07:05:12 PM
I'm not quite sure I understand this.  Could you please elaborate?

It is vital that the combined computational power of all honest Bitcoin generators is higher than that of any single attacker. In the future there will not be any inherent reward for generating a block -- generators will only get the fees from transactions. Possibly this fee alone will not pay for enough computational power. It may be better to continue "taxing" all existing bitcoins at a very low rate by continuing to print more bitcoins forever.

Probably Bitcoin can survive without this change, but I think a more perfect system would include it.


Title: Re: properties of an ideal digital money/commodity
Post by: chaord on January 25, 2011, 07:24:14 PM
I'm not quite sure I understand this.  Could you please elaborate?

It is vital that the combined computational power of all honest Bitcoin generators is higher than that of any single attacker. In the future there will not be any inherent reward for generating a block -- generators will only get the fees from transactions. Possibly this fee alone will not pay for enough computational power. It may be better to continue "taxing" all existing bitcoins at a very low rate by continuing to print more bitcoins forever.

Probably Bitcoin can survive without this change, but I think a more perfect system would include it.

Ahh!  I see what you are saying now.  Thanks for the clarification.  Yeah, I too have wondered about this.


Title: Re: properties of an ideal digital money/commodity
Post by: ribuck on January 25, 2011, 07:48:17 PM
In the future there will not be any inherent reward for generating a block -- generators will only get the fees from transactions.

When the reward drops from 50 to 25 BTC per block, we will see whether or not it has any significant effect on the difficulty level. Until then, I don't think anyone needs to worry or speculate, because hard evidence will come soon enough.


Title: Re: properties of an ideal digital money/commodity
Post by: caveden on January 26, 2011, 08:44:05 AM
You know what would be nice?
Try to apply the same table to gold (or precious metals in general), just to see how better bitcoin stands. :)
Maybe even comparing to government fiat money too, but that might too much humiliation... :D


Title: Re: properties of an ideal digital money/commodity
Post by: abstraction on January 26, 2011, 11:17:47 AM
Possibly this fee alone will not pay for enough computational power.

Won't the market regulate the fee by itself and provide enough computational power in doing so? I think the scenario will be processing centers not accepting transaction fees below (A) if the product (volume * floating average fee) falls below the operating cost (B) (or some small % above (B)). Some people will want to have a very fast transaction (like when paying at a store) and will accept a fee for the priority. Others can tolerate the wait and offer a smaller transaction fee. Every mining bank (transaction center) will compete for the higher priority transactions since they offer more. At some cut off point (A), some mining banks will not accept transactions below that level. Other more efficient mining banks will accept transactions below (A) until they reach a point (C) where they, too, can't afford to operate. It gives the incentive to find more efficient processors to either better compete in the same fee tiers as everyone else or undercut the market fee price.

Right now, it's not worth it for me to mine for coins, but it might be worth it to accept 0.009BTC transaction fees and leave my client running, providing computational power. Have I correctly understood how fees are processed?


Title: Re: properties of an ideal digital money/commodity
Post by: ribuck on January 26, 2011, 12:03:44 PM
Right now, it's not worth it for me to mine for coins, but it might be worth it to accept 0.009BTC transaction fees and leave my client running, providing computational power. Have I correctly understood how fees are processed?
Provided you understand that you only win the fee if you solve the block (in which case you also get the 50 generated coins).


Title: Re: properties of an ideal digital money/commodity
Post by: joe on January 27, 2011, 02:43:41 AM
Fees will not work because they are easily circumvented by pooling and doing back-end bookkeeping, such as mybitcoin already does now. What will really happen--and this would be more fair than transaction fees--is miners will collectively decide to continue generating (and accepting) new coins past the 21 million limit set by Satoshi.

At first this seems like it is breaking a core principle of bitcoin, but it is completely within the spirit of bitcoin. The new dynamic inflation rate will be reached based on the agreement of 51% of CPU power. If miners set the inflation too high, the currency on that high-inflation chain will lose credibility, thus the return for miners will lose real value. If miners set the inflation too low, you have the original problem of insufficient mining activity. So the end result will be a fair inflation rate that coin users can live with and miners will work for.

It deserves a discussion on what happens when the chain branches:

Let's say I branch the chain today to provide a bigger reward for new blocks, say 100 BTC. Then you'll have 2 chains: the main chain and the new chain. At this point, all coins in current and future existence will fall into 3 categories:
(1) valid on both chains (default, all previous coins are valid on both chains)
(2) valid on only the main chain (coins generated on main chain, 50 BTC reward) and
(3) valid on only the new chain (coins generated on new chain, 100 BTC reward).

The price of BTC is currently 0.40 USD. The price for type (1) coins will remain at 0.40. The SUM of the price for type (2) and type (3) chains will equal 0.40 USD. The difference in prices in type (2) and (3) coins will determine which chain the market eventually accepts.

So... for right now, we don't need to do anything. But when the major mining software makers decide to allow different block rewards other than the 50 BTC, 25 BTC ladder set up by Satoshi, then we will have to reprogram the regular clients to deal with branched chains gracefully and communicate to the user which branches their coins belong to.


Title: Re: properties of an ideal digital money/commodity
Post by: theymos on January 27, 2011, 03:44:12 AM
Having two chains purporting to be the "real Bitcoin chain" will be really messy, and I think the market will avoid taking that step unless the subsidy ends up being very wrong. It could happen, though. I'd prefer to anticipate the future market as much as possible in order to avoid that.

Quote from: abstraction
Won't the market regulate the fee by itself and provide enough computational power in doing so?

Not really. The block subsidy is a mandatory tax on all bitcoins, whereas the transaction fee is controllable by senders.

Right now the network is paying for 50 BTC of mining per block (not counting fees). If we instead pay 25 BTC per block, then we will get less computation, and we will be that much more vulnerable to attack. The reduction in computation happens because everyone who does not make 25 BTC of profit right now is eliminated after the reduction.

If there is no subsidy, then there will need to be 500 transactions (with 0.01 fees) per minute for the network to get the same amount of computation that a 50 BTC subsidy would buy. Otherwise, the computation provided by miners will be reduced because blocks will be worth less and people will be eliminated from the market.

This entire issue may be meaningless if it turns out botnets can compete against custom hardware, since botnets have no cost and will mine for nearly any amount.


Title: Re: properties of an ideal digital money/commodity
Post by: FreeMoney on January 27, 2011, 05:38:34 AM
I just had an idea. It's probably not a possible change, but maybe, interesting anyway.

Right now you only pay to get your tx included in the chain, but the tx benefits in terms of security for every block added after it.

It would be cool if when you paid a fee it went half to the miner who included it, 1/4 to the next block finder, 1/8, 1/16 until below some limit. This way each user can increase the incentive to put his block deeper and deeper, potentially for a while, but still for a finite fee.