Saturday, June 18, 2011

Bitcoins Considered as Virtual Gold

A commodity money, such as circulating gold coins, has two substantial advantages and two substantial disadvantages in comparison to other forms of money. The first advantage is that it is less dependent on information and trust. Gold was, until modern times, by a substantial margin the densest known metal, and it has several other distinctive characteristics, making it reasonably straightforward to distinguish gold from gold plated lead or even from gold that has been seriously debased by alloys. The value of a commodity gold coin depends on fineness and weight, and weight is easy to measure. So it was practical, indeed common, for multiple issues from multiple sources to circulate together, exchanging at rates determined by relative gold content. Knowing that a coin was from a reliable mint was convenient, but not essential.

The second advantage is that a commodity currency is less subject to manipulation than a fiat currency, whether token coins, such as most currently circulating coinage, or paper money. The issuer of a fiat money can manipulate its value either as a source of income—financing government spending via the printing press rather than taxation—or for other purposes. Inflating  a currency benefits debtors, including indebted governments, at the expense of creditors; one reason for Greece's current problems is that, without a currency of its own, it does not have the option of that form of de facto default. And manipulation of the money supply can be used for other political purposes, such as temporarily lowering the unemployment rate via an unanticipated expansion. Thus the value of a fiat currency is potentially less predictable than the value of a commodity currency, and predictable value is an important feature of the currency in which parties contract for future payments.

The flip side of this advantage, however, is that the value of a commodity currency is determined by external factors which may themselves be difficult to predict, especially over the long term. New gold discoveries, improvements in the technology for extracting gold from ore, changes in non-monetary demand for gold, all can result in changes in future price levels that are not easily anticipated. Thus the value of a commodity money is less predictable than would be the value of a fiat money controlled by competent and benevolent agents.

The other disadvantage of a commodity money is that someone has to produce the commodity. Labor and capital are expended in locating and mining gold, a real cost. David Ricardo, writing about two hundred years ago, pointed out that a tax on gold mines whose output was used entirely as money in the taxing country and provided all of its money was a burden-free tax—not only did it impose no excess burden (cost to taxpayers above receipts to government), it imposed no burden at all. A smaller amount of gold was mined but the value per ounce was greater, resulting in the same total value of gold money at a lower production cost—and it is the value, not the weight, of money that mostly determines its usefulness. Fiat money represents the limiting case, where "production cost" is in effect all tax.

There is, as I have discussed elsewhere, at least one interesting and attractive intermediate case: A private fractional reserve money. Its value is determined by the value of the commodity on which it is based, so behaves like the value of a pure commodity money; the one difference is that monetary demand plays a lower role, with each ounce of gold supporting the equivalent of multiple ounces of currency. And, for the same reason, it sharply reduces the cost of providing a monetary system compared to the cost with commodity money. But, unlike a pure commodity system, it does depend on trust and information, on being reasonably confident both that your bank note was really issued by the bank it claims to be from and that that bank can be trusted to pay off on its promises.

The reason to raise these issues now is the appearance of bitcoins, a private, decentralized system of anonymous online currency. I have been interested in the subject of anonymous ecash for many years but have no particular expertise in the current incarnation; readers who do are invited to correct any errors in my account.

Bitcoins, as I understand them, are, like gold, mined—not with pick and shovel but with computer power. Each consists of a piece of information which requires a considerable input of computing power to generate. In equilibrium the value of one bitcoin tends to the cost of producing it, just as the value of an ounce of gold tends to its production cost. Thus, like a commodity currency, bitcoins consume real resources in their production. And their value has both the advantage and disadvantage of a commodity money—not subject to deliberate manipulation, but vulnerable to changes from external causes such as reductions in the cost of computing power or mathematical improvements that make it easier to deduce the information used to create one. Like a gold coin, a bit coin can be tested for validity by the user—a process that involves both a check of its mathematical characteristics and the use of decentralized mechanisms to prevent the double spending problem.

The details of how the bitcoin system works should be findable elsewhere online. While I have an avocational interest in the relevant mathematics, I am not an expert in it, so thought it more useful to focus on the economics.

19 comments:

Miko said...

As I understand it, the algorithm used to design the problems that must be solved to create bitcoins is sensitive to the amount of time required to solve the problems and beings creating more difficult problems if they're being solved too quickly.

Jon Leonard said...

The mining protocol is more subtle: The system calls for a certain amount of bitcoins to be created in a given week, and the number of bitcoins generated per computation changes accordingly. This week's hash collision is worth less bitcoin's than last week's, both because the protocol calls for less new money each week, and also because more people are trying to mine bitcoins.

Other parts of the protocol are equally clever, but I haven't had time to think through all the consequences.

Ken Simpson said...

Great summary. It's nice to read the views of a scholar on what has been a very sensationalized topic in the press.

In my view, the fundamental cost of producing Bitcoins will asymptotically trend toward the cost of electricity required to compute SHA-256 hash values. At the moment, the state of the art is FPGA-based computation, with some groups experimenting with small runs of custom processors to achieve the lowest theoretical power consumption.

My company, LargeCoin, is taking this to the next logical place, a fully custom ASIC with high upfront investment to maximize the computation to power ratio.

Bitcoin will revolutionize commerce as much as the Internet. And like the Internet, many will criticize it in the early days for its inadequacies, which will be rectified by entrepreneurs in due course.

Peter said...

Another drawback with gold is that central banks might screw up and hoard too much gold. Sumner argued that this was the initial cause of the great depression (the Fed and the French were the screw ups).

http://oxlib.blogspot.com/2010/05/scott-sumner-its-complicated-great.html

I wonder how much the green movement are going to hate bitcoins if they become very popular and valuable. A high value will make people spend more energy to create more bitcoins. I suppose they will argue for higher energy taxes (like Ricardo). But today we have the drawback of using both gold and energy for reasons other that monetary.

But fractional reserve banking might lower the value of bitcoins in the future. How high might the money multiplier become for a completely virtual currency without any regulations?

Anonymous said...

I wonder about the mining process of bitcoins. I thought it helped handle the transaction log that verifies each transaction of bitcoin. If it does then as more bitcoins are being transferred then more computing power will be needed to handle it. The mining process seems to produce a product that has no value. If you mined for traditional commodities like gold or platinum then you have something that be sold to goods manufacturers. The mining process seems more like "minting" to me and the end result is a piece of crypto-currency that will have the value people think it is worth. Not much different than traditional fiat currency albeit created a decentralized authority which can be considered a big plus.
It seems to me that 80 picaflops of computing power being used to mine bitcoins (according to the grid computing wiki article) might be more valuable than the bitcoins that are being produced. It would be nice if bitcoins could be used to "store" computing power that could be used later. But how something like that could happen is beyond me.

Anonymous said...

Paul Bohm argues that the real value of Bitcoin is as an infrastructure for decentralized services, including PKI (public key infrastructure) and global namespaces (see namecoin). This means that the calculations performed by bitcoin miners actually add value, by making such services possible.

I still find bitcoin rather dubious, partly because I don't see how the network nodes can keep up with a constantly growing ledger of bitcoin transactions.

RP Long said...

I'm surprised to read that part about Ricardo. His argument about gold mine taxation was obviously deeply steeped in his labor theory of value, and we all know what happened to that...

As with any tax or central plan, the question isn't whether there is value at all, but rather value to whom, relative to what, at whose expense?

Perry Metzger said...

I am someone who does understand both the crypto and (much of) the economics and I am intensely skeptical of Bitcoin even though my political orientation would make me desperately want it to work. However, I think one has to be especially careful about one's beliefs when one's biases lead one to intensely want something to be true.

I would get into my objections in detail, but let me state just a few.

Bitcoin transactions are not anonymous in spite of all claims otherwise, and if anyone tells you they are, they don't understand how the system works. There are people who are about to get very badly hurt because of that. (There are ways to build a system on top of bitcoin that get around this by involving trusted third parties -- but once you have that layered on top you have no incentive not to base the system on gold or pig iron instead of integers.)

Right now, the trading price of bitcoins is much higher than the cost of minting them. That tends to make one feel that the market is far from operating efficiently. Indeed, I have friends who are very skeptical of what they call the "tulip market" for bitcoins who are giggling all the way to the bank week after week.

Also right now, the price of bitcoins is regularly fluctuating 50% in the course of a day or two, leading one to feel that they are not even remotely close to a stable store of value. Perhaps this is just because they're thinly traded, but I suspect there is a bit more to the problem than that.

Last, and most damningly: I'm seeing lots of people who don't really understand how the system works or what the limitations are diving in and setting up startup companies. This feels like it cannot end well -- again, it feels like a speculative bubble.

I have a much more detailed critique, but that's a good start.

Patri Friedman said...

There is a huge difference between mining bitcoins & mining gold. When you mine gold for money, the result of those resources expended in mining is that you have some gold, which has value for lots of other uses. If the price of gold is high because of its use as money, then some of this mining may be at a loss, but not a complete loss.

When you mine bitcoins by expending computer time &electricity, you end up with a number which has no inherent value for any other use. It's a complete loss - the resources have been wasted.

Seems to me that this makes it much worse than a commodity backed currency. And more stable than a fiat currency, but more wasteful in production.

David Friedman said...

Responses:

Re Ricardo. He didn't have a labor theory of value, despite the common assertion that he did. For details, see Stigler's article: "David Ricardo and the 93% Labor Theory of Value." Or take a look at my old lecture notes on Ricardo, webbed at:

http://www.daviddfriedman.com/Academic/Course_Pages/History_of_Thought_98/Ricardo_notes.html

So far as I can tell, his analysis of the effect of a tax on gold mines was economically correct.

Response to Patri: Gold that is mined and used for jewelery or to plate electrical contacts is providing valuable non-monetary services. But the gold that is mined, minted, and circulated as coinage is providing the same monetary services as bank notes, and producing it is a lot more expensive.

Anonymous said...

"In equilibrium the value of one bitcoin tends to the cost of producing it, just as the value of an ounce of gold tends to its production cost."

No, no, no. Bitcoin is a pure bubble - it has no non-monetary value. So in equilibrium, its value is exactly zero.

Note: this is not true of government fiat money. It has value as long as the government is functional, because the government demands taxes be paid in its money and nothing else. Nothing is forcing people to value bitcoins.

Lex Spoon said...

I really like the idea of ecash, but basing it on compute power strikes me as highly problematic. To get the properties of a currency that you desire -- such as controlling inflation and avoiding spending the same money twice -- you end up wanting a trusted third party. Once you have such a third party, you may as well have a mint and use straightforward protocols against it.

If mining currency becomes a valuable way to spend compute power, there will be competitive pressure to use quite a lot of compute power to mine currency. The equilibrium state can only be that a large fraction of the world's compute power is going towards mining currency. That's quite a lot of dead-weight resource usage. I would expect more compute power go into making currency than to every other usage combined. Why not?

There are additional challenges as well. I don't know the details of bitcoin, so I'll speak in general.

First, the total amount of compute cycles put into currency mining is going to be steadily increasing over time. It is hard to see how to make a stable source of value out of such a base. Bitcoin sounds like it uses some sort of trusted third party to generate a grand divisor for the world to use from week to week. However, such an approach appears to concede the original goal of a distributed ecash. What is to stop the centralized divisor calculator from simply lying?

Another oddity is that different devices run the computations at different rates. One result of this is that you would almost certainly not use a cell phone to compute new currency. Indeed, you would probably not use any personal device at all, because the people who will win this game are the ones with specialized hardware. Thus almost everyone will end up buying the money from someone else, in stark contrast to the idea many people have that the device itself can generate the money it needs.

Ecash is great, but lets plan on having some sort of mint to administer the transactions.

Douglas Knight said...

To elaborate on Jon Leonard's comment, because the number of bitcoins mined in a given period is fixed ahead of time, improvements in computing do not effect the value of bitcoins.

Raphfrk said...

"Nothing is forcing people to value bitcoins."

You may not be forced per se, but for certain types of transactions, you may want to use a system like bitcoin (for example micropayments).

This creates demand and this is where the value comes from.

The real issue is the possibility of hyper transfer. For example, if a currency came out that was better in some way than bitcoin, everyone might switch instantly to it.

The bitcoin system is based on "proof of work". You can think of the chain as a mathematical object that they are constantly adding to.

In order to create another object of equal complexity, you would need to do the same amount of work that has been done so far on the bitcoin chain.

The reason gold is valuable is that it has certain properties and also has history.

The Bitcoin system has the property "virtual currency with the largest total proof of work" and also "oldest virtual currency".

Bitcoins are similar to diamonds compared to graphite. Both are made from carbon, but diamonds are (provably) crystals.

People would be cautious before switching to a currency that has only existed for a month, from bitcoins. Bitcoin's lead will increase over time. It will get older and the proof of work will increase.

The largest proof of work "title" is unlikely to be lost instantly. If the value of bitcoins slowly dropped, it would still be worth using as a currency and depending on how fast it drops, wouldn't necessarily give hyper transfer.

The way they have set it up, there will only ever be at most 21 million bitcoins, so it is more likely to experience deflation.

Coins are created on a schedule. Blocks are created around once every 10 minutes. The miner who manages it gets some bit coins, however that drops over time. Eventually, the miners would get almost nothing and would be funded by transactional fees.

The currency will experience inflation is the rate of coin creation is less than the increase in demand.

One issue with the system is that transactions don't take effect until the next block is created.

If you submit a transaction, miners will include your transaction when trying to calculate the next block.

10 minutes later, on average, one of them will get a "strike" and announce a new block containing your transaction.

It is possible that another miner who didn't include your transaction will also strike around the same time. This creates 2 provisional "best" blocks and they break the tie with whichever one gets the next valid block.

This means that transactions take a while to actually be accepted by the system. It is recommended that you don't consider the transaction validated until 6 blocks have passed. This gives 1 hour.

Breno said...

Mining is a terrible word to describe the process of getting a bitcoin.

Lotery is a better word to describe.

Everytime some lucky guy solves a block he won a few bitcoins.

Diferent from gold where you can get more gold by using more resource. You can't get more coins by using more energy or cpu time.

There is no cost of procution, bitcoin is FIAT money.

The diference from regular FIAT money is that bitcoin users trust the algorithm.

Is more like what PGP did for privacy right and freedom off expression. Bitcoin did to money.

KineticReaction said...

"but vulnerable to changes from external causes such as reductions in the cost of computing power or mathematical improvements that make it easier to deduce the information used to create one."

Changes in the cost of computer power or electricity would make no difference to how many bitcoins are created.

The rate at which bitcoins are created is fixed mathematically, and the difficulty in creating them adjusts automatically to keep production at a constant rate.

Value increases in bitcoin will result in more resources being used to mine them which will increase difficulty, but the direction of causation is one way. A decrease or increase in the resources being used to mine them has no effect on the rate at which bitcoins are produced and therefore the value of bitcoins.

Anonymous said...

"When you mine bitcoins by expending computer time &electricity, you end up with a number which has no inherent value for any other use. It's a complete loss - the resources have been wasted."

The Bitcoin network uses computational power as fuel. If nobody is pumping electricity through computers for the network, it won't work. The 50 bitcoin reward for mining is an incentive to get people to provide that power.

The electricity isn't wasted. It's used to create something of value - a functional exchange network.

Andrew S said...

The marginal cost of compute power to bad actors is near zero. Spammer botnets control 10-20% of the world's computers. Currently they spend most of their time on network limited work (spam) and thus have their GPUs and CPUs available for this sort of computing.

With resources that match existing bitcoin mining, you can create fraudulent transactions.

said...

A lot of ignorance in this thread, please check out this:

New to BitCoin? Start here!
https://bitcointalk.org/index.php?topic=7269.0