Proof of Work
In a way Bitcoin is not the first Proof of Work monetary system that humanity used. Arguably until recently all money were based on ‘wasted’ work, like for example shell money systems or Rai stones.
Haliotis shells “were traded all down the West Coast from Alaska to Mexico” (Quiggin p299) But among the Yurok, the haliotis shell was common enough in the local environment that it was only used whole, as a pendant with minor treasure value. Olivella was also locally abundant and used liberally as an ornament, but not as substantial money or treasure. With both fungible money and non-fungible treasure, we again see the signature economics of collectibles at work: the unique interplay between supply and demand, in particular the demand for scarce supply, which distinguishes a collectible from a normal commodity.
from Conflict and collectibles among the Yurok
Precious metals also form such systems — they are pricey because producing them is costly. If we somehow find a cheap way to produce/extract gold, discover a big deposit of gold or make transmutation cheap, the price of gold would go down. Currently the annual revenue of all gold mining is about 2800 x 32000 x 1200$ ≈ $100 billion (World_Gold_Production_1900–2014, gold price), and since gold mining is a competitive activity we can conclude that the annual cost of mining gold is also at that level. Using gold as some kind of reserve value store costs us at least $100 billion annually (plus storage costs). This seems wasteful — and so seem the shell money systems — but during the long history of humanity they worked — the value lost in production of the money was later recuperated in multiple transfers.
Bitcoin alone cannot be a money system for modern society — because it can only facilitate about 7 transactions per second. Americans make about 33billion credit card transactions annually — that means more than a thousand per second on average, and probably orders of magnitude more in peaks. I don’t know how many transactions are still made with physical money — maybe even more than with credit cards, and there are also wire transfers etc. Even if in bitcoin we introduce bigger blocks and do other optimizations the throughput of the current system would still be orders of magnitude more than what bitcoin can provide.
But perhaps it could play a role similar to gold. If not as the base currency — like in gold standard, then maybe as some kind of reserve (as gold in the current world economy). Or it can work as a settlement layer for a Lightning Network (its hard to say — but most probably this would mean for bitcoin a role not much different from gold in gold standard). In some ways bitcoin is better than gold — it is harder to counterfeit, easier to transfer big amounts, easier to hide etc, it will have a more stable supply and at some point in the future no supply at all — 0% inflation. In other ways it is worse — first of all it is much more fragile: gold can function even if it is no longer mined, in bitcoin mining is mandatory, the protocol might have bugs, it will not work in case of a total civilization collapse (when we lose our communication lines) or even in a more mild economic crash — when the demand for transactions using these reserves suddenly grows. This means that bitcoin would not replace gold — but it could have its smaller internet niche.
I can imagine bitcoin in that value reserve business in the future. Especially in the far away future (after block 6930000) when miners reward will come entirely from transaction fees. Then it will all be balanced out — the transaction initiators will not pay in fees more than the their subjective value of the transaction — the costs will be paid by the services. It will still be kind of wasteful with all this electricity burning. With gold we are in a similar place — we dig it from the ground, with so much effort, and then put it into vaults back underground. Maybe there is no better way. But with bitcoin it is probably worse. Transaction initiators will have to take part in an auction — so they’ll pay a fee that will be close to their value of the transaction, miners will earn that fee — but they also work in a competition so they’ll spent nearly as much money as the fee. In effect nearly all the value of transactions will be burned. With gold the transactions capacity channel is not limited and the transactions don’t need to pay fees close to their value to get into that channel.
In that far away future, when bitcoin becomes internet gold there will not be many bitcoin transactions (maybe still 7 per second), but they will transfer huge values and thus generate big value themselves. But I am convinced that currently Bitcoin and other cryptocurrencies as systems cost more than produce and it is a problem — because it cannot go on like this for very long.
It is not easy to estimate how much value people get from using Bitcoin and maybe I am missing something important — but my intuition is that it cannot be more than the extravagant cost of Bitcoin mining.
The best current ASIC miners have efficiency of about 0.1 Joule per 1⁰⁶ hash. The current hashrate of the bitcoin network is about 5*1⁰¹⁵ hashes per second. That means it uses at least 5*1⁰⁹ Joules of electricity every second, i.e. about 20Twh yearly. This is a conservative estimation — because most used miners have much worse efficiency — but it still places it at the level of Azerbaijan. And electricity is only a part of the cost — mining rigs, that become obsolete in a few months, also cost money and there is human work required to run the mining operations. These aspects are hard to estimate — but there is another way. In competitive markets MC=MR — so, just like with gold, a good estimate of the whole cost of maintaining the Bitcoin network would be some average of the total mining revenue — and that is approximately constant (for now — the mining reward will change about Jun 2020). There are about 1700 bitcoins mined every day— if we multiply this by the current price of bitcoin (lets take $2000) that means more than $3 millions per day or more than a billion dollars annually, plus transaction fees. This is again more than quite a few countries GDP . And this number will grow if we get more bitcoin adoption and the price rises.
Billion dollars a year is one hundredth of the price we pay for gold and it might be justified if bitcoin was already playing the role of digital gold. But it is far from that. It is still used mostly for short term speculation. Long term value store is also kind of speculation — you don’t buy gold to use it, but you buy it because you believe someone else will later buy it back from you. But with gold it is a much longer term speculation. It is not clear if it will not be replaced by some other cryptocurrency. The inflation will drop in the future — but now it is still rather high for long term value store — 4% annually (for comparison gold inflation is about 2.8/183 ≈ 1.5%). It is natural that the 8 years of bitcoin history cannot compare with the thousands of years of gold history. It is also self reinforcing — if people expect bitcoin to fluctuate then they will sell if they see its price dropping, adding to the amplitude. And the volatility makes it less suitable for long term storage.
Bitcoin is also used as a payment processor. As I argued above it is not a very good payment processor. Its throughput is dwarfed by other payment processors (like VISA with peak capacity over 50K tps). It is also not very usable — because of the need to wait for confirmations. But with some work can be quite anonymous — beating all other online money processors for transfers requiring that. It is probably quite important — but hard to estimate. The growing fees (inevitable because of the limited capacity) are shrinking this market.
That billion dollars a year is directly paid by the miners — but in the end they must be balanced out by new money flowing into the system — or the bitcoin price will fall. I believe that the extravagant price of bitcoin mining will soon drain the pockets of bitcoin enthusiasts and we’ll have bear market again for many many years.
This is why I have high hopes for Proof of Stake systems. They are much more complicated and we are far from full understanding of their game theory mechanics — but it is likely that they might be much cheaper to run.
Update: Fixed the calculation of price of a ton of gold — it is a hundred times more.
