Bitcoin is An Energy Hog
And its value seems tied to that fact
If you’re not familiar with how bitcoin works and why miners need to exist, you’re in luck. I’ve got a quick and simple overview for you. This piece used to be a small part of the article, but I’ve rewritten most of it and expanded it.
The cooperation and competition between cooperating groups for mining bitcoin has resulted in a lot of wasted energy. The power consumption of the network is still growing. As the network grows, so do the power requirements. On the day of December 22, 2017 Bitcoin mining :
- Totaled 99,502,553 kWh for the day.
- Consumed up 690,990 kWh for each block.
- Expended 57,582 kWh for each Bitcoin.
This one day of use required enough energy to power 82 Back to the Future DeLorean time machines — if they actually existed.
These values are trending up each day as more people join the network . The cost of this much energy varies based on your location . The current cost of producing one Bitcoin on residential power in NYC is $13,365. If one uses commercial power, it drops down to $6,650. The price of one Bitcoin on that day closed at $15,695.79.
Is there a correlation between power consumption and the cost of bitcoin. There definitely seems to be. At first, my analysis used the NYC cost of electricity, but severely overestimated the value of Bitcoin during the last year and a half. Using the residential cost of electricity in the US and China averaged over 2014–2017, we see an interesting correlation :
The price of bitcoin from around 2015 until mid-October 2017, tracks with the price of electricity in China. From August and October, it correlates with the average US cost of energy. In November, it goes on a space walk to a new energy level. By the start of December, it moves with the average cost of electricity in NYC. Here’s the detail view of the last 6 months:
Correlation does not equal causation, but what a correlation! Bitcoin appears to be a store for the expended value of electricity in areas with a lot of financial interest in Bitcoin. If this correlation is strong enough, it means the Bitcoin will always be bounded by the cost and availability of electricity. It also means making a lot of money off Bitcoin can only happen through investing in mining infrastructure in regions where electricity is super cheap. Of course, one could always manufacture a hype and artificially raise the value through modifying market sentiment.
Ch — Ch — Ch — Ch — Changes!
The next reward halving in 2020  could cause bitcoin to completely fall over if the value of bitcoin doesn’t grow in proportion to the cost of mining. If it all works out, there are decades left until all Bitcoin are mined and the network will have to run off transaction fees. Energy consumption is what I see as the biggest concern.
The next halving could cause a transient effect in the network where the number of machines mining on the network drops dramatically. This may have the effect of miners selling off their Bitcoin and the value falling. The difficulty value will drop, which will cause the amount of energy used to plummet. If the correlation above holds, that may actually cause the value of Bitcoin to plummet more. This is no big deal — long time Bitcoin holders know it has had its ups and downs.
Another possibility is the price of bitcoin doubling. This would allow all the miners to stay in the game, maybe with a slight dip before a large increase in the value. They wouldn’t have to drop out, but energy consumption would continue increasing around the same rate it is today. If the above Bitcoin value to average cost of energy correlation holds, it means the energy consumption could end up doubling.
The above correlation may also be spurious. In that case, it would be vary unclear if the value would drop. If the value is tied to the number of people using the network, it is possible the value of Bitcoin will stay the same. Cutting down the number of miners would decrease difficulty, meaning fewer hashes would have to be tried. Unfortunately, that means fewer people are involved in ensuring the network can be trusted.
One of the easiest, yet most hotly debated changes is increasing the block size. Increasing the block size has two effects: 1) It will allow more transactions to be processed per block. It will increase the number of transactions verified every 10 minutes and appended to the blockchain. 2) It allows more data to be stored in each transaction. Storing more data is contrary to the goal of processing more transactions, as data associated with each transaction limits how many transactions can be stored in a block.
There was a bit of a compromise made with this in the 0.13.0 version of Bitcoin Core. It basically stated that associated data is now counted as 1/4 the size it used to be. This effectively increases the number of transactions stored in a block and effectively increases the size of the block. Unfortunately, the added efficiency is not that much.
Another idea would be to move the witness data off the chain entirely. It would have to be shared over another mechanism. Perhaps a separate chain or just a plain old DHT. Then the protocol would function almost like it did before segwit, and all the transactions could be appended to the blockchain as fast as they would fit. Of course, this does add additional headaches that many people would probably rather not deal with.
The next idea to be considered is changing the effort required to try solving the PoW. This is actually what Bitcoin Gold(BTG) did when they forked Bitcoin. They made it easier to use commodity GPU hardware to mine and verify trust in a distributed fashion. The value of BTG is currently around $325. There is no data on energy consumption, but last I checked, they had 331 nodes. No one knows how it would have played out if Bitcoin actually changed their PoW scheme. Would the added trust have added value to the network? Who knows? It will be interesting to see how BTG changes over time.
Other blockchain tokens use other schemes besides proof of work. One token, called Chia, is looking at using proof of pace. At one point, there was mention of verifiable delay functions, but their website doesn’t mention it any longer. Some use proof of stake — which basically assumes someone with a large stake will not act against it’s own interest. In Hashgraph, each node talks about what it knows about and what it has sent to other nodes until all nodes can prove every node knows everything they know — like a proof of consensus.
We will likely see a lot of innovation in this realm. It will be very interesting to see which scheme wins in ten years. Perhaps we will continue living in a mosaic of different blockchain based tokens. Perhaps something new will take Bitcoin’s place as the dominant coin in the market. Maybe Visa will just continue being the default network.
- To compute cost of creating one Bitcoin, I established a correlation between the number of TH/day and kWh/day as value conv in TH/kWh. To compute the value use: 60*60*24*TH_s*conv*energy_cost/1600