In the last video, we have discussed that a critical factor for the survival of public blockchains like Bitcoin and many other cryptocurrencies is maintaining a healthy base of nodes. It's this simple. No nodes, no network, no blockchain. So it makes sense to consider some simple economics on the decision of a miner to join a public blockchain and stay in the game, and look at some potential ways that we could tweak the parameters to make the system more appealing to the nodes and more efficient for the users. If you're thinking about launching your own coin or token, perhaps to compete with Bitcoin and the others, this should be something of a first-order concern. If you think about it, the decision of a Bitcoin node to participate is not that complicated. A rational agent will participate if there's an expected profit to be made. Specifically, a miner will participate if it's expected revenue is greater than the expected cost. The revenue comes from the new coins mined and the transaction fees received, and how likely a miner will be getting those depending on several key factors. How many other miners are there? The more miners there are, the more difficult the puzzle is, and the more difficult it is to win the new coins. On the other side, how big is the demand? If there are a lot of users competing for the limited block space to send their transactions and the system is very congested, the transaction fees will go up. Also, because all of these rewards are paid in Bitcoins, you're most likely convert it back to fiat money to spend. What price do you expect the Bitcoin to have in the future? If you think the price is going to the moon, you're more likely to take the chance and participate. If you think the price is going to zero, you'll be much less likely to participate or stay. So a miner will take an expectation of all that revenue factors and balance it with the costs on the other side of the equation. Unlike expected revenue, the participation costs are much simpler. Like any business endeavor, there is fixed cost and a variable cost. The fixed cost is your hardware computing power, the computers and mining chips that a miner needs to buy, and the cost of real estate to host these hardware. The variable cost is predominantly energy. Guess what? Doing all these trillions of hashes takes a lot of energy and that could be quite costly. So let's take a look at each of these cost components separately and derive some simple implications. First, let's look at the hardware. If you're interested in crypto, at this point, you might be thinking about getting into the mining game with your desktop or laptop computer. Don't. Even the fastest CPU can only run a couple of thousands of hashes per second compared to the trillions and trillions of the entire pool. So it's basically useless at this point. Same thing with the graphic cars. Some people set up a home rack with four or five cars to my only bitcoin. That's mostly a waste of hardware and energy because although GPUs get a bit higher hash rate, it's still very tiny. The only way to mine Bitcoin these days is using dedicated chips. Application Specific Integrated Circuit or ASIC chips that are designed to do one thing and one thing only take a lot of hashes. Unlike CPU or GPU which have to do other tasks like running your computer, these dedicated chips devote their entire computing power to carry out that single function. So you can make it very small and state thousand of them in mining machines. This is by far the most common way that Bitcoin is mined. Moreover, the design of the proof-of-work puzzle ironically encourages a significant degree of centralization. Think about it. If your probability of getting selected to post the block is proportional to your computing power, it makes sense to band many small nodes together as one large unit. Consequently, most of the Bitcoin and other cryptocurrency mining is done in mining centers or mining pools, both are businesses that's run up in the wake of the crypto boom to take advantage of the economies of scale. First, a mining center is like a factory. This is probably what you saw in the news in the last couple of years. The business model is very simple. Buy up as many mining chips as you can, house them in a location where electricity is cheap and hopefully with a reasonable cool temperature like near a hydroelectric dam or ice land, then fire away. A mining pool is exactly like what it sounds. A pool of miners banded together as one unit. Most individual miners join mining pools. This is also operated as a business, where there's a manager node that bands individual miners together and listen to the transactions on their behalf. The manager node builds the blocks then delegates the hashing duties to individual miners in the pool. This effectively achieves the same power concentration as a mining center, except the individual nodes could be located all over the world. If the pool wins the reward, it's split among the individual miners according to how much hash power they have contributed to the pool. Of course, it's a business and the business model is simply charging a management fee, a percentage cut of the reward that the pool has received. If you think about it, both mining models actually run against the original idea of decentralized consensus and further illustrates a critical limitation of proof-of-work. It is an extremely inefficient form of consensus building and nothing illustrates this further than as energy usage. The annual energy used to run these trillions and trillions of hashes and to keep these chips cool is a staggering amount of more than 70 trillion watt hours. This is more than the annual energy usage of over 150 countries in the world. To put it in another way, it takes about 600 kilowatt hours to process a single Bitcoin transaction, and that one transaction alone can power a typical US household for three weeks. By comparison, the visa network takes about one-third of that, 200 kilowatt hours, to process one million transactions. So on a transaction by transaction basis, visa is over three million times more efficient than Bitcoin. Therefore, in order for the growth to be sustainable, it's imperative for the blockchain developers to come up with a new generation of decentralized consensus algorithms that are just as good at reconciling the data but significantly more efficient. Some of the examples are also called practical Byzantine fault tolerance algorithms and include protocols such as proof-of-stake. At the current stage, however, many of these are still work in progress. So while we expect to see many exciting breakthroughs in the future, currently, most public blockchains still use proof-of-work. So if you put them together, the economic implication is that we are more or less currently at a market equilibrium in terms of the Bitcoin network. That is initially the network is very small with very few miners for a long time. The mining puzzles are not too difficult and the reward was pretty large. The initial barriers of entry, therefore, was quite low. As Bitcoin and blockchain in general began to receive media and public attention, more miners joined the system, driving up the difficulty of the puzzles significantly. This combined with the inherent inefficiency of the proof-of-work algorithm resulted in a significant degree of consolidation in the node network with most of the mining being done by either big mining centers or big mining pools. So the network effectively becomes much more centralized and they expect the profits are probably quite close to the expected costs. Therefore, the barriers of entry for Bitcoin mining now is likely much, much higher and the market is significantly more competitive. Consequently just like gold-mining throughout history, the bulk of the economic rent is probably earn not by the miners themselves, but by makers of the mining equipment and by the operators of the mining pools. The economics of this is fascinating, and if you're interested in having a deeper dive into this aspect, I've included some useful resources here.
