Nothing is cheaper than proof of work (Part 1)

Preface: Wasted Resources

I wrote a post about Proof of Work (POW) and mining a while ago. The first half got a lot of attention but (obviously) wasn’t very convincing. For example, CCRG (Ethereum’s research group) used almost entirely academic theory about Proof of Stake (PoS).

So I'm going to expand on the first half of that article in the hopes that it will save some energy for those people (and the few journalists/hypocritical "investors" who associate the value of Bitcoin with internet money) who are wasting their time on these invalid ideas.

Outline

1. Reiterate the core theoretical concept (marginal benefit equals marginal cost).
2. Generalize all human activities by explaining economic rent.
3. Explain why rents greater than zero are incompatible with peer-to-peer currencies.
4. Give some specific examples (e.g. Tendermint, Delegated Proof of Stake DPoS) and assume they follow Bitcoin’s issuance schedule.
5. Summarize the debate about all release plans.

(Again) Auction: One hundred dollars

“$100, done! The winner is this miner wearing a Peercoin shirt.”

I used to think:

A: Issuing 50 new bitcoins per block... is equivalent to:

B: Sell these 50 bitcoins to the highest bidder at the auction

…This will continue until its marginal cost equals marginal revenue ( MC=MR ). After all, every hash has a cost, and every hash slightly increases the chances of each miner winning 50 bitcoins. If any action (not just hashing) can increase the chances of winning 50 bitcoins, then these actions will be repeated until the total cost of these actions is close to the total value of 50 bitcoins.

When we apply this to pure proof of stake (POS), this principle implies an attack phenomenon known as "stake oppression", while the proof-of-work version ("try multiple blockchain histories until you find the one that grants you the tokens") is significantly less incremental. This incremental work is not measured (that is, Bitcoin's "difficulty"), and it does not clearly show that "total amount of work" = "total expected value of block rewards".

For proof of work, you can think of it as a disagreement about which New York Times issue (out of 1,000) we should read tomorrow, while for proof of stake, it’s a question of which of the 6,000 issues are the real New York Times issues (out of 1,000).

The history of abandoned blockchains (the “ NYT堆栈”) is not widely known, so you don’t know about it, but their creation worked. As time goes by, more and more desperate developers push for such work, but the results are getting harder and harder to see: all the proposed proof-of-work alternatives have been re-labeled as “masked proof-of-work”.

We can see that changing the level of chaos (complexity) in the proof-of-work algorithm does not affect the connection between "how the world works" and "the impact of the blockchain". Ultimately, every individual "miner" (or anything else) will 1. estimate their own expected return, 2. deduct costs, and 3. work for the difference.

And "expected benefit" is the only variable parameter in this system (not "expected cost").

“Rent” always makes the production cost (MC) equal to the sales price (MR)

This concept of "marginal cost" = "marginal benefit" applies to almost every transaction in this world, including all future blockchain technologies.

Measuring costs

A sensible person might immediately argue that, in fact, MC = MR, how can such a thing exist? (For example: "I know that the convenience store near my home sells my beer at a higher price", "I run a small business, we usually raise the price by at least 3%, this 3% includes water, electricity, rent and my family's salary, I make a profit from this... Of course, I don't know how much profit I can make")

MC = MR is an eternal truth, because this is how economists measure costs. Specifically, by artificially introducing "rent", the equation can be refined as: MC_rent + MC_non-rent = MR.

This may sound like cheating (to an economics novice), so I will start with the overall discussion of rent, assuming that, for P2P money, the rent is always equal to zero (as opposed to some centralized system like a central bank/checkpoint that can be controlled or owned, forcing people to pay non-zero rent).

What is rent and what is the meaning of its existence?

As explained in economics textbooks

Suppose a seller has a patent on an item (only he can sell it). The seller can guess how the buyer will profit from the patent and estimate the size of their profit to bargain. If this asking price (MR) exceeds the cost (MC) he spent, the difference can be regarded as "rent" ( MR-MC ). Assume that this amount can be sold in a certain period of time, which determines the price of renting the patent (to someone). Although the seller has no cost to use his own patent, he will lose an opportunity cost if he does not rent it to others. Therefore, we have a new production cost (MC), which includes the patent that can be rented, which is now exactly equal to the sales price (MR).

This may seem impossible, but it is the only logical way to measure costs. Imagine that you spent $10 million on research and in the process discovered some drug that could be produced for $0.10 but sold for $200.00. Did the "cost" of that drug really only cost $0.10? Why is the "cost" different from what we measure, and is the difference in price a patent royalty, or a duplication of research?

Cough medicine

Let us try to take this concept to an extreme.

When I have a cough, the marginal revenue (i.e. benefit) of taking a dose of cough medicine is very high... I might have paid $50 a night for it. But I'm lucky: my total cost of medicine for a whole week is barely more than $5. So, MR>MC?

However, there are a lot of "rents" at play here, and this time they belong to me (the buyer), not the seller. I realize that my illness is private (only to me). I also realize that neighboring stores or competing products may adjust the price of medicine against me if all stores know about my urgent need for cough medicine and start preventing others from buying my medicine and preventing other stores from negotiating with me, etc. My response to these measures by the store is to stock up on medicine when I am healthy, have others buy it for me, or buy it at a competitor's store. If the store knows that I am willing to pay $200 for the medicine I need for 72 hours, then they will increase the price, but only I know at this time, so I don't have to pay an extra bill for it.

Although it looks like MR>MC, what actually happened is that I bought a course of "medicine" plus saw a specialist, which resulted in a good "cough cure". And when I get sick again, I will be able to do this (this medicine did me no good when I was healthy) because I already know that this medicine will cure my illness. Therefore, at this time, MC = medicine + sickness_rent = MR.

The rental is private

MC = MR applies only to decisions that are close to the agent and that the agent is involved in. The formula ignores external factors (because external factors can be problematic).

Today, you may be forced to “rent” something from someone else, and over time you will have to find alternatives (and you will pay a high price for not having economic sense at this point… you have to learn the “smarts of renting” and understand “what is the best deal” because you are not yet able to use it freely). In fact, if you are an entrepreneur, you can do the opposite: think about whether these rents can give people what they want, at a reasonable discount. This will motivate scarce resources to go where they are most needed.

at last

"Rent" means "exclusion", which is contrary to peer-to-peer (P2P) networks

If all peers are equal, how can anyone be forced to assume (endure) a subordinate position?

Rent only exists when the seller has something that its competitors do not, so the competitors need to pay "rent" to the seller. However, "exclusivism" and "peer-to-peer networks" are irreconcilable.

The problem does not only exist in the steady-state operation of the protocol, it exists in all areas, including the initial setup of the protocol. A complete peer-to-peer network protocol must have P2P operation and P2P origin. However, a protocol that allows significant rents raises questions: Who are the owners of these things we rent? Why them? Why do we rent these things?

The only entity on coinmarketcap that can earn rent is Ripple, and the reason for this is that it is not P2P (it is a licensed, inimitable “peer-to-peer” that has a lot to do with banking).

I hope this isn't too hard to take. Now that you know what we're talking about, let's get back to the point.

Everything is Proof of Work

In the absence of rent, we know from the issuance plan of "25 BTC / 10 minutes" that every ten minutes, 25 BTC will be lost for anything.

For simplicity, all P2P systems assumed in this section can issue at the same issuance schedule (i.e., 50 units every ten minutes, halving approximately every four and a half years). The following section explains why the issuance schedule is irrelevant.

Is a “working independently” agreement possible?

If all cryptosystems were to issue tokens periodically, without creating an incentive to “waste” those tokens of equal value, then the cryptosystem would have to issue them in a way that is independent of all possible human actions. The way tokens are awarded must have a Spearman correlation correlation of zero with anything that humans can influence.

This is not possible. The decision to use the software is within human control, and the protocol is invisible to non-users. Therefore, it is impossible to distribute money out of thin air... which makes so-called "waste" completely inevitable.

Changing time and ownership (no effect)

Imagine a design where 1. who gets the 50 BTC reward can be randomly determined in advance, 2. the result is kept secret and announced a long time later, and 3. the 50 BTC is paid out. At this point, when estimating MR, can no one calculate it?

Wrong.

opportunity

It is true that as time passes it will be too late to calculate MR accurately, but now is also the best time to calculate some future MRs, and it does not matter if MR and MC occur at different times.

(Did you know that Bitcoin’s proof-of-work rewards can’t be spent for 100 blocks? That’s because, if orphaned blocks are created, those coins are completely gone. Not only that, miners are most capable of double-spending, so it’s wise to keep an eye on the coins they do own.)

Randomness

Do you know how Bitcoin uses strong randomness to select the creator of the next block? Randomness prevents alliances between block creators, making double-spending attacks infeasible. The randomness is so strong that no one can predict who will discover the next block, or even when it will be discovered!

Make “work” useful

“We repurposed the PoW as a heater…”

First, repurposing is a good idea; second, humans need heat. Third, such an arrangement is highly conducive to achieving Bitcoin’s goals.

However, what does this do to MC and MR? Instead of the expected $100 (BTC)/x hashes, we chose $100 (BTC) + ($5 worth of heat)/x hashes. If the previous spend was $100 (on the hardware-efficient frontier), the spend will now increase upwards, to $105 (eventually, Bitcoin's difficulty will adjust). Inefficient miners (who don't use the wasted heat) will be eliminated, and the total spend will still equal $100 ($95 BTC + $5 worth of heat).

“Mining heat” simply makes the hardware more efficient, making them more difficult and complex at the technical level (energy usage/block).

“Proof of work is not just about double sha-256 calculations, it can also do something useful for society, like finding prime numbers…”

This logical argument is perfectly sound. Miners will not get windfalls from finding large prime numbers, so it will not increase their MR, nor their MC. However, from a social perspective, we will get additional benefits, but "we" do not need to increase the total cost. This is the argument for externalities mentioned above.