Hacker News Reader

> then there is simply no way for anyone to send $99 to try to get at that randomnes

can't I send one of the 10 people $99 today, and he pays me out in 10 days? Can't that person rent the fact he/she is on the list of 10 out. I am new to economic terms, but it seems your example even better shows his MC_rent + MC_nonrent = MR. Where he says in p2p systems MC_rent should be 0, because it's exclusionary. Again .. your example's main feature is that it excludes all but 10 people.

> can't I send one of the 10 people $99 today, and he pays me out in 10 days?

But if he already received the $100, why would he trade his $100 for your $99?

Are you saying the recipient might want $99 now rather than $100 later? Let's say that the donor doesn't give any advanced notice; the recipient doesn't know they've been selected until they see $100 appear in their bank account.

The example given is a bit fuzzy about what it is showing and what assumptions are being made. This example is pretty bizarre because we have actors behaving economically irrationally (the $100 provider) and beneficiaries (I'll call them listees) who don't seem to have volunteered. I think it might be meant to simulate the creation of new money, but I don't accept the situation as being similar enough to be sensible (because no money is being created).

Mr-give-away-$100 is burning resources (his cash reserves) to create an asset for 10 people (the potential to be given money randomly in the future, which is a strange financial asset). That asset is absolutely saleable on a secondary market and will be priced at some value south of $100.

We can deduce who is on the list by observing who gets the $100 payments and deduce the probability after a few days of observation. The listees have incentive to provide this information as asset + secondary market is better for them than asset + no secondary market.

Upper bound: Nobody is going to pay >$100 for a promise of $100 (I lie, that does happen, but only in rare and extreme circumstances. We'll ignore those).

Lower bound: If the listed individual needs money NOW for some reason, which does happen in practice, then they may be willing to sell at any price >$0.

Mechanism: A contract. They work well for events that happen in the future.

Motive: The listees are giving up a small amount of value to smooth out their income stream.

Results: When the $100 is transferred out of Mr-give-away-$100's account it will immediately be transferred to someone who bought it for some value $100 > X > $0; at a guess probably in the high $90 range.

The person who was gifted a free asset will get rich, but that can be realised before the actual payment is made.

Such contracts would not really add any net costs to the system, since they cost virtually no resources. Whereas if I send $100 to the winner of a PoW contest, then $99 worth of electricity will be expended.

Even if the contracts did add some costs to the system, PoW systems can have similar contracts. I could join a PoW pool and agree to share any $100 rewards I receive with the pool.

Alternatively, we could just avoid the matter of contracts by replacing $100 with $1. $1 is small enough that nobody would need to think about diminishing returns, so there would be little point in spreading out the risk.

> there is simply no way for anyone to send $99 to try to get at that randomness.

Sure there is: black bag your house and bug your laptop, or remotely compromise it.

If the original article's argument is basically correct then it means that for any alternative to PoW such as PoS compute-grinding work may be replaced by attempts to game the system or attack the system's security.

I'm also skeptical of the claim that PoW is necessarily more expensive than fiat currency issuance. To account for the cost of fiat you must include the cost of all the security, administration, and governance that is required to enforce the integrity of the system. A dollar costs a lot more than the paper it's printed on (or numbers in a database).

The Ethereum wiki is wrong on this point in my opinion.

Even disregarding grinding attacks, the value diverted to coin purchases ultimately has a cost in goods and services. It diverts economic activity to non-economically productive activity in cycling capital into and out of deposits.

I think Proof of Stake could potentially be better than Proof of Work, but the point about cost being equal across validation methods is correct in general in my opinion. There are specific circumstances where it is not true, like if producing the mining resource creates negative externalities.

Where I think the article is wrong is in neglecting other aspects of consensus algorithm efficacy, like the potential security benefits from Proof of Stake totally aligning the incentives of owners of mining capital (which in the case of PoS is the network coins) with the success of the network.

> the value diverted to coin purchases ultimately has a cost in goods and services

I see what you mean, but that opportunity cost applies to all currencies. Every dollar we hold in currency (be it fiat, PoW or PoS) could have been invested in productive businesses. People tend hold substantial amounts of currency anyway since it's convenient.

If we assume that people are going to hold about $10 trillion in currencies anyway, then it makes sense to ignore that opportunity cost when comparing PoS to PoW.

> Proof of Stake totally aligning the incentives of owners of mining capital (which in the case of PoS is the network coins) with the success of the network.

I agree -- aligning validators' and stakeholders' incentives has some nice benefits. It means we can have validators vote on parameters like minimum transaction fees, and the outcome of the vote should reflect what's best for stakeholders as a whole. In a PoW system that wouldn't work well -- the miners would likely vote for minimum fees that were higher than required for secure validation.

>>I see what you mean, but that opportunity cost applies to all currencies.

I don't mean that. I mean the actual process of converting liquid wealth into/out-of ether staking deposits, and of sacrificing liquidity to maintain stake deposits. That is not a free activity.

Given the earnings accrued to anyone with staking deposits, there will be competition to increase the share of one's wealth that are held as staking deposits, and that requires increasing frequency of these activities.

The cost of all of these activities will approach the revenue generated by holding ether.

Absolutely there's an economic cost; that's sort of the whole point, so it's expensive to attack. However:

- The economic cost doesn't have to be an energy cost, which means it can be more environmentally friendly than Bitcoin.

- With PoS you have the option of making attacks expensive by penalizing the attacker, which lets you achieve a given level of security at lower economic cost to honest actors.

This is actually addressed in the article.

See section: IS A “WORK-INDEPENDENT” PROTOCOL POSSIBLE?

> If any cryptosystem is to periodically release coins, without immediately creating an incentive to “waste” an amount equal to the value of those coins, the cryptosystem is going to have to release the coins in a manner which is totally independent of all possible human activities. The coins will have to be rewarded on a completely effort-blind basis. The coin-reward must have a Spearman correlation of zero with everything that mankind could influence.

If you can make it so that there is no way for human influence to affect who gets the coins, then of course marginal revenue will not necessary approach margincal cost. However, it's quite hard to design a system that doesn't have at least some way to influence who receives the rewards. And once you have some way of paying to modify the rewards, then marginal cost start creeping closer to marginal revenue ...

You're right, I missed that, but

> The coins will have to be rewarded on a completely effort-blind basis

He's right that an effort-blind system is impossible, but that doesn't matter as long as the expected costs of influencing the block reward exceed the expected benefits. And it is is fact possible to design a system where the expected costs will always exceed the expected benefits, so putting effort toward influencing the block reward is unprofitable.

The profitability of a PoS grinding attack depends on the transaction fee, the minimum account age, and the certainty with future blockchain states can be predicted. In a well-designed PoS system, those variables will be chosen such that grinding attacks have negative expected value.

> And it is is fact possible to design a system where the expected costs will always outweigh the expected benefits,

This is the key point, and also why the author's analysis is much too broad: as stated, the argument applies to all social structures in general.

For example, under the author's analysis, PoW itself should not be able to exist, because if $10 of economic value is generated by the blockchain, then at least $10 of energy needs to be burned in order to keep the blockchain secure. This is wrong; it ignores the high real-world cost of dispersed attackers to acquire the (dispersed) surpluses from an attack.

Yet we see that in practice social structures exist and produce net surpluses greater than their maintenance cost (in fact, these two facts are equivalent, since a social structure that did not provide an energetic surplus would not survive in dynamic equilibrium).

The ability to punish rule-breakers as a large multiple of the magnitude of their attempted rule-breaking is what allows social structures (including markets) to gain the necessary leverage; otherwise there can be no surplus derived from creating structure, because to get $1 of social surplus, you need to burn $1 of surplus.

Therefore, the real question is whether dPoS creates structures with higher leverage to punish cheaters than PoW.

One other thing to keep in mind is that cost also includes the liquidity cost of staking coins. You are literally printing more lottery tickets for every additional amount of coin you stake. And liquidity cost is a real cost as you could have instead invested that money into a money making bushiness. So marginal cost can also approach marginal revenue in terms due to liquidity costs as well (grinding attacks aren't necessary).

That's true, but PoW currencies have the same opportunity cost -- if you buy Bitcoin you can't invest the same funds else.

Let's say you were forced to invest all your savings in either a PoW or a PoS currency; no other investments are possible. Say they both have the same block reward, which is essentially inflation tax.

Since you'll have to pay inflation tax either way, wouldn't you rather invest in the PoS currency, where the tax is redistributed to stakeholders including yourself? Rather than the PoW currency where the tax goes toward hardware and electricity bills.

There's a third possibility in your example though: Invest in mining the PoW currency.

> That's true, but PoW currencies have the same opportunity cost -- if you buy Bitcoin you can't invest the same funds else.

That's simply not true. You could potentially loan out your Bitcoin to a company which would play to use that money to create a profitable business. (Of course, it's a bit trickier to do that with Bitcoin due to the deflationary nature, but that's another story.)

When you stake your coins, that's not quite possible.

True but at the end of the day, someone is going to be holding the coins, and that entity is better off holding PoS coins that PoW coins.

We can tweak the scenario to say that you're forced to buy and simply hold one of the currencies, without loaning them out or otherwise touching them for 5 years. Then clearly PoS has the advantage.

The quoted text does not support your thesis - come on! It admits that C=R might be true - but argues that this is not enough.

I see how you could read it that way based on the first half-sentence, but if you read the rest, it's saying that C=R is not always true.