Proof of Stake is bad

for decentralisation.

Given that Web3 is all about user ownership, which cannot happen truly without decentralisation, it’s a logical assumption to say that the most important factor when deciding whether something is “good” or “bad” in Web3 is its degree of decentralisation. Now, I have a question…

Do we define something by its qualities, or by its outcome?

If you’re a proponent of equality of opportunity, you would tend to agree that as long as a system equally fosters opportunity, then there is equality in the system. However, if you’re a proponent of equality of outcome then you would not see the system as having equality (given that outcome will likely not be equal).

When it comes to decentralisation, is it fair to say that a system isn’t decentralised just because validation power has consolidated in the hands of the few?

I would argue it depends on the context of the conversation. If we are discussing “the most decentralised Layer 1 right now” then it is fair to look at the current outcome of the distribution of validation power to create some rating hierarchy.

However, if we are discussing “the consensus mechanism that most fosters decentralisation” then I believe we should look at the opportunity to be decentralised. Hence, Proof of Stake is bad: there is fundamentally less opportunity for decentralisation.

Proof of Stake’s fundamental issues with decentralisation

1. Proof of Stake starts off centralised

First and foremost, given that PoS requires validators to stake coins in order to validate, these coins need to exist before the chain can even go live. Sure, the team and investors that own these initial coins can sell them, or if they’re generous give them away, to the users, but that doesn’t stop a PoS chain from being completely centralised from the very beginning.

In PoW, however, the coins spawn in due to mining, thereby avoiding centralisation from the start. This leads perfectly onto the second key point.

2. Proof of Stake’s detrimental feedback loop

Secondly, PoS networks are secured via their own product - the coin. Once someone reaches 51% of the total coin supply, they cannot be dethroned since the consensus mechanism favours those with more coins. If this entity restakes its earnings, it will keep increasing its share of the total coin supply until it reaches 100% (as per the power-law distribution).

PoW on the other hand relies on hardware that is external to the system. If someone does acquire 51% hash power and keeps reinvesting their earnings into more mining equipment, a new entity can still come in with GPUs to bring down the attacker’s majority share. In PoS that literally cannot happen.

3. Unlikely abdication

Thirdly, in PoS there is no reason to sell the rewarded coins because validators don't expend more electricity than what it takes for the computer to simply be on.

Whilst PoW miners need to constantly manage operational expenses, the PoS validators can turn on auto-restake and forever keep their throne.

Proof of Stake defence

In PoS’ defence, it does a good job of disincentivising malicious behaviour by introducing a sunk cost - collateral; something PoW doesn’t have.

If a bad actor wanted to alter transactions or disrupt the network, said collateral will get slashed. If they wanted to shut down the L1 entirely, they will face an even larger monetary cost given the collateral (slashed or not) becomes completely worthless if they win.

On the other hand, in PoW, the bad actor simply loses out on the reward and faces cost of energy. After shutting down the L1, they can also simply re-utilize / sell their mining hardware, meaning the sunk costs are reversible.

This disincentivisation is enough to keep most attackers at bay and is sufficient to get the L1 to a market size whereby even bigger institutional attackers have to think twice about losing the collateral they would need to take it down. Combined with its speed and other benefits, PoS is still a solid option.

That being said…

Key thoughts

It seems that PoW is fundamentally better at decentralisation, but PoS is fundamentally better at disincentivising malicious behaviour. But the whole idea of decentralisation is to disempower bad actors by letting everyone and anyone be a part of the validation process.

Decentralisation is a way to bypass tyranny of those in control by giving control to everyone instead of a select few. Whilst PoW seems to facilitate this by segregating the product of the system (transacting value) from its source of power (mining), PoS seems to simply impose monetary fines for tyrannical behaviour of those in control.

If PoW’s problem was that it is too slow to use on a mass scale, then we must call a spade a spade and admit that we will sacrifice the principles of the Web3 paradigm for accessibility.

But does PoS need to be good at decentralisation? No. PoS L1s, despite prioritising speed, are still the foundation for decentralised protocols and therefore incredibly crucial to the mantra of Web3. The issue is that all protocols and dApps are secured by the L1s, most of which are fairly centralised.

But do we really care about decentralisation? Evidently, not on the infrastructure level. Beyond the L1s, the internet service providers, cloud storage hosts, microchip manufacturers, and so on, are also all centralised. How deep should you go?

It seems we just care about owning our stuff. But what does that even mean when we’re discussing the stuff on the World Wide Web?