⚠️ Validators Up 85%, Nominators Down 70% staking rewards

[Megadot]

Maybe one of the ecosystem’s brilliant minds like @jonas can explain this to me.

Before the DAP staking changes, validators were earning around 23 DOT per day. Now, with 10K DOT self-bonded, validators are earning 43 DOT per day, an increase of almost 85%. Yes, 85% more.

At the same time, nominator’s rewards have dropped by around 70%. Yes, 70% less.

How is it possible that validators are making almost twice as much while nominators are earning dramatically less? How is this a better staking model?

What’s the point of staking as a nominator if the economics are shifting so heavily in favor of validators?

This ecosystem makes absolutely no sense.

[CryptoJoe]

As a nominator if I had invested in the SP500 instead of DOT my yearly yield would be 14-16% while here on DOT it is 6%, having this much risk with this so low yield doesn’t compensate for holding this token.

As a nominator it is not worth it to hold DOT since it yields half of the SP500 with more than twice the risk, seems like a bad deal and if you account for inflation, real yield for nominators is not 6% but 3%.

There needs to be real use case for the token and not just staking.

[Megadot]

I think I’ve finally found the flaw in Jonas’ calculations.

The entire model appears to assume that running a validator costs $2,000 per month in server infrastructure.

Here you can see the official hardware requirements for running a validator: 32 GB of RAM and 2 TB of SSD storage.

These requirements are nowhere near what would justify a $2,000/month infrastructure cost for the vast majority of operators. Can someone explain where that number came from? Has nobody at Parity or the W3F pointed out that this simply isn’t representative of what most validators actually pay?

Maybe $2,000/month makes sense if you’re running infrastructure in a very expensive region like APAC with unusually high bandwidth costs, but that’s certainly not the norm.

Here are several hosting providers and their monthly pricing (some with unmetered traffic), compiled by @kukabi in this post

A large number of validators run their infrastructure in Europe using providers such as OVH, Interserver, Vultr, and others, where comparable servers typically cost around $200/month, not $2,000. Here’s an example of OVH’s pricing:

If the economic model used to redesign staking is based on infrastructure costs that are off by an order of magnitude, then it’s no surprise we’re seeing the outcome we have today: validators have seen a significant increase in rewards by 85%, while nominators have suffered a 70% dramatic reduction in their rewards.

I genuinely hope someone steps in, corrects this broken economic model, and remembers that nominators are supposed to be stakeholders too, not just an afterthought.

@OliverTY @thewhiterabbitM @bill_w3f @sorpaas @kianenigma @jonas

[Cyrill]

My 2 cents. Why I think the $2000/months is in reality not as bad as you frame it:

1. You can’t just look at some random monthly server costs. Need to think about redundancy, monitoring, security, disaster recovery. So on top of the monthly server rental, you need to factor in costs for more infra, additional services, and some human time that needs to be compensated.
2. You said it yourself: maybe there are actually regions that are much more expensive. The $2000 dollars a month needs to work out for everyone, if we want to have geographical distribution.

I’m not trying to argue with the overall promise of the post, thanks for raising it, I think it’s a valid concern and maybe this number can be fine-tuned to bring down costs further.

[Megadot]

These statistics may not be fully up to date, but I highly doubt that more than 15% of validators are running in Asia.

Around 80% of the network is hosted in Europe and US, where infrastructure costs are significantly lower and simply don’t match the $2,000/month assumption used in the calculations.

If the majority of validators are operating in regions with much lower hosting costs, then the economic model should reflect that reality, not the costs of a small minority of deployments.

[Nukeme3]

As I understand it, the costs you’ve outlined are for the cloud service rents, just the cost of borrowing the necessary hardware to run a validator, which simplifies the hardware maintenance costs but does not account for the software maintenance, security, and additional administrative and support costs.

Presumably, this could be done by a single person, but it’s more likely served as a team of 2-3 domain experts who share administration duties across a small cluster of validators.

Running a validator node is not a passive turn it on and walk away activity like Bitcoin mining. $2000 month is not an unreasonable cost estimate for the skilled labor in the EU/NA

[Megadot]

Running a validator requires the same type of infrastructure setup as for running a collator: configure the server, setting up firewall rules, installing monitoring with Grafana/Prometheus, downloading the binaries, and getting the node running. Even the initial chain synchronization is now much faster thanks to warp sync.

After the initial setup, the maintenance process is also very similar: monitoring the node and applying occasional binary updates, which usually take only a few minutes per month.

So why are collators considered to have operational costs around $300/month according to this last proposal, while validator costs are being modeled at much higher levels? If both require similar infrastructure, setup, and maintenance effort, can someone explain the reasoning behind such a large difference in the cost assumptions?

I get the impression that nobody has really stopped to calculate the actual operational costs based on real-world data.

We’ve seen this before with the IBP Bounty, where pricing calculations were based on assumptions around AWS costs, while many operators were actually running their infrastructure on much cheaper providers.

If the goal is to design fair economics, the model should be based on the costs that operators are actually paying, not on expensive cloud assumptions that don’t reflect the reality for most of the network.

[cpiccard]

The recent changes to staking and validation are yet another masterpiece from the genius behind the core-time sale auction mechanism. The only thing missing is a Dutch candle auction. I’d give it a solid 9.5/10.

[Nukeme3]

I understand the point you’re making. Validators and collators share similar work activities and failure modes.

But there’s a difference in that collators are an operational cost to the parachain, whereas validators are a cost to the entirety of the Polkadot network. The risk of collator failure is a loss of block output on a single parachain. It may also be an exploitable failure impacting users of the parachain.

But loss of a validator node, or worse compromise, places the entirety of the network at risk. Hence, the importance and effort made to harden the NPoS over the last 10 years.

It doesn’t come down to raw costs, but instead to risk mitigation.

To give a personal example, I work as a controls engineer at an auto manufacturer. Much of my time is spent monitoring systems, making small changes when new component part numbers are released, generating some reports, rebooting computers that IT buggered up, and answering questions about machines that people could find in the manuals.

99% of the time, my work could be handled by just about anyone with a Google search tab open. Except the 1% when SHTF and the entire assembly plant grinds to a stop because something, somewhere, decided to stop working. Now, suddenly, production managers, maintenance managers, manufacturing managers, quality managers, and their entire teams are dead in the water because no one knows what went wrong, let alone how to fix it.

My career is based on learning and understanding complex, critical systems, mitigating failures, and recovering to standard operation when they inevitably fail.

The price premium of well-selected validators is based less on the tangible, “what do they do?” and more on the, “what do they know?”

Granted, if anyone has updated, accurate validator total operating cost assessments, I’d be interested to see it.

[Megadot]

The issue is not whether validators deserve fair compensation. The issue is whether the DAP model accurately reflects the real costs, risks, and responsibilities involved.

Before these changes, validator compensation was shaped by market dynamics. When DOT was around $3.50, many validators were earning around 250 DOT/month (~$875/month), with different commission structures competing in an open market.

Instead of allowing the market to continue determining the cost of operating a validator, the new model introduced fixed cost assumptions with significantly above real-world expenses. Most validators run on providers such as OVH, Vultr, InterServer, and other European hosts, where the required infrastructure costs a fraction of the amounts used in the model. After the initial setup, monitoring, and occasional upgrades, validator operations are mainly about maintaining reliability.

The model also creates a clear inconsistency. Validators and collators run similar infrastructure, require similar expertise, and maintain similar software stacks. Yet collators are modeled at around $300/month, while validators are assigned costs closer to $2,000/month. The recent Asset Hub incident demonstrated this clearly: when a critical parachain stopped producing blocks, transfers were halted, XCM was affected, and applications depending on it were disrupted.

The result of the DAP has been an ~85% increase in validator rewards, with nominators paying the price through a roughly 70% reduction in returns, all based on cost assumptions that appear unrealistic and disconnected from actual operating conditions.

[OliverTY]

@Megadot

The result of the DAP has been an ~85% increase in validator rewards, with nominators paying the price through a roughly 70% reduction in returns

Nominators are not slashable anymore since the recent staking updates. Their stake therefore does not improve the security of Polkadot anymore. Surely this justifies lower APY.

Nominators are not “paying the price”. Nominators are just delivering less value to Polkadot which is why they get less rewards.

[Megadot]

If the justification for higher validator compensation is security and risk, then the model should measure security and risk, not introduce an arbitrary $2,000/month operating cost assumption as the basis for the calculation.

Security risk and operational expenses are different things. If validators are being compensated for risk, that risk should be quantified. If they are being compensated for costs, those costs should be based on transparent, real-world data.

Using inflated cost assumptions to justify a security premium creates a model where the outcome is effectively predetermined.

I understand the argument that removing nominator slashing changes the value proposition of being a nominator. However, that does not justify replacing a market-based system with fixed assumptions that appear disconnected from reality.

Before these changes, validator economics were shaped by competition: operators set commissions, nominators chose where to stake, and compensation was influenced by supply and demand.

The main concern is the scale of the change.

An ~85% increase in validator rewards combined with a ~70% reduction in nominator returns is not a minor adjustment. It represents a significant redistribution of rewards.

The issue is not whether validators should earn more. The issue is why the solution was to introduce fixed cost assumptions far above actual operating expenses.

If the real cost of running a validator is closer to $300–$600/month, then compensation should reflect that reality. It should not be based on much higher assumptions and justified by claiming that nominators provide less value.

Nominators may have a different role after the staking changes, but they still contribute to network participation, decentralization, and economic security. Their rewards should not be reduced based on a cost model that has not been properly validated.

[OliverTY]

@Megadot

I understand the argument that removing nominator slashing changes the value proposition of being a nominator. However, that does not justify replacing a market-based system with arbitrary cost assumptions.

I dont think that is the justification. It are two separate things.

I also agree with the fact that validator rewards should be set to a realistic value, whatever that may be.

[Megadot]

I agree that they are two separate things, and that is exactly my point.

If security/risk is the justification for higher validator compensation, then it should be evaluated separately from operational costs. If operational costs are the basis, then they should be based on realistic market data.

The concern is that the current model appears to combine assumptions that are difficult to separate: a security argument alongside a fixed operating cost assumption that seems far above what many operators actually pay.

I appreciate that you also agree validator rewards should be set at a realistic level. That is the core issue: the model should arrive at that value using transparent, verifiable data rather than assumptions that may unintentionally shift too much of the reward pool away from nominators.

[dandan]

Validator rewards have increased, and more value is being retained in the Treasury. Who are the main beneficiaries of this redistribution? Also, who currently operates the largest number of validator nodes?

[Nukeme3]

I think I understand you a little better now. The concern is more about running with an unverified assumption rather than actual accounting. Not simply that the hardware rental costs are a fraction of the payout. Do I understand the correctly?

It might be a bit too early for a knee-jerk reaction, especially as there’s discussion of reducing the active validator set, but assessing the real cost of running a validator and a possible parameter adjustment to distribution seems reasonable. Additionally, this information will be necessary anyways when validators are compensated with a stablecoin from the DAP.

[sorpaas]

See also DAP staking changes significantly reduced Polkadot's security -- a wrong premise, and a bet against the industry

But honestly, here we don’t even need any theory for this. Just look at who’s benefiting: W3F/Parity themselves single-handedly runs 100+ validators. Plugging in their calculations: $2000 * 100 = $200,000 per month. And they’re the entity who’s authoring, proposing, and voting for DAP.

By the way, for the theory part we’ll write a more detailed response. The problem with DAP actually goes quite deep and it’s a bet against the whole industry. Not just $2000/month (good argument, but we don’t even think that’s the central failure). There are actually two particular things we can observe over the next months to years (which we won’t discuss, as network security is at stake). Therefore we’ll be able to confirm whether we’re correct, and whether the $2.01M security floor will materialize. The recent BonkDAO incident means a realistic attack becomes possible (for them it costed $4M recoverable funds).

[Maxx]

Hah they removed my post, what else is new. Censorship on Polkadot Forums and I didn’t even say anything bad.

[Gr33nHatt3R]

@Maxx Your post was generally fine, except for the mention of investing. Please review the Code of Conduct. You’re welcome to repost your general complaints as long as the investing reference is removed.

No Investor / Price talk
No market/price/investor talk of any kind is allowed on the forums. We urge you to avoid terms which violate this, E.g “ICO”

[cpiccard]

Seems the non-profit / charitable W3F is trying to squeeze out as much as possible from whatever is left of Polkadot. They control 100 validators, they control the treasury. Hence this is where the rewards go.