This is the continuation of the discussion in this topic.
The question is whether parachains should have staking&inflation.
Arguments against staking on parachains were, that parachains don’t need staking since their security is ensured by the relaychain.
Arguments for staking were, that there needs to be a decentralised selection of collators and collators need to be rewarded for their work.
What about block rewards? Many parachains choose not to reward collators through inflation, but rather by giving them a percentage of the block fees they produce. This is done using pallet collator-selection itself without any other custom pallets.
The KILT chain does this as well. It pays a portion of the transaction fees to the collators. This is also necessary to incentivize collators to actually include transactions. Transactions are additional work. If they get the same rewards with and without transactions they could choose to simply not include transactions.
Transaction fees are enough if they are high enough. The collator must earn more than they pay for the infrastructure. If that’s the case you are good.
another angle pro staking is that it might be a way how could parachain tokenholders capture value in legally sensitive way. Staking business has some recognition in law/taxing today, so you could choose it purely for pragmatic reasons.
I think the high level principle is that chains should avoid “ponzi-nomics”.
If users are going to stake + inflate your token, there should be clear justification around the value being generated which is equivalent to the tokens being minted. I think the issue is that most people have an “inflation by default” mindset, which is just one of many pillars of the shitcoins we see all over the blockchain space.
One of the key reasons Solo-chains have inflation is to refund + incentivize humans to spend resources to secure the chain. Since Polkadot provides security to all the connected Parachains, inflation for this purpose is not needed.
I do think that it can make sense to have some staking + inflation to help select honest collators, which can help censorship resistance and data availability of the network, but the value of this is quite a bit lower than what you would expect from traditional staking. Additionally, I think there is some justification to aggressive inflation rates for the purpose of bootstrapping your network, but there should then be a plan around reducing or eliminating that inflation over time, similar to philosophy around Bitcoin block rewards.
I think there are arguments that non-inflationary (or even deflationary) tokens are have better long term economics than their inflationary counterparts, however too many people are fixated on accruing free interest on their tokens without actually providing any value. I would suggest to parachain teams to not get wrapped up in the “APY” hype-cycle that we see in the current wave of blockchains, and instead focus back on the principles of game theory, economics, and utility / value.
It all depends on the purpose.
Staking in Polkadot is generally associated to PoS consensus: staking ensures security here.
But staking can have many different purposes, a good example would be Astar having dApp staking to support dApp developpers, in which case staking makes sense with the parachain model.
Staking on collators doesn’t make much sense because this is a reproduction of PoS, but parachains are not PoS blockchains.
Staking on collators actually creates unnecessary competition between collators who have to spend their begging for nomination instead of focusing on their collator role which is ensuring performance and decentralization. This is all what matters for parachains.
Parachains still need to be decentralized to ensure their own resilience but staking doesn’t solve this.
This question often comes back because of one main reason in my opinion: collators are over incentivized. Distributing a much smaller reward can drastically decrease unnecessary inflation while rewarding fairly collators for infra cost + work.
A fair reward regarding actual infra costs of a collator would be a $ equivalent 3 digit. The $ 4 or 5 digit we can see sometimes doesn’t make sense compared to relay chain rewards (see Polkassembly proposal for reference).
While I agree the inflation should stay low (and specially it should have counter-part to avoid ponzi), I think it is necessary in some cases.
A strong part of the security is done by the Relay, but some elements are still done by the collators, like avoiding censorship, assuring block production, availability…
And also some part of the chain like Randomness. Currently we can only offer guarantees proportional to the amount of token that a collator would lose to “influence” the output.
I’m not for or against staking in general, I think it mostly depends of your model. But it is important to re-evaluate it overtime and to have mechanism in place to balance it.
Given that parachains pay for secure blockspace already, it makes sense for them to balance their total spend for that service with the spending that they need to provide to collators. In principle, since liveness and censorship resistance aren’t uniquely attributable faults it is unwieldy or perhaps impossible to slash for - slashing is where the bulk of the risk and therefore the reward comes in. With that framing, I think it’s clear that collators both take on substantially less risk and substantially less reward for the work they do, and their payment should be closer to the ‘real’ costs that they incur: running nodes, monitoring, upgrading, etc.
It’s also not necessarily the case that block rewards for collators need to be paid in an underlying ‘native’ token: they could easily be paid in DOT or a stablecoin as well. We should separate ‘staking’ and collator set selection from inflation, conceptually.
Whilst it’s tempting to class “inflationary by design” coins as shitcoins you could also argue that chains that mint billions of coins from the outset, with or without inflation, are also within that category but for reasons of centralisation. I don’t think either of those points of view are helpful though, as there could be legitimate reasons for both approaches.
I’m not taking sides in that debate and it’s somewhat off-topic. What I would like to do is outline the approach that we have taken - and would welcome comments, these ideas are not all set in stone.
In essence we have taken a multi-step approach:
We minted a very limited number of coins (<3M), with transaction fees being burnt. This will create deflationary environment for our coins.
We also know that there is a real future risk that if the parachain is successful then the burn rate could create a liquidity issue.
This trend will be seen long before it actually happens and we have planned to introduce countermeasures which may include a managed inflation to offset the effect of the transaction fee burn rate or assigning the transaction fees to collators or a combination of both.
We expect that staking will be part of the mix even though it will not be done for reasons of securing our parachain.
The point of the steps is to allow us to measure the effects before we introduce the countermeasures so that they can be aligned with the reality on the ground.
one more example of PoS on parachains. We at Mangata introduced consensus for these reasons:
bigger decentralization. You give up control over nodes and don’t operate them from company, it reduces legal risk
secure the network against economic attack vector of MEV. We introduced new security considerations
I see the consensus need, because relaychain don’t do consensus about ordering. We use the PoS because we need some mechanism how to establish what the correct ordering looks like. Without it any collator could suggest any ordering (even if shuffled in Mangata’s case), but you would have no way to say which ordering was correct, since both blocks would have “correct” validity from the relaychain point of view.
Now you could say and introduce ordering rules verification on relaychain, but… should it be the responsibility of relaychain? I can imagine co-existing multiple ordering protocols operating on several parachains. We see at least two trends now in crypto - MEV democratization and MEV minimization
I think this could generally apply to even rollups on ETH, sequencers have the same problem, so they’d have one more reason to introduce consensus and own token
While there is no consensus on un-included transactions (by definition!) it is possible for a Substrate runtime to limit the set of valid orderings of transactions within a block - isn’t that the route to limiting the effects of MEV?
I used to think that a network was as decentralized as its most centralized piece. Among many other elements, the collator set is a key (relaychain agnostic) element to measure the level of decentralization of a parachain.
Staking is a good answer with a proven track-record but not the only one.
For inflation, I tend to think that the tokenomic should be adaptive. I naively guess that before defining the economic dynamics of their tokens, each parachains ran some model based on some assumptions. These assumptions may be inaccurate in the long run and need to be revised. As a decentralized network is a living beast with a powerful governance tool, even if you started with an inflationary model, you may end up being deflationary after 1 year of activity.
My answer would then be that parachains can have staking & inflation.
Over-paying collators in token can be detrimental in the long run :
- Collators have cost in fiat
- The token is generally also used for governance
I guess, some simple answers to these questions can help :
- How much, as a network, did I pay to secure one transaction in the past week?
- What is the average cost of running a collator for a week?
- What is the premium the network is paying? Does it make sense ?
In a dream world, these parameters can be re-adjusted via governance based on some TWAP/VWAP token price analysis.
Prevention of VER (value extraction by reordering) does basically that- there is an order dictated by a runtime and block executors can not deviate from it- otherwise produced state will be invalid. However this mechanism is only a part of MEV prevention and does not protect from value extraction by denial. To prevent VED, collators have obligations which are not enforced by the runtime. Since they can misbehave by willingly ignoring the obligations, and the misbehavior can be cryptographically proved, this is the mechanism where the stake became necessary.