A criticism of the ATOM 2.0 white paper and proposal #82

I will start by saying that this is not “just a signaling proposal” in my opinion. It is setting the direction of the whole Cosmos Hub. This is the exact intent actually. If I understand correctly what “signaling proposal” means in this case, it means that the Cosmos Hub is embracing the proposed changes and components, with the assumption — or rather faith/trust — that even if they are somewhat underspecified they will be eventually refined to the point they become good.
I argue that this is nonsense. Because a “good on paper” component, especially if underspecified, does not hold any guarantee of success.
And the ATOM 2.0 paper doubles down by presenting several underspecified components and changes whose efficacy, in concert with each other or on their own, is unproven.
So ATOM 2.0 is in essence a bundle of orthogonal things — because yes, they are orthogonal and could be explored and voted independently — each of which is underspecified, and hence we can’t say if they work or not. So how can we say already that they work together?

It also seems to me we are twisting what a “signaling proposal” means. Any proposal should be subject to heavy scrutiny and criticism, at all times. The Hub is secure because of its stakers, and letting something pass with oversight can set a precedence. An idea proposed to the Hub governance should at the very least be specified, with clear arguments on why its benefits or success are supposed to unfold. Yet this proposal does not do any of these things. It actually uses many forward-looking statements that are presented either as facts or as obvious/deterministic consequences of some choice or component discussed in the paper. It is even said in the disclaimer at the end to not trust these forward-looking statements, yet many seem to be doing it.

But let’s go in order. I will now try to rationally criticize some parts of the paper that I deemed relevant, with the best of my capabilities. So hey, I could be wrong. My capabilities are limited.

I invite the reader to have the ATOM 2.0 paper handy as I am going to refer to it directly, and following might be hard without keeping an eye also on the actual text of the paper.

1. Introduction

In page 2 there is the first forward-looking statement:

In this new role, the Cosmos Hub becomes a secure platform

Yet the paper proposes 2 things — monetary policy changes, and liquid staking — that arguably alter the security properties of the Cosmos Hub in ways that haven’t been fully explored and discussed, and for which we don’t know the results. How can one claim that the Cosmos Hub will remain a secure platform? What if it becomes unstable? Attackable?
Does Interchain Security hold any value proposition with an insecure Hub? This is undiscussed. Risks and implications actually are never
discussed here. We only find a plethora of forward-looking statements. When presenting “very bold offers” as Ethan Buchman stated in an episode of the Epicenter podcast, you are at least expected to discuss risks and implications, evaluate a number of possible alternative outcomes and making an hypotesis on what is more likely to happen. Or these “bold offers” should not be presented to on-chain governance, they are not ready.

All I see here is a lot of forward-looking statements sold as facts. No discussion on risks, no implications.
This pompuous-looking paper that wants to mock academic papers and confuse the common holder maybe after all is not a paper at all.
Is this perhaps more marketing than an actual vision for the Hub? One has to ponder.

Still in page 2, when discussing about Liquid Staking the author’s claim is

a critical part of the proof of stake economy

To substantiate this affirmation, the paper references to a 2 years old report by Chorus One, one of the validators of the Cosmos Hub. Yet the report presents the possibility of this becoming true, but do not perform this conclusion. Being a report, it does not provide any definitive conclusion as it was out of scope. Using this report as support for this statement is misleading.

In page 3 the diagram in figure 1 color-codes the so called layers of the ATOM 2.0 vision so that it seems that Interchain Security enables the Interchain Scheduler, and Liquid Staking enables the Interchain Allocator.
Is this true? Am I seeing things?
Was it not intentional? Well I argue it was.
While the first if-then-can (ICS->Scheduler) makes sense, the second (Liquid Staking->Allocator) not so much.

Liquid staking is even mentioned in the first page of section 4.2, and staked ATOMs used explicitly as the asset for economic coordination instead of ATOM.
And if that’s the case, then you tell me what is the relationship between
Liquid Staking and the Allocator, because I can’t see it.
If you replace staked or bonded or liquid staked ATOM with only ATOM everywhere is mentioned, the whole Allocator section retains all its meaning. So the dependency between Liquid Staking and Allocator is simply forced.
I can’t actually see, to be completely honest, how Liquid Staking fits in the whole ATOM 2.0 vision.
I mean I don’t want to argue on the merits of Liquid Staking here.
But I argue that it could be surgically removed and treated independently and the rest of the “vision” would not get altered.
I have actually here deconstructed its implied relationship with the rest of the “vision”.

To me Liquid Staking is a clearly contentious outlier that at the very least should be explored independently.

Still in page 3

With these primitives, the Cosmos Hub will become a self-propagating economic engine that drives the expansion and integration of the Cosmos Network

Again, forward-looking statement that is fundamentally unsubstantiated. Truth be told, there is nowhere to be found an actual clear, succint explanation of why these components do indeed work together. Let alone work so good together to provide this triumphant result for the Cosmos Hub.

Page 4, point number 2, states:

In turn, these projects expand the Scheduler’s addressable market.

This statement assumes that the projects funded through the Allocator will all succeed or (rather and) all offer potential for MEV extraction, or at least in large numbers.
Otherwise why this circularity would be so central in the “vision” for the Cosmos Hub.
It is so vitally important for the “vision” that they are bundling together the
two components (Scheduler and Allocator) as if they depend on each other. But it is yet to be proven that this circularity can actually materialize, at least in the way the authors envision it.
Most importantly, even if it does, it might be minimally impactful, or actually damaging.
Risk-taking includes also (obviously) the possibility of failure. This is true
for anything, and so yeah, even the Allocator. And what if the Allocator’s
performance is a net 0? what if the number of projects that end up not succeeding is higher than the ones that do?
Then the Allocator becomes only a place to waste the revenue obtained via the Scheduler.
I stress out that this assumed-to-be factual statement is yet again unsubstantiated.
On page 5 the authors immediately use one more time a forward-looking statement

The result is a renewed role for ATOM as preferred collateral within the Cosmos Network.

Again, for how much it may sound exciting, it’s unproven. My argument for why this is was explained above.

2. Securely Scaling Cosmos

The chapter starts with:

The Cosmos Hub is the most secure pillar of the interchain.

It is true today. And the whole paper — not just this section — builds up heavily on this premise.
We all agree that this is the case. Yet at the same time the paper introduces
proposals for changes that effectively alter the security model of the Cosmos Hub.

This is actually stated by the authors themselves when discussing Liquid Staking at the bottom of page 7

Liquid staking requires providers to custody assets, and therefore represents a significant change to the Cosmos Hub’s security model.

The overarching argument here, for how I understand it, is that the Cosmos Hub is secure. We can leverage its security to do amazing things, plus more. Yet we introduce a significant change to its security model— actually more than one if we include the monetary policy changes — shifting the security of the Hub from provably secure to “yet to be seen”.
So I do see a contradiction here. Regardless of the merits, and the potential mitigations of these risks.
If you build your arguments around an assumption that you invalidate yourself, then the arguments break. So your entire buildup does not hold.
I am not saying here that this or that is bad or dangerous.
I am arguing that there are gross inconsistencies in the reasoning of the paper.

2.2. Liquid Staking

The section starts with:

A practical consequence of the current competition between staking and external uses of capital is that most staking assets are confined to their originating chain, which hinders cross-chain composability. Therefore, full economic integration of the interchain requires liquid staking.

The therefore-requires in this statement is like pulling a rabbit out of a hat. This cannot be simply stated like this and requires providing arguments. What are the alternatives? Osmosis is doing superfluid staking. Not saying it is the solution for the Cosmos Hub, but it shows that the therefore-requires relation to liquid staking is really loose. In addition, to support the first part of the statement the authors reference to a document authored by Zaki Manian, outlining a vision for the Cosmos Hub that is factually old and superseded, hence incorrect and should not be used as reference.
Moreover, in the first point of that article, which is the one perhaps more relevant to this discussion, the author states:

Presently, all ATOM stakers get compensated with more ATOMs for helping to secure the underlying network.

It is not true. Factually not true. The ATOM inflation works in the reverse. It penalizes non-stakers for not staking. There is no compensation happening, as there is no value created. Given the intentional use of compensated, I argue that this statement shows that this basic concept about ATOM was not understood by the author at the time of writing.
So referencing to this document as justification before the therefore deprives the statement of strenght instead of adding it, which is what a reference is generally used for.
You could, and should, find another one. This goes on top of the unproven assertion made by the therefore-requires.

Right below the previous statement the paper also states:

The user experience and capital efficiency improvement offered by liquid staking is so substantial that liquid staked assets are expected to become the dominant transactional medium — particularly for use-cases exogenous to the originating chain.

For what I know — and I have written and reviewed peer-reviewed papers in academia — usually a statement like this one deserves a reference. The use of expected to become necessitates supporting evidence that analytically shows this trend, and allows an estimation by projecting such trend into the future. Neither the ATOM 2.0 paper does this, nor uses a reference.
In academia this would result in a “strong reject” from a reviewer almost straight away, or science would be made of guesses.

Since the benefits of liquid staking effectively makes these applications unavoidable, the Cosmos Hub must ensure that the liquid staking system is as safe and decentralized as possible by providing for it in-protocol.

We are not sure that providing for Liquid Staking in-protocol is the best solution. Ethereum — for example — does not do this, and it already has liquid staking. There is no existing blockchain that do this to this date.
I am not saying it isn’t true. I am not saying Ethereum is doing it right, or it is safe. All I am saying is that yet again, unproven relations are stated as facts. This is a pervasive attitude throughout the whole paper.

I already stated, and I will repeat, that Liquid Staking deserves its own separate discussion.

3.1. Issuance

However, keeping 2/3 of the monetary base staked is a substantial capital
inefficiency that limits growth and hinders cross-chain composability.

This statement is plainly wrong. This assumption actually forms the basis of the PoS mechanism.
I think there is actually a confusion in the language in the first place. Or at least I see some ambiguity. If by monetary base we refer to the total underlying capital value of the ATOM supply, then staked here is confused with locked, because with the proposed Liquid Staking the ATOMs would (and should) remain staked while the capital would be unlocked as it gets transfered to the derivative asset (doesn’t it actually mean that the capital is getting moved away from ATOM, hence the chain is reducing its security?). If instead by monetary base the authors intend the ATOM supply, then the statement is saying that the Cosmos Hub target staking ratio of 2/3 is inefficient (calling for basically “unstaking”). Yet it is literally the only way to keep the Cosmos Hub secure.

I think in general this statement under scrutiny does not actually make any sense. I do see either a gross mistake (yet this gross mistake is made in an important on-chain proposal) or a more alarming misunderstanding of the security model of the Cosmos Hub. Or I am completely wrong and invite comments.

For this reason, the Cosmos staking economy is on the verge of a transformation
that will impact both the proof of stake security model and validation businesses built on inflationary rewards.

The impact on the security model is stated again. Yet Interchain Security
(and a prosperous future of the Hub) is only enabled when
the Hub remains provably secure. As I stated before, this exposes a
big fallacy in the narrative of the whole paper.

On page 9:

As a safety measure, if the staking rate ever falls below 2/3,
the new monetary policy will pause and the original monetary policy will resume, incrementally increasing issuance up to a maximum percentage of supply until the staking ratio again exceeds 2/3, at which point issuance will return to the steady state tail emission rate

In my very humble opinion, this proves that the current model is the one that holds, while the proposed monetary policy changes are experimental. So they alter the security guarantees of the Hub. This contradicts…you know the drill, this is the third time by now.

4.1. The Interchain Scheduler

I find ironic that one of the authors of the paper referenced at the beginning of this section claimed that:

it makes absolutely no sense to vote on Atom 2.0 atomically (all or nothing) — some of the ideas could be executed on quickly whereas the others are at
a level of R&D that they basically are citing my papers.
And not that I’m knocking my papers — but this is like saying you can have ImageNet tomorrow right after Cybenko proved the representer theorem (there was a 30+ year gap between those events!)

Yet, let’s look at the Scheduler.

On page 13 the scheduler is described.
In point 1 it is stated:

Chains may sell as much blockspace in the marketplace as they wish, above some minimal threshold.

How do chains decide how much space to sell in advance?
What happens if this locked space goes unsold? Does this mean that blocks might be produced with empty space making the chain possibly incur in suboptimal performance?
I don’t have these answers, but the authors seem to not have them either.
Yet these are actually important questions, at least in my opinion. And they
need to be answered before I can personally say “yeah, this Interchain Scheduler makes sense”.
If these basic questions are unanswered, then how the scheduler might (or might not) work is unknown. Therefore, so is its effectiveness.

Moreover, there are — I believe — performance implications for block production when adding the IBC-messaging required to excercise the right of using block space sold via the Scheduler. And unless I am missing something obvious, this aspect is not analyzed, nor mentioned, at all.
If a sovereign chain evaluated this performance impact as too high, it might choose to not join. The irony is that likely the chains that decide not to join as they desire to run as fast as possible might be the ones where the most MEV could be extracted.

4.2. The Interchain Allocator

I will not analyze in depth this section, but will provide some comments.

I think that how the Covenant would practically work is too unclear.
Incentive alignment can truly be achieved via creating a relation of interdependence first, and then there can be the involvement of capital to strengthen the relation. Unless the two things are done at the same time of course. But capital alone does not really create incentive alignment. It is merely an investment.
The authors indeed propose a series of points as candidate for the Allocator’s mandate (page 16) intended to provide guiding principles for projects funding. So the interdependence that is necessary outside of capital is indeed acknowledged.
Yet it is unspecified how the Covenant would actually help in this regard. But this is also due to the extremely short explanation of the Covenant itself.

There is in general too much emphasis on capital throughout this section.
The Rebalancer is another proof.
The mechanism by which the interdependence part is kept intact is unclear.

The reality is that the whole section is unclear or heavily underspecified to be taken seriously. Less than 1/3 of this section is dedicated to explaining the actual components of the Allocator (without images it is close to a half-page). The remainder does not discuss how the allocator works, but rather its uses and supposed effects, as if it is already here and working.

It seems that the Allocator is proposing making the Cosmos Hub become a VC firm. Some may say “ok, good”. Maybe. Maybe not. But again there are clear risks in creating and adopting a component such as the Allocator that deserve their separate, independent considerations.

Conclusions

I tried to provide a review of the paper as if it was subject to a scientific
peer review round. From what I can see, the paper does not hold and the proposal should be rejected. I don’t condemn the authors of the paper. I instead applaude them for the effort put in. Unfortunately effort does not translate into results all the time. This paper presents interesting ideas, yet they deserve to be treated independently, and substantially expanded so that the community understands each of them and how they are actually supposed to work, before approving them for the future of the Cosmos Hub.

Feedback is most welcome.
I recognize I might be wrong, but I invite the reader to argue on why.