Oxford proposal - adaptive issuance discussion

Research and Development
1

Greetings to all forum members!

As the cooldown phase of the Tezos Oxford proposal nears its end, we would like to express our thoughts and concerns and open the floor for discussions.

Based on our analysis, we believe that the proposal implementation may have negative economic consequences for both bakers and delegators. This is primarily due to a significant reduction in inflation rewards, despite the proposal’s main goal of increasing the value of XTZ and improving the economic situation within the ecosystem.

Having been part of the Tezos ecosystem from its inception, we at Everstake are deeply interested in the network’s growth and well-being. Therefore, we believe it’s important to emphasize the following points:

- Decreased inflation
1. Reducing inflation doesn’t necessarily guarantee an increase in token value. While it may have a positive impact on token value, what if it doesn’t react as initially anticipated, especially considering the current market conditions? That’s why it’s crucial to consider all possible scenarios and prepare a backup plan with thoughtful solutions to address potential challenges. A substantial reduction in the inflation rate may prompt many retail and institutional stakers to sell their XTZ due to the decreased APY, potentially causing a negative impact on the price.
2. This will also lead to significantly reduced rewards for bakers. Currently, the inflation rate stands at approximately 4.59%. After the proposal is accepted by the majority of voters, inflation will fluctuate between 0.05% and 5%. In all cases, bakers’ profitability will see a significant decrease in the long term unless there is substantial XTZ price growth. When it comes to smaller bakers, they are already facing challenges in maintaining their infrastructure, while there’s no active inflow of new bakers, and the reduction of rewards will only worsen the situation.

- Delegation vs. the new staking process
1. Our primary focus here is on the user experience and interests: what the new staking process will look like and how user-friendly it will be, as this is what determines whether delegators will want to transition to stakers. Also, once again, keeping in mind that stakers’ funds will be subject to slashing.
2. There might be a potential risk that only a tiny percentage of users will switch to staking. It’s crucial to engage the community further, listen to users’ opinions, and gather user feedback. With reduced rewards, delegators might tend to seek higher returns and switch to staking, however, the transition to stakers comes with increased risks, which can, on the contrary, push users away.

We are deeply interested in the implementations that will contribute to the evolution and prosperity of the ecosystem and would like to see more discussions around the potential implications of Adaptive Issuance.

9 Likes
2

Interesting. Didn’t know that stakers funds are at risk of slashing. That actually could lead to people that want to stake, to gravitate to big and reliable bakers and not staking with new bakers, making it unattractive for starting a new bakery. They could only try to get delegators with low fees. But low fees & reduced rewards are not really a argument to create a bakery.

Protocol Level Payouts & risk of getting slashed is a odd combination.

3 Likes
3

Thanks for getting this discussion under way.

Most tokenomics analysis point to XTZ’s inflation and uncapped supply as a major factor on its poor market cap action, so in theory anything that attempts to curb the inflation should, in theory, be a positive change in the long term.

There is no question that the Adaptive Issuance will affect bakers. My understanding from the proposal details is that the goal (at least initially) is to hit a 50% total staking target (from the current 70%). This means a 20% reduction in total staking. We will lose bakers (well, staked funds), and that is by design.

As you mentioned, it’s likely that those previously staked funds will hit the market, possibly lowering the price in the short term, but it’s anyone’s guess the effect a lower inflation rate will have in the long term. Again, tokenomics theory point to it being a good change.

The question in my mind is where the 50% total staking target came from and whether that is the optimal threshold for the network today. Intuitively a less disruptive approach would be to start with a higher target (closer to the current 70%) and slowly lower it over time.

IMO it would be great if Tezos proposals allowed not only a yay/nay vote, but also a discussion on the (potentially tweakable) parameters of said proposals.

1 Like
4

My 2 cents on this and Tezos in general, similar to what i’ve posted on Twitter

I’ll be the first to admit that I know next to nothing about tokenomics, so i’m only coming at this from 1 side, and i’m fully aware I don’t know everything. The side i’m coming at it from, is having spoken to real users. I’m a mod in a discord channel for an NFT project, and I join the ones from our big partners (man utd, mclauren, the gap, various games etc), to offer help to new users of Tezos. Heres what i’ve seen:

  1. Not once have I seen a real user mention our current inflation rate in a negative light. Vast majority were unaware of it. A handful of times new users have asked me to explain how it is you earn rewards from delegating, when I explain it comes from inflation … nobody cared. They all thought that it was a cool feature and found it interesting

  2. Every single person i’ve spoken to hates our dApps. They are buggy, slow, limited feature set, lacking anything that could be considered onboarding etc. Go into these discords and search for the names of our big projects, its the same in every channel

    • Why can’t I do X?
    • How do I fix X?
    • I keep getting gas exhausted, how do I fix this?
    • What does this error mean? it makes no sense?
    • Where are all the stats pages like on openSea/uniswap?
    • Why won’t support reply to me?

  3. Last time I checked, Solana (and others) had inflation rates higher than Tezos, yet they are permanently 40+ places above us in market cap rating

  4. I’ve seen dozens of developers complain about how difficult certain things are in Tezos, with a learning curve far greater than any of the others. Often having to build the tooling they need from scratch, or having to build custom versions of things that already exist to solve a problem not being addressed. This one has affected me personally, and it needs attention.

Personally, i’m not buying the idea that reducing our inflation is going to lead to any meaningful increase in price action. I understand the general concept, less new tokens should lead to greater prices. But doesn’t any inflation decrease the price unless tokens are being burnt from usage? If users continue to find experiences on Tezos poor, and don’t want to USE the chain, then I only see downsides to this proposal.

One of the things driving our current usage is the idea that you can get X for free, because your rewards paid for it. Yes i’m aware of the fact that due to inflation its not entirely true, but thats how users think of it. I don’t see how reducing these rewards is going to increase price, because I see users decreasing their usage alongside it. Resulting in the same, or a net loss if users think they’ve lost the only thing keeping them here

I firmly believe, without a shadow of a doubt, our problems all lie in poor UI/UX and experiences. There are many things on Tezos that are still, to this day, insanely complex for frontend devs. We have dozens of unique/propriety things in tezos and no low level libraries to support all these (e.g. Michelson SDK, 1 library that supports all our curves/hashing, offline forger/parser, drastic improvements to run_operation RPC, etc). Good frontend skills are entirely different to the skillset needed to read whitepapers + build low level SDKs. To me, this is quite simple: better tooling = tezos more accessible to different developers = better apps = more users. To me, this needs to be our prime focus. I personally would love to see protocol development paused, and have core teams fill in these gaps before pushing too far forward. Many of the features we already have aren’t used in any meaningful way, yet we keep piling on new features. We need to make tezos a more attractive place to build and use

9 Likes
5

The current delegation rate is 70%, but it’s liquid staking, which is different from actual staking. The actual amount staked is one tenth of that (7%), and is currently provided by bakers only.

The goal is to bring this amount up to 50%. This is achieved by

  • allowing bakers to put an arbitrary amount at stake, rather than the current 10% of delegations.
  • allowing delegators to also contribute to the stake of their bakers
  • having amounts staked earn twice more than amounts delegated.

We might still lose bakers due to lower rewards, but it’s worth keeping in mind that these rewards come from newly minted tez.

5 Likes
6

I’m beating that drum for a while now. They won’t listen. Its like a circus that plays for itself, blinded by ever new attractions they keep building for their own joy.

5 Likes
7

Following our previous statement regarding Adaptive issuance, we would like to express additional concerns which include increased slashing parameters, the risk of overstaking, and two recently discovered issues in the implementation of Adaptive Issuance.

Furthermore, in this post, we would like to share our suggestions for changes to the original proposal that may mitigate the negative effects of reducing inflation.

  1. Overstaking
    1. Bakers will have the option to decide whether they accept stake from stakers, and if so, up to which fraction of their bond (the maximum enforced by the protocol is up to 5 times the baker’s own stake). Even though there’ll be a limit for bakers to set when it comes to the stake from stakers, stakers will still be able to exceed it, resulting in a baker being overstaked and thus being subject to reward penalties.
      The main concern here is that there’ll be a limit that stakers still will be able to exceed, which will result in reward penalties for both stakers and bakers, and it is not possible for bakers to avoid being overstaked when accepting stakers.
      You can read a more detailed explanation of overstaking in the article: A Walkthrough of Tezos’ New Staking Mechanism | by Nicolas Ochem | The Aleph | Aug, 2023 | Medium
  2. Increased slashing parameters
    1. While slashing incidents are extremely rare in Tezos, the new staking mechanism implies a significant increase in the slashing amount. Currently, there is already a substantial penalty for double-attesting. Why would we want to increase the risks further? Whether it’s a double bake or a double attest, it’s not just about the financial risks; it also concerns bakers’ reputation, therefore, no baker would intentionally engage in such behavior.
  3. Another contributing factor here is two recently discovered issues in the implementation of AI, affecting issuance and bakers’ rewards. However, they are expected to be addressed only in the next proposal.
    We believe it is better to address the issues in the first place, make the necessary alterations, and not to rush into implementing AI in its current state.

The biggest point of our concern is still a huge influence of the Adaptive Issuance implementation on inflation and therefore on the validator’s incentives. Although it won’t be implemented right after proposal approval, it’s still going to cut XTZ rewards both to delegators and validators (bakers) in the future. This may make Tezos look way less attractive to both retail holders/bakers since investing in other chains with higher APR would have better economic outcomes.

Therefore we want to share our suggestions for changes to the original proposal which may neglect the negative influence of reducing inflation.

1. Gradually decreased minimum issuance

Rapid changes in the XTZ issuance may drive away both regular community members and bakers, therefore weakening the chain security and causing troubles to the entire ecosystem. Decreasing staking income without implementation of any additional XTZ utility and making direct influence on its price will hit the economic motivation of participating in the network.

The easiest and simplest way to make sure incentives won’t be too low is raising the starting minimum nominal issuance from 0,05% to, for example, 3%. The goal is to ensure that bakers and stakers are still getting incentivized properly while simultaneously adjusting inflation based on the staked ratio. Minimum nominal issuance can be adjusted by further proposals according to the market and network conditions, and the main idea is to implement the decrease gradually instead of instantly, because the possible fluctuations of staked % along with dynamic/static ratio may be unpredictable. We believe that this approach will make the XTZ community more confident in the network incentives and make transition to the new economic model more smooth.

2. Price rate, as addition to dynamic and static rate

The current Oxford Adaptive Issuance model offers to change inflation based only on the fluctuations and amount of staked funds, but this approach may be quite doubtful in case of major price volatility/continuous trends. The original idea was to increase the XTZ value by limiting its supply on the market, but this does not always work, especially given the current market conditions. The main risk of this approach is the possible price decrease, which along with the inflation cut may have a negative economic impact both on regular holders and validators independently of the staked XTZ ratio in the network.

The price rate is the new addition to the inflation rate formula, which is designed to respond to changes in the price of XTZ tokens on the market. The price rate would depend on the current market price of XTZ and its primary purpose is to counterbalance fluctuations in the token’s value. When the price of XTZ rises significantly, the price rate would reduce the inflation rate, and when the price drops, it would increase the inflation rate. That way it will work as an additional security layer along with dynamic and static rates. Initially, the price rate can be bound to the target price of 1,5$ or 2$ per XTZ, but it’s also a subject to discuss. Later, the target price may be changed by the community voting according to the current market conditions and other reasons. The price rate alone should be able to hold the inflation high enough to compensate for the possible price decline in a way that even if the staked ratio will be close enough to 48-52% target, the inflation may remain on the same level as it would be with a staked ratio of 10-20%. The main idea is to incentivize moving XTZ away from the market to staking in case of the price dump, therefore cutting supply and increasing demand.

5 Likes
8

Thank you @Anna (and all) for contributing to the debate on Adaptive Issuance and Staking. Here follow a few remarks on the discussions above:

On feature activation. Having a separate feature activation vote for Adaptive Issuance and Staking allows the community to debate these features at their own pace, without blocking the pipeline of protocol development. In Oxford’s case, it disentangles the debate from reactivating Timelocks, and the continuous improvement of Smart Rollup technology.

Ultimately, it is the bakers who will decide if and when these features activate – in the same way it is bakers who inject and vote for protocol proposals.

On reducing inflation. Indeed there are no guarantees that reducing issuance should increase token value. It is not however what Adaptive Issuance targets. Its aim is to formalize the relationship between issuance and the need for staked funds, ensuring that issuance matches the actual security budget of the chain and the incentives to encourage participants to stake and produce blocks, but nothing more.

Adaptive issuance does not guarantee lower issuance either. For example, if the staked funds ratio stays below 48% (as is now), issuance would grow to its nominal 5% maximum and would stay in that value (or quite close to that value) unless the staked funds ratio grows over 52%.

In the end, Adaptive Issuance ties issuance to an on-chain market for security deposits, and enables the protocol to adapt automatically to changes in this market’s conditions.

On delegation and staking. The new staker role (which is a complement to and not a replacement for the delegator role) entails some risks for users – and again, delegators have to weigh in on whether staking is the right option for them, or they would rather keep their delegations liquid.

Several design choices have been made so that rewards align with risk taken, encouraging baking over staking, and staking over liquid delegation.

On slashing stakers. Staked funds constitute, by definition, the security deposit of the network, and hence they need to be: (i) frozen for a sensible, predictable period of time (here, ~ 2 weeks); and (ii) slashable in the event of misbehavior. Delegations increase decentralization, and reduce the cost of consensus liveness attacks, but they are not at stake.

It should be noted, however, that slashing of staked funds is quite rare – it is more common for bakers to e.g. lose attesting rewards for failing to reveal nonces than being slashed for double-signing events.

3 Likes
9

Amen. Why someone has not offered you a Tezos dev funding consulting position, at this point, I can’t fathom.

2 Likes
10

Thank you for all the tools you’ve been building and fighting for on Tezos. Stuff like TzGo is absolutely essential to us at Tez Capital! That’s the kind of stuff that should see increased funding and that should be seriously considered for on-chain inflation funding. That’s what we would be talking about if we could somehow leave this bizzaro experience we all collectively seem to be caught up in.

3 Likes
11

@Anna

- Decreased inflation
The inflation will only significantly go down if there is success in getting lots of tez to bake and stake. If there is so much success that we’re effectively taking 50%+ of all tez off the market (vs 7% right now) for 7 cycles, it’s not crazy to think that other good things are happening like token appreciation.

Check out the graphs at the bottom of this blog post to get an idea of what kind of emissions we’d be looking at with different levels of staking On Oslo vs. Oxford and Adaptive Issuance – On Tezos Governance

- Delegation vs. the new staking process
Risk of slashing with informed consent should be fine and is a reason for stakers to really hold their bakers accountable and maintain a direct line of active communication. You say there is risk that people don’t switch to staking, in which case we keep our current emissions pretty much intact—a little lower I believe.

The process to stake will be just like the process to delegate. The user will have the option to do either and the explanation will be provided on the differences and on the risk of staking. We will provide a staking option immediately at https://tez.capital when the time comes. Popular wallets will be updated promptly I’m sure.

- Overstaking
Overstaking and overdelegation in general are problems that exist with our delegation/staking model. The silver lining, perhaps, will be that stakers will be much more involved in their situations, given the extra inherent risk and lack of liquidity when putting their tez at stake. This is also a challenge for wallets and dapps to explain the concept of overstaking and overdelegation.

- Increased slashing parameters
Like you said, slashing is rare and bakers care about their reputations and tez to do it either on purpose or via lack of due diligence. The main point for me is that the baker’s tez and staker’s tez is at stake together and it’s locked, off the market. This tez is accountable for the security of all other tez.

- Two newly discovered Adaptive Issuance bugs
I agree, it’s better to address this before implementing Oxford and that’s a stance we all should hold long term. We have 1-2 extra months but we don’t have the ability to mess up and stall our chain by mistake even once. I know these bugs don’t have that ability per se, but rushing all the time to push protocol changes even if they have issues is a dangerous mindset. We’re no longer in a position where “we must rush” to catch up to anything. There is so much of stuff we could be fleshing out instead, see previous response by @simonmcl.

- Price rate, as addition to dynamic and static rate
Relying on some kind of oracle as part of L1 operation is going to lead to problems down the line. Locking up your tez for 4 weeks is hard to manipulate but manipulating the xtz price is not as hard. That’s at least 2 systemic issues with price rate as part of the calculation.

- Gradually decreased minimum issuance
If we gradually decrease issuance as described, that essentially disrupts the free market forces we want to be at play. If people want to stake more, they should be able to and the issuance should change accordingly lower. If there was some kind of cap like 3% and people kept staking anyway, that’s overstaking and overdelegation all over again.

To make this idea work we’d need to set different xtz at stake target levels per each gradual minimum issuance level.

The main thing for me is that inflation goes down significantly only if staking is a huge success. If staking is a huge success, we have an additional 43%~ of tez that cannot be sold without waiting ~2 weeks.

You’re saying that you’re afraid of rapid changes in xtz issuance but the only reason we’d see such a rapid or permanent decrease is if there is big interest and participation, which is a very positive signal.

4 Likes
12

At last some sane voice on this matter. I wonder how people are accepting a protocol change with known issues. There should be some hard quality control in place . I completely agree with @Anna views

2 Likes
13

Could you expand on this point? When I was crunching some numbers, at 30% Tez staked I got issuance of ~0.7% (6 times lower), and it continues to decrease as staked XTZ grow. Dynamic rate is quite low to ensure substantial issuance growth and over that period bakers will remain under significant pressure to cover their costs.

Here is a spreadsheet I made. If you find mistakes in there, let me know!

I’ve shared more detailed perspective on Oxford AI that we have at P2P ORG right now in a recent forum post. I believe that AI parameters should maintain positive baker economics even considering that market conditions will get worse. Putting XTZ self-bond to work should make sense from the opportunity cost perspective as well (e.g. versus locking funds in vanilla ETH staking).

P.S. AI reflecting the price of XTZ might be something to consider but at this point of time to me it looks like an over-complicated mechanism that will have to rely on oracles and might result in harsh issuance fluctuations (maybe not but some modelling will be needed). In any case, we should continue this conversation around AI parameters whether the current proposal is approved or declined.

2 Likes
14

The caps in delegation and staking are here to prevent the emergence of staking monopolies. It’s a good thing, and unique to Tezos.

Protecting bakers from overdelegation and overstaking in-protocol is challenging. But even if it wasn’t hard to do, we shouldn’t. As you are saying, the current model keeps everyone engaged.

3 Likes
15

What I really do not understand is seeing bakers that voted Yay in former periods, voting Nay after, or not voting having said that they would do that. For example, rhe multi-protocol validator Chorus One is voting Nay now, while it voted Yay in the former period… but it has communciated the reasons behind that vote to their delegators or consulted them previously? I do not think so.

Anyway, there is still a separate voting procedure to confirm each amendmentment after Oxford proposal is activated. Why not approving Oxford with all their improvements and keeping this discussion during the “P” proposal?

2 Likes
16

Voting ‘Yay’ for testing and ‘Nay’ for going live should the tests be unconclusive is something one can understand. But it’s true it’s best to explain why.

2 Likes
17

Indeed, there is nothing wrong to change the sign of the vote… but has this been discussed/agreed with their delegators?

2 Likes
18

What reward penalties are you talking about? If the available staking space is exceeded, the extra tez just turn to normal delegations.

That is a legit concern, since the community doesn’t like patching such stuff (which is fair), and if you are afraid that sooo many bakers will immedietely turn their toggle votes to “YES” that AI will be activated before the “P” upgrade that would patch the issues, then an alternative would be to let this voting phase fail and immedietly inject the patched Oxford version.

It’s amazing to see how both you (Everstake) and P2P who are multichain validators that is just running a bussiness out of the chains, have the same concerns and points. I bet with the amounts you have staked the never ending issuance is very sweet but it’s clear as day that the current model is having a bad effect on the PA of the coin.

Thinking that the supply shock from the big reduction of issuance along with the big increase in locked tez will have no positive effect on the price after a while is just nuts to me. Just hold your rewards to sell during the bull market when the demand is increased like bitcoin miners do. Bitcoin having the halving every 4 years is a great example of how to create scarcity around an asset which in turn creates more demand.

Reducing the issuance from 4.6% to 3% is like shooting blanks. A big fat zero in terms of effect. Realize that with low issuance you will be holding an asset that has become MUCH more scarce, much less liquid and with the same (or bigger) demand. Users will keep buying the same amount of tez to buy NFTs, use in DEFI or to speculate and hodl (probably even more with the new tokenomics).

This is a terrible idea in my opinion. This kind of action would require some sort of oracle integration which is just too risky to do. I dont think the L1 chain should have anything to do with any kind of oracles at the protocol level.

2 Likes