"Flaw" in DTAO design / Concerns with Deregistrations #1975
Description
Is your feature request related to a problem? Please describe.
DTAO creates unfairness through it's asymmetric halving of TAO and ALPHA that compounds over time.
The unfairness becomes especially relevant with deregistrations in mind, and should be considered before implementation.
Let's consider two identical subnets, both receiving 10% of total emissions (those don't account for slippage):
Subnet A (Created Today):
- Receives 0.10 TAO/block
- Faces 1.0 ALPHA potential selling pressure (including a 0.41 ALPHA guarantee)
Assuming the ALPHA price is also 0.10 TAO, the 0.41 ALPHA equals 0.04 TAO
The net liquidity injection thus is 0.06 TAO and 1.41 ALPHA (given we add ALPHA)
Subnet B (Created After 5 Halvings):
- Receives 0.003125 TAO/block
- Faces 1.0 ALPHA potential selling pressure (including a 0.41 ALPHA guarantee)
Assuming the ALPHA price is also 0.10 TAO, the 0.41 ALPHA equals 0.04 TAO
The net liquidity injection thus is -0.036875 TAO and 0.44125 ALPHA. The TAO injection would not be negative in practice, but it would end up being virtually nothing/none.
While the TAO injection halves the ALPHA emitted remains same for new subnets, becoming increasingly "destructive".
The second problem would be "implemented" by the deregistrations/dissolving of subnets:
NAV/ADR was the initial criteria proposed for the deregistrations, but isn't being considered anymore as it doesn't work in the future.
Assuming there are is no burned ALPHA the ADR is calculated as ALPHA_out / ALPHA_in, and what it effectively shows is not just the ratio between the ALPHA distribution, but the ratio between the current ALPHA price and the liquidation price.
Think of it this way if each block subnet A gets 1 ALPHA injected and 1 ALPHA emitted it would end up with a 1:1 ratio and thus an ADR of 1, that means the liquidation price would equal the 0.10 ALPHA price. (with root proportion etc. the ADR could end up being higher)
Now let's take subnet B to compare against again, if it would get 0.03125 ALPHA injected and 1 ALPHA emitted, the ADR would end up being 32, that means the liquidation price would equals 0.10/32 or 0.003125 TAO compared to the actual ALPHA price of 0.10 TAO.
This means that the liquidation would become increasingly more harmful towards ALPHA holders of late-stage subnets, where the liquidation guarantees a 96.875% loss in value. (That is all without looking at the market/speculating, just the effects created by the chain).
Combine all of that with deregistrations based on the EMA price and one end's up with post-halving subnets being less attractive through disproportional behavior from the chain, compared to early-stage/today's ones. Ensuring that the newer/younger the subnet (in TAO halvings) the lower it's ALPHA price would be and thus only they would be considered for deregistrations.
Describe the solution you'd like
A "rule" got already changed (well remains to be changed) where ALPHA injection halves with the TAO block emission, so that the new subsidy mechanism works correctly, compared to how it scaled with ALPHA block emission until yet.
One solution that exists would be to scale ALPHA block emission with the TAO block emission, and thus eliminate the asynchronous ALPHA halvings. Instead of independent per-subnet ALPHA halving schedules the ALPHA emission would scale globally.
This would fix the selling pressure scaling proportionally with the liquidity injection, the ADRs would normalize towards ~1.0 meaning it could be reintroduced as a deregistration criteria and would solve the compounding difference between ALPHA price and liquidation price. Resulting in subnets facing equivalent "challenges" relative to the context of their emissions. (even if 20 years apart).
There are other consequences this creates, mainly that post-halving subnets will not be able to reach their 21 million ALPHA supply.
Describe alternatives you've considered
One could eliminate the idea of deregistrations entirely in DTAO, and while this would solve the potential liquidation problem, and unfair deregistrations through asynchronous root proportion selling pressure. It doesn't solve the underlying structural issues through DTAO but "hide"/remove symptoms rather than the root cause.
We could also maintain the current independent ALPHA/per-subnet halvings, to preserve the intended 21 million ALPHA supply, but this results in structural unfairness (as described) and increases the barriers to entry for newer subnets.
Additional context
No response