dTAO launched a month ago: a meme-filled subnet with a broken tokenomics.

Author: donn

Compiled by: Shenchao TechFlow

I have always been fascinated by novel tokenomics. Watching how crypto protocols adjust their incentive mechanisms is always intriguing, and sometimes they seem very enticing—until they inevitably crash. So, when Bittensor launched its dynamic $TAO (dTAO) system on Valentine's Day, I was immediately drawn in.

The idea is simple: to provide a new and more "fair" distribution method for the issuance of TAO between subnets.

But just a month later, the problems became apparent. It turned out that the seemingly perfect design does not always operate as intended in a free market.

dTAO's operating mechanism

Here is a simplified review of how dTAO works:

Each subnet has its own subnet token ($SN), existing in the form of a native TAO-SN UniV2 type pool. Confusingly, although users "stake" TAO in exchange for SN, this functionally makes no difference from "exchanging" TAO for SN. The only difference is that users cannot add liquidity to the liquidity pool or directly trade between subnet tokens (e.g., SN1 → SN2), but can do so through TAO as an intermediary (SN1 → TAO → SN2).

The issuance of TAO is distributed proportionally according to the price of the subnet SN tokens. To smooth out price fluctuations or prevent price manipulation, the system uses a moving average price.

The SN token itself has a relatively high issuance volume, with a supply cap of 21 million, similar to TAO and BTC. A portion of the SN is allocated to the TAO-SN liquidity pool, while the remainder is distributed to stakeholders in the subnet (miners, validators, subnet owners). The amount of SN allocated to the TAO-SN pool is intended to balance the issuance volume of TAO in the pool, thereby stabilizing the price of SN (in terms of TAO) while increasing liquidity.

However, if the above calculation indicates that the number of SNs required by the subnet exceeds the maximum issuance of SNs (based on the issuance curve of SNs), then the issuance of SNs will be capped at the maximum value, and at this point, the price of SN (in TAO) will increase.

The core assumption of this mechanism is that subnets with higher market capitalization create more value for the Bittensor network, and therefore, they should receive a larger issuance of TAO.

However, the reality is that the tokens with the highest prices in the crypto market are often those that have the most attention, hype, Ponzi characteristics, and marketing resources. This is why L1 public chains and memecoins always tend to have relatively high valuations.

Although the starting point of the mechanism design is good, the assumption that those subnetworks that generate value through income will use part of their revenue to buy back SN tokens, thus driving up the price and obtaining more TAO issuance, seems somewhat naive.

is filled with Meme coins subnet and uncontrolled token economics

Before the launch of dTAO, I had discussions with several crypto analysts about the obvious flaws in dTAO's token economics — that is, higher market capitalization does not equal higher income or greater value creation.

But I didn't expect that this theory would soon be validated in practice. The free market operated in a "wonderful" way.

Just before the upgrade, an anonymous person took over subnet 281 and transformed it into a meme coin subnet called "TAO Accumulation Corporation" (abbreviated as "LOL Subnet"). This is obviously unrelated to AI.

It was stated on the now-deleted GitHub page:

Miners do not need to run any code; validators score them based on the number of subnet tokens held by the miners. The more tokens a miner holds, the higher the issuance they receive.

The actual situation is as follows: Speculators buy SN28 tokens → The price of SN28 rises → SN28 obtains more TAO issuance → If it exceeds the issuance limit of subnet tokens, the price of SN28 continues to rise → The issuance of SN tokens is proportionally allocated to the holders of SN "miners" → People buy more SN to obtain more TAO → The price rises further → The Ponzi cycle continues.

As a result, the issuance of TAO officially began to fund... memes! At one point, the SN28 subnet even became the seventh-ranked subnet by market capitalization.

But why couldn't SN28 take over Bittensor? Centralization saved the situation

Within just a few days, the Opentensor Foundation ran customized validator code using its root stake to incentivize people to sell SN28 tokens, causing its price to plummet by 98% within a few hours.

Source: Bittensor discord

The SN28 subnet token plummeted by 98% following the actions of the Opentensor Foundation.

Essentially, the Opentensor Foundation plays the role of a centralized entity, preventing the free market from exploiting the dTAO mechanism. This centralized intervention is currently feasible because we are in a slow transition period from the old TAO issuance mechanism to the new dTAO mechanism.

Transition from TAO's old mechanism to dTAO

The old mechanism of TAO allowed up to 64 validators to stake TAO on SN0 ("Root Subnetwork") to vote on who could receive the issuance of TAO.

This mechanism itself has triggered a series of incentive issues arising from the power held by large validators (such as the Opentensor Foundation, DCG Yuma, Dao5, Polychain, etc.). For example, theoretically, they could direct the issuance of TAO towards the subnet they invest in or incubate, or towards the subnet where they run validator nodes and earn TAO rewards.

Top validators shown on taostats.io/validators

Therefore, breaking away from this mechanism is a good step towards decentralization. I appreciate that the team has chosen a more decentralized reward mechanism, even if it means they may lose part of the issuance.

When the SN28 event occurred, dTAO had just been launched for about a week, so SN0 (the blue line in the figure below) still controlled about 95% of the issuance, allowing the Opentensor Foundation to intervene.

However, about a year later, the control over the issuance of SN0 will drop to around 20%. This means that if an event similar to SN28 occurs again, it will be nearly impossible to intervene through SN0. In this case, Bittensor may transform from a "decentralized AI" project into an incentive network for meme coins.

During this transition period, the power to control emissions will shift from the old mechanism (SN0 or "root attribute") to the new dTAO mechanism ("alpha attribute").

Admit it, this is not just a meme.

Even if we assume that people are rational enough in a bear market not to dive headfirst into meme coin hype, there is still a chance that Bittensor will evolve into a general-purpose incentive network that is completely unrelated to AI.

Imagine a thought experiment: someone has launched a subnet specifically designed for decentralized Bitcoin mining (although this is not a novel idea). The goal of this subnet is to incentivize Bitcoin mining in a resource-efficient manner, while using the mined BTC as a recurring income to repurchase the subnet token SN, in order to obtain a greater issuance of TAO.

Therefore, TAO has transformed from a decentralized AI project into a general incentive project, and the issuance of TAO is only used to subsidize various random operational costs (OpEx) of enterprises, rather than moving towards a specific goal.

Technically, it can be said that this aligns with the original intention of the Yuma consensus mechanism, as Yuma consensus aims to reach a consensus around any "subjectivity" work, not necessarily limited to AI. However, this lack of a clear goal makes the entire system seem... meaningless.

Final Thoughts

The dTAO model was launched just a month ago, and cracks are already showing.

The incentive mechanism of the free market indicates that without any centralized power, Bittensor may no longer be an AI project, but rather an "attention network" dominated by meme coin subnetworks, or a "general incentive network" led by profit-oriented enterprises that use the TAO issuance to subsidize operating costs without making substantial improvements to the Bittensor network.

I believe that the network needs a true "objective function" to unify the goals of all subnets. However, it is clearly very difficult to find a clear objective in the field of AI (especially General Artificial Intelligence, AGI) - as we have encountered various challenges when running fair large language model (LLM) evaluation frameworks... This is also the reason why the Yuma consensus was created to work on "subjectivity."

As the saying goes: "Tell me the incentive mechanism, and I will tell you the results."

In previous versions, I mentioned that the issuance of TAO is proportional to its market value, whereas in fact it is proportional to its price. This error has now been corrected, thanks to @nick_hotz for the correction.