Introduction

As of writing this, Bittensor only has 1 token, TAO, which is the commodity underpinning the entire network. With the introduction of the first Bittensor Improvement Template (BIT001), the earliest iteration of Dynamic TAO has been revealed. Note that the proposal is unlikely to pass in its current form, and the Opentensor Foundation has already suggested they’ll use community feedback to improve upon it. Before understanding Dynamic TAO and the place it will have within the Bittensor network, we must first understand TAO as it is today.

The Present Role of TAO

Bittensor is an incentivized intelligence network, and those who provide value to it are rewarded in TAO. Some examples of value-providing entities are miners, validators, and subnets. Miners are rewarded for completing tasks resulting from validator queries to subnets, validators are rewarded for confirming the output quality from miners based on the subnet owner’s validation mechanism, and subnet owners are rewarded for running value-producing incentivized intelligence networks on top of Bittensor. TAO is also used by miners to register on each subnet, in order to become a miner on the subnet. The TAO that miners use to register is then recycled and burned back into the unissued supply, extending the TAO halving date. The current TAO emissions schedule is 7,200 TAO issued into circulation per day, with one block being produced around every 12 seconds. After the first halving (approximately September of 2025), the emissions rate will be reduced by 50%, meaning only 3,600 TAO will be issued into circulation per day until the following halving occurs. This continues until the maximum supply of 21 million TAO tokens has been emitted into the network.

TAO is often referred to as being the “bandwidth” of the network. The more TAO a validator holds, the more likely they are to get their queries to subnets answered the quickest and with the best quality responses (by the best performing miners on that subnet). Nominators who delegate their TAO to a validator boost the validator’s “bandwidth” on the Bittensor network. In exchange, the nominators earn a collective 82% of all TAO received by validators. Note that receiving the collective 82% of TAO does not mean that nominators receive an 82% yield on their delegated TAO. The yield from delegating TAO to validators varies based upon the percentage of the supply delegated at any given moment, as well as the performance of each individual validator. Because of this, it’s important for nominators to delegate their stake with high-performing and trustworthy validators.

As subnets are assigned more weight relative to other subnets, and therefore receive greater emissions, the TAO rewards for participants in that subnet increase, and vice versa. Rewards for network participants are as follows:

• 41% goes to validators

  • 82% of the 41% goes to nominators who delegate their stake to validators

  • 18% of the 41% goes directly to the validators

• 41% goes to miners

• 18% goes to subnet owners

Mechanics and Drawbacks Within the Network

Under the current implementation, TAO emissions are determined by the root network, which is subnet 0. The root network consists of the top 64 active subnet validators by stake delegated to them. Each of these validators manually set their weights for the subnets, which is essentially the validators saying how valuable they see each subnet to be. A higher weight assigned by validators relative to other subnets means a higher perceived value within the network. The subnets then receive emissions based on the weights from validators, relative to the stake represented by each validator. This means that if 2 validators set the same weight for a subnet, the validator representing more TAO will induce more emissions for that subnet. Emissions are simply the number of TAO that are distributed to each subnet per block.

Though this system has been relatively effective up until this point, there are few drawbacks. These are some that I’ve identified:

• Centralization: By having the weights and emissions determined by just 64 validators, it poses a centralization risk. Validators within the root network control a key component the Bittensor network itself, as determining subnet emissions is critical

• Inefficiencies: The process of setting weights and determining emissions happens largely off-chain (which is also contrary to decentralization). It’s a manual process that requires a lot of time, at least for those who take the time to truly learn about the subnets they do or don’t assign weights to

• Emission Mechanics: In its current form, validators on the root network are not penalized for taking any particular approach to setting their weights for subnets. This means that validators who can’t or don’t want to put in the effort to learn about all of the subnets, and therefore don’t push toward the best outcome for the Bittensor network as a whole, faces no repercussions. As a result, Bittensor experiences a suboptimal approach to determining emissions through manual weight setting

• Subnet Quantity Limitations: Stemming from the inefficiencies of validators determining emissions through setting weights manually, this means that only so many subnets can exist before it’s unreasonable for validators to perform proper due diligence on them all. In turn, this disincentivizes the network from adding more subnets, as doing so will only congest this process further

• Competition Mechanics: The Bittensor network is still in it’s infancy (I know; it’s hard to believe). The entire ecosystem runs on incentivization and competition mechanics from top to bottom. The current configuration allows for subnet turnover (one subnet registers by deregistering another) in a potentially inefficient way. The most efficient way of determining values for subnets is through the market, which isn’t the case at the moment

• Miner Incentives: Miners on any arbitrary subnet earn TAO from the outputs they produce. Presently, there’s no direct incentive for them to hold on to the TAO they receive from mining, other than staking to receive more TAO or speculating on the value of TAO within the ecosystem. It can be argued that the TAO is valuable due to governance and being used as “bandwidth” within the network, but there’s no direct consequence of miners disposing of their TAO tokens

What is Dynamic TAO?

Per the current (original) Dynamic TAO proposal, BIT001, each registered subnet is assigned their own token. These subnet tokens are referred to as Dynamic TAO. The original (network-wide) TAO token is referred to as Global TAO. Dynamic TAO is a commodity like Global TAO, but on a per-subnet level. When registering as a miner on a subnet, miners will be able to use either Global TAO or Dynamic TAO to do so. Additionally, Dynamic TAO is used to determine emissions across subnets based on the distributed estimation of how valuable the Dynamic TAO token for a given subnet is, or will be, within the Bittensor network. A subnet’s Dynamic TAO is created upon the subnet’s registration, as is the subnet’s liquidity pool with an initial 50/50 weight between Global TAO and Dynamic TAO. The pool’s initial liquidity is bootstrapped by the subnet owner during registration. As more network participants obtain Dynamic TAO, the number of free circulating Global TAO is reduced, increasing the scarcity of Global TAO within the network.

Dynamic TAO Tokenomics

Each Dynamic TAO token will have the same tokenomics as Global TAO, beginning on the day of its creation. This means that Dynamic TAO will share the same emissions schedule and halving time frame as Global TAO. Note that this does not mean that Dynamic TAO tokens will have the same halving date or emissions rate as Global TAO, as Global TAO was created prior to any Dynamic TAO tokens. Dynamic TAO will simply follow the same emissions and halving schedules as Global TAO, which will start on the day the Dynamic TAO is created.

Dynamic TAO/Global TAO Liquidity Pool

Understanding the liquidity pool between Dynamic TAO and Global TAO is critical for people trying to understand the BIT001 proposal. The good news is that it’s a fairly simple concept which I’ll do my best to explain. The liquidity pool mechanics are similar to that of an Automated Market Maker (AMM), such as Uniswap. As the amount of one token within the pool goes up, the exchange rate to the other token goes down, and vice versa.

Let’s use an example to explain this in a simple manner. If you and a group of friends each choose to make 1 of 2 baked goods, let’s say cookies and muffins, you can all come together to trade them. For this example, there’s initially an equal number of cookies and muffins that have been made. If more people want to eat cookies, then more muffins will be brought and traded for cookies. As this continues, the amount of cookies available decreases due to demand, and people will have to trade more muffins to get the same amount of cookies. In this scenario, cookies have increased in value relative to muffins, and muffins have decreased in value relative to cookies. With their increased value, people may bring more cookies to trade at this higher exchange rate, causing the values of muffins to cookies to trend back to an equilibrium. The cycle continues in either direction based on supply and demand. This example is demonstrated by the advanced graphic below:

Though the example may not align perfectly with the Dynamic TAO/Global TAO liquidity pool concept, the fundamental aspects are similar. If participants value Dynamic TAO (cookies) higher than Global TAO (muffins), they’ll deposit more Global TAO into the pool to receive Dynamic TAO. As this continues to happen, it will cost more Global TAO to get the same amount of Dynamic TAO, since there’s less Dynamic TAO available than there originally was when the pool was weighed at a 50/50 split. One reason Global TAO holders may swap their Global TAO to Dynamic TAO is because they believe the value of the particular subnet’s Dynamic TAO token in the Bittensor network is greater than or will increase against that of Global TAO. The opposite is also true. If it’s believed that the subnet’s Dynamic TAO is or will be less valuable within the network, participants will swap their Dynamic TAO back to Global TAO through the liquidity pool. Participants swapping into Dynamic TAO is a bet that the underlying subnet is, or will be, valuable. Contrast to this, swapping from a subnet’s Dynamic TAO to Global TAO implies participants believe the subnet’s value is low or will be lower in the future.

A key point to note about the per-subnet liquidity pools is that the Dynamic TAO tokens can only be exchanged within these liquidity pools. They cannot effectively be listed on any third-party exchanges (unless the back-end of these exchanged used the subnets’ liquidity pools), which keeps the liquidity within these pools on the Bittensor network.

Dynamic TAO’s Effect on Decentralization and Emissions

I believe the most attractive aspect of introducing Dynamic TAO is the decentralization that comes along with it. Rather than using the root network and a subset of validators to set weights, and therefore determine emissions, for subnets, the market itself is tasked with doing so. Each subnet’s liquidity pool is used to determine the emissions distributed to that subnet. The more Global TAO and less Dynamic TAO within the pool, and therefore more value placed on the subnet’s Dynamic TAO and the subnet itself within the network, the greater the emissions for that subnet.

Emissions are distributed both in the form of each subnet’s Dynamic TAO and Global TAO. Under this new model, nobody will directly earn Global TAO. Instead, Global TAO is deposited into the liquidity pools along with 50% of the respective subnet’s Dynamic TAO emissions. The other 50% of Dynamic TAO emissions will go to the miners and validators performing work on/for the subnet.

As any holder of Global TAO can deposit these tokens into a liquidity pool to receive that subnet’s Dynamic TAO, anyone and everyone can participate in determining the emissions for each subnet. The result of this is the elimination of the root network in its current form, as it will no longer be used to determine subnet emissions. It’s difficult to stress just how groundbreaking this change will be. This is the next iteration of the Bittensor network continuing to move away from pseudo-decentralization, taking an enormous step toward the Opentensor Foundation’s goal of Bittensor being fully decentralized.

Dynamic TAO’s Relationship With Subnets

The competitive mechanics of Bittensor will improve drastically under this new system. As the market, not validators, will decide how emissions are distributed within the network, subnets are further incentivized to produce as much value as possible. Markets are touted as the most efficient model for determining what has value, and this will likely be demonstrated through this new emissions model. As subnets compete to receive emissions from network participants by providing more value, this leads to an overall value increase of the Bittensor network itself.

When a new subnet registers on the network, the owner bootstraps the pool’s liquidity by depositing Global TAO and “buying” the initial Dynamic TAO token supply. The price of the Dynamic TAO in relation to Global TAO will be the same as that of the previously deregistered subnet. During the registration of a new subnet on the network, the subnet with the lowest 30-day moving average price of its Dynamic TAO to Global TAO is deregistered. The new subnet is then registered in its place.

An issue that may arise is the concentration of emissions to a relatively small group of well-performing subnets. The result of this is that smaller, developing, and/or newly registered subnets may be “starved” of emissions, making it difficult for them to stay operational. Additionally, the 18% of emissions that currently goes to subnet owners might be eliminated, though this hasn’t been decided yet. If it is, it will put further stress on the subnet owners to receive emissions from the market.

Another problem is the ability for malicious subnet owners to “rug pull” those who deposit Global TAO into the subnet’s pool, in exchange for the subnet’s Dynamic TAO. In theory, a subnet owner could be very good at marketing their subnet (whether or not they intend to actually develop the subnet), convince participants to deposit Global TAO into the liquidity pool, and then trade their own Dynamic TAO holdings for the Global TAO deposited into the pool at a higher exchange rate. The likely result of this would be greatly diminishing the value of the subnet’s Dynamic TAO, while the subnet owner ends up with more Global TAO than they deposited. In this case, the people who deposited Global TAO into the liquidity pool will be left with less Global TAO than they deposited if they were to swap back, leading to a loss of Global TAO compared to their initial holdings. This example can also be done with a subnet owner acting on privileged information and swapping out of their subnet’s Dynamic TAO token and into Global TAO prior to an event that decreases the value of the Dynamic TAO within the liquidity pool, such as the subnet ceasing its operations.

These potential issues outline some risks for Global TAO holders who intend to deposit Global TAO into liquidity pools in order to receive subnets’ Dynamic TAO. If network participants want to be sure that they never run the risk of ending up with less Global TAO then they started with, they can simply hold Global TAO and not use a subnet’s liquidity pool to swap to Dynamic TAO. In doing so, however, they will have no say over the emissions distributed to subnets.

Dynamic TAO Incentives and Disincentives for Miners

Under the new system, miners will receive Dynamic TAO as rewards instead of Global TAO. Though a simple change, it has a vast impact on the network. As a subnet’s emissions are determined by the amount of Global TAO in its liquidity pool compared to Dynamic TAO, miners are disincentivized to swap their Dynamic TAO to Global TAO. If they choose to do so, it will decrease the amount of Global TAO in the subnet’s liquidity pool, thus diminishing the emissions on the subnet they’re mining on. This incentivizes miners to hold on to their Dynamic TAO, or to at least be strategic in how they swap it back to Global TAO.

If a miner truly believes in a subnet (and especially if they want to be a whale within that subnet), they’ll want to begin mining as early as possible, for as long as possible, and hold on to their Dynamic TAO, as to not diminish their mining rewards. Miners will have more skin in the game on the subnets they mine on, as they’re now directly exposed to the performance of these subnets. An effect of this is the free circulating supply of Global TAO being reduced, thus increasing the value of Global TAO within the network.

Dynamic TAO Delegation and Validation Mechanics

The BIT001 proposal iterates on how delegating works within Bittensor. Rather than nominators delegating a single token to validators, market participants who have deposited Global TAO into subnet liquidity pools in exchange for Dynamic TAO can delegate the Dynamic TAO to validators of their choosing. Validators don’t have the ability to transfer the Dynamic TAO delegated to them.

Validators determine which subnets to validate on according to the quantities and diversity of Dynamic TAO tokens delegated to them. Validators will prioritize validating on the subnets with the largest quantities of Dynamic TAO delegated to that validator. This is because the nominees who are delegating their Dynamic TAO with the validator are telling the validator what subnets they value most. As a result, validators are incentivized to validate on these subnets. Failing to do so will result in a loss of delegations, as nominators will move to validators who validate on the subnets they hold Dynamic TAO for. A loss of delegated Dynamic TAO results in a loss of validation rewards, along with a lower priority when querying the subnets for which the delegated Dynamic TAO has been reduced. Like miners, validators will have more skin in the game, as they will have to acquire Dynamic TAO on the subnets they validate on in order to stay competitive. Because of this, more Global TAO will be reduced from the free circulating supply, as it’s held within subnet liquidity pools.

With this new system, the yield for people delegating their Dynamic TAO will heavily rely upon the performance of the subnet they hold Dynamic TAO for and the validator(s) they choose to delegate to. There is no more “risk-free” yield for nominators delegating their stakes to validators. Market participants are now incentivized to research subnets as much as possible and assign the most accurate value to them, which is done by swapping between Global TAO and the subnets’ Dynamic TAO. Different subnets will have varying performance, with some doing amazingly well and others failing. In turn, the yields and values of Dynamic TAO tokens relative to global TAO will fluctuate. If a Dynamic TAO holder wishes to receive any sort of yield for their Dynamic TAO tokens, they must delegate the Dynamic TAO tokens to a validator. As stated above, if network participants want to be sure that they never run the risk of ending up with less Global TAO then they started with, they can simply hold Global TAO and not use a subnet’s liquidity pool to swap to Dynamic TAO.

tl;dr

• Each subnet’s Dynamic TAO token will be representative of the subnet’s value within the Bittensor network, according to the market. Dynamic TAO can only be swapped for Global TAO within subnet liquidity pools, retaining value within the ecosystem

• Emissions will be determined by the market. The result is a more decentralized, efficient, and democratized mechanism for rewarding network participants. This is, in my opinion, the most important aspect of the proposal

• Miners and validators will have more skin in the game and are incentivized to acquire and hold on to the Dynamic TAO tokens for the subnets they mine/validate on. In doing so, the amount of free circulating Global TAO will be drastically reduced

• There are some newly proposed risks, such as:

  • Potential “rug pull” opportunities for malicious subnet owners

  • Those who swap their Global TAO to a subnet’s Dynamic TAO may end up with less Global TAO than they initially had, based upon the subnet’s performance

• It’s yet to be determined if subnet owners will continue to receive 18% of the network emissions

• The BIT001 proposal is not set in stone. The Opentensor Foundation is actively listening to community feedback, and will try to benefit both the network and its participants to the best of their abilities

My Thoughts on BIT001

As a whole, I’m very pleased with the direction Bittensor is heading in, and I feel that BIT001 strengthens the network. I think the largest benefits to this proposal are the decentralization and efficiencies of having markets play a larger role within the ecosystem. Markets are, generally, extremely efficient. All network participants will be incentivized to provide value to Bittensor, bolstering the worth of Global TAO within the network. Having emissions determined by all Global TAO holders is important, as it puts power into the hands of token holders.

I believe there are some incentives for bad actors to game the system, such as creating subnets with low value propositions but good marketing techniques. Subnet owners, even the trustworthy and productive ones, may feel that potentially losing the 18% of emissions they’re currently receiving is unacceptable. Additionally, Global TAO holders may not want to swap into Dynamic TAO if they run the risk of ending up with less Global TAO then they started with, even if they’re earning a yield for delegating the Dynamic TAO they’d receive.

Ultimately, even in it’s current form, I feel this proposal would benefit the network greatly. I do, however, believe it will have revisions prior to being deployed. Notably, subnet owners may want some, or all, of the emissions they’re used to receiving, and Global TAO nominators may want a “risk-free” or a “lower-risk” method of delegating their stakes. One such method of doing so is a basket-based approach, where Global TAO holders would be able to swap into and, potentially, delegate all subnets’ Dynamic TAO tokens automatically. Regardless of the decision by Opentensor, it’s clear they’re pushing Bittensor to be the best version of itself. I very much appreciate the vision of Opentensor and their willingness to communicate with and take feedback from the members of the community. I hope this (quite long) post helped people further understand the Dynamic TAO proposal, as well as some positive/negative aspects of the current state and potential future state of Bittensor. It’s clear to me that Bittensor has a bright future ahead of it, and I’m grateful to be a part of the journey. Thanks for reading!

Thank you so much for reading! Please note that both inτaolligence and the author of this post both hold a position in Bittensor. We appreciate everyone who sacrifices their valuable time to do so. If you have any questions, comments, corrections, or anything else you’d like to tell us, please reach out here: [email protected]. Additionally, if you’d like to further show your support through a donation, you can donate to any of the addresses below:. All donations will go toward improving the quantity and quality of this newsletter’s publications. Thank you!

TAO: 5EEvipyF2wkL9neNwswajNMtQ1e4UWGxGxubbNVWv3h6MGZc
ETH/ERC-20 tokens : 0x1eaAeD5d4A6570c6A02C242adedeb88BE7aE1079