Problem Statement
Bittensor’s ethos is about open-source collaboration, decentralization, and aligning incentives between subnet operators, miners, validators, and investors. However, the current model allows scenarios where centralized entities can exploit the network in extractive ways.
Example Scenario:
-
A company (e.g., Google) creates a subnet advancing state-of-the-art Al.
-
Investors fund it, emissions increase, and the subnet attracts the best miners.
-
The output is SOTA, but the company eventually stops open-sourcing results, hoards value, and misaligns with Bittensor’s ethos.
-
Even if token price corrects, the company can privately subsidize miners, keeping control.
-
The community cannot easily respond because subnets are difficult to replicate rapidly. This creates misalignment between: Subnet operators (who may extract value), Investors and stakers (who expect openness and value accrual), and the broader Bittensor ethos
(decentralization, transparency, and open contribution).
Core Idea: Open Subnets
To counteract misalignment and prevent extractive behavior, subnets should function more like open-source software ecosystems :
Cloning & Forking: Any subnet
can be cloned or forked, just like GitHub repos.
Open Incentive Mechanisms (IMs): Incentive mechanisms are also open, forkable, and refined collectively by the community.
Iterative Design: Incentive
mechanisms evolve via issue tracking, PRs, and continuous refinement, similar to open-source projects.
Market Competition: If a subnet operator becomes misaligned, the community can fork the subnet, improve incentives, and eventually, outcompete the centralized version.
This transforms subnets into dynamic ecosystems rather than static, team-controlled silos.
Decentralized Dynamic Subnets:
Market discovery Marketplace
A marketplace where experts from each domain analyze and discover inefficiencies, pain points, and potential optimization opportunities(basically problems) in markets. and propose an initial subnet implementation example( the solution to the problem). Such as: the incentive mechanism, different roles needed for the subnet development and maintenance etc. The initial implementation example will be solely used to inform market participants’ decisions ( is this problem worth solving? Is a subnet the best way to solve it?), because they will be the ones voting (with their Tao) for subnets that will be launched.
To avoid spams from entities that discover the so-called problems, for each submission, some amount in Tao has to be recycled. And to avoid Overwhelming the blockchain with data, these entities will have to submit, alongside the problem title, a link to the detailed article. from arXiv, X etc ( the competition and incentives will dictate the “standards”. As these competing entities will be rewarded if a subnet is launched from their propositions. So they will be incentivized to be the most creative, and clearest to attract market participants). The entities will need to “collateralize” the links they submit to the blockchain. This way, after some validation, if the links are "fishy ", their collateral gets slashed.
Mining the architecture
Now that we have “subnet ideas” (Market inefficiencies or problems). Market participants, could explore the propositions through a bittensor blockchain explorer.
Markets will vote for the ideas that are worth solving. Basically any Tao holder will be able to contribute some Tao to the “idea’s” wallet ( controlled by no one). And if an idea has enough votes ( enough Tao to fund the subsequent steps), then the idea becomes “minable”, that is the protocol could incentive ( on a sub-subnet on root, for example) miners and validators to create a "good " design for the subnet to be launched ( this will become perpetually incentivized and optimized on a sub-subnet of the subnet that will launch. To continually improve and optimize the subnet’s architecture and incentive mechanisms ). Through open source mining, miners will iterate over and enhance the subnet’s architecture(on testnet )for some time. When the testing period ends ( a time interval that is enough to get a good architecture), the best miner as of that time, will have the opportunity to guarantee that the subnet is ready to be launched. The miner will basically put some non negligible amount of Tao as collateral. If the subnet slot gets purchased, then launches, and everything is " good "(to be defined), then the miner that guaranteed the quality of the architecture gets rewarded with some bounty from the subnet funding wallet ( the one that market participants funded previously). And if there is any issue, then the miner’s collateral gets slashed. If the best miner , for some reason doesn’t "guarantee " the quality, then mining continues until the best miner guarantees that the subnet is ready to be launched.
The launch
After the subnet gets registered, some time will be given to miners and validators to prepare for their mining and validation activities, based on the last best architecture and incentive mechanisms, prior to the subnet’s registration.
The DSSNs
Each decentralized subnet, will need several sub-subnets, with their own incentive mechanisms to function properly.
DSSN = Decentralized Sub-SubNet
DSSN1: The commodity
Miners and validators will work on producing the subnet’s commodity.
DSSN2: the team
At the architecture design phase , the list of "team members " ( essentially the needed roles to operate, maintain, and grow the subnet) needed, will be specified (and can be expanded in the future if needed). And each role will have a set of UIDs competing against each other to produce the best version of the work assigned to that role. Example : if a subnet needs a developer, and let’s say 5 UIDs are available for “mining development”, then these miners will compete for each task assigned to that role. And the highest scoring miner’s work will be chosen ( and that is for each task). The work of these miners will be open sourced ( so that new miners can get inspired by the top miners). And just like on every subnet, when a new miner registers, the worst scoring miner gets de-registered. And if some roles need their own separate incentive mechanisms, then a sub-subnet should be created ( and could have already been created at the design phase).
DSSN3: A proxy for selling
Miners on this sub-subnet are arbitrage entities, that purchase bandwidth/output( a percentage of the subnet’s output), by sending alpha tokens to the subnet’s wallet.
Competition ? These entities compete with each other by bidding higher to get the most of the bandwidth (no emissions are distributed to these miners). Scoring? Validators calculate the percentage of bandwidth each entity gets based on all time transferred alpha tokens to the subnet’s wallet ( all though this could be computed by a function). For example if all time “sales” (the amount of alpha sent to the subnet’s wallet) = 1000 tao, and we have two mining entities. The first paid 800 tao and the second paid 200 Tao , then the first entity gets 80% of the bandwidth and the second 20%. And if a third party comes and payes 1000 tao, then the third entity gets 50% of the bandwidth ( all time sales are now 2000 Tao). The first entity gets 40% ( 80% of 50% which is the initial 1000 Tao).
If the subnet has the most efficient and the cheapest ( basically, is the best) way to produce the commodity, then entities will compete to get access to it and sell it for a profit on the market. And the subnet will dynamically get the best deal, based on the quality of the output and competition from the arbitrage entities ( the more miners join , the more old miners have to pay to get access to the same percentage of bandwidth). Bandwidth value is dynamically calculated based on demand. And one thing is sure, these entities will not overpay for the output.
DSSN4: incentivizing Innovation and fine tuning the design
On this sub-subnet, miners will perpetually iterate over , optimize, enhance ( perhaps, discover newer and completely different architectures) and improve the subnet’s architecture and different incentive mechanisms. Example: If for some reason the subnet needs marketing, miners could discover this and propose a way to incentivize it ( and will get rewarded for being creative , in addition to their emissions).
Miners will, maybe, test their solutions on the testnet . And after each month (for example) the subnet’s architecture gets updated with the best miner’s architecture ( with some collateral and bounty rewards, to make sure this will not break anything). This ensures, the subnet gets continually enhanced and optimized.
Incentives
- The problem identification : everyone is incentivized to identify and report problems ( market inefficiencies or even newer markets. The so-called subnet ideas ) to the market. These miners will be rewarded (either get dividends or a one time payment from the subnet), if investors choose their “ideas”.
- The architecture and design : people will adapt, and new markets will form around architecting and designing decentralized dynamic subnets. And entities will specialize to be the experts in doing so, if incentivized enough( maybe through a sub-subnet on root).
- Alpha holders : every single alpha token( from both emissions and revenue from sales) not used for the subnet’s operations, gets distributed daily to alpha holders, based on the quantity of alpha they hold. This way holders are incentivized to hold the alpha token, because the more they have the more they receive .
- The investors ( that funded the subnet launch): theses should receive more than normal alpha holders, when alpha tokens are distributed. And any free alpha in the subnet’s wallet is distributed daily. If for some reason the subnet needs funding, investors can fund it again, and get the same privilege ( receiving more than normal alpha holders) as the investors that have launched the subnet.
TL,DR
The traditional approach is : founders identify a problem, propose a solution to it , then pitch the idea to investors. And potentially receive funding ( or even fund it themselves).
The new approach will be a reverse pitch approach for solving market problems, in a decentralized and permission-less way. Problems in markets are identified, investors choose the problems they want to be solved, then pitch them to the entire world, and then eventually get the best founders to solve these problems.