Human, All Too Human

In 1878, Friedrich Nietzsche published a book of aphorisms whose title was also its argument: Menschliches, Allzumenschliches. Human, All Too Human. He had spent five years writing it, breaking from Wagner and Schopenhauer on the way. The thesis was simple. Strip away the metaphysical pretensions, God, Truth, objective morality, and what remains is recognizably human. Vanity. Habit. Interest. Resentment. Ambition. The will to dominate, dressed in whatever language the age finds respectable.

A century and a half later, a blockchain network promising decentralized machine intelligence arrived with what sounded like the opposite of Nietzsche's diagnosis. Intelligence would be machine rather than human. Coordination would be protocol rather than politics. Value would be discovered by market consensus rather than allocated by gatekeepers.

That promise was worth taking seriously. It is still worth taking seriously. A world where intelligence is produced only inside a handful of frontier labs and state-aligned institutions is a world where the most important technology of the century becomes another enclosed commons. Bittensor offered a different picture. Miners producing useful intelligence. Validators measuring it. Subnet teams building specialized markets. TAO distributing ownership to the people doing the work as well as those investing.

Some of that picture has been built. Templar made decentralized training credible. Grail made decentralized post-training credible. Covenant-72B proved that large-scale training over the public internet was not a fantasy. Basilica and other subnets built useful infrastructure. At least some of the builders are serious. I know for a fact that at least some of the output isn't vapor.

But the mechanism that decides who gets paid for that output is not machine intelligence.

It is human power, wearing protocol clothing.

The Charter

Bittensor's Delegates Charter contains a sentence that deserves to be taken seriously.

That is the kind of line you write when you believe the thing you are building should outgrow you. It is also the kind of line you write when you understand the central danger of decentralized infrastructure: a system whose security depends on the virtue of its founders, delegates, foundations, or largest holders is patronage wearing protocol clothing, however well-intentioned the patrons.

The Charter promises governance "in the hands of the many rather than the few." It promises transparency as the basis of trust. It promises participants the right to control their own destiny. These are the values Bittensor needs if it wants to be more than an AI incentive market with a governance layer attached.

This essay treats Bittensor more critically than I have before. Earlier this year, I described parts of the network as heirs to the cypherpunk tradition: builders still writing code while the rest of crypto drifted toward speculation and institutional capture. That judgment was sincere. The subnet mechanism has produced technical work that deserves attention.

But technical output and censorship resistance are not the same property.

Eric Voskuil's Cryptoeconomics provides a harder test than mood or sentiment. His cryptodynamic principles identify three interdependent forces required for censorship resistance: risk sharing, dedicated energy cost, and power regulation through external markets. A system missing any one of them cannot deliver the value proposition of permissionless money or, by extension, permissionless infrastructure. The framework is deductive rather than sentimental. It does not ask whether the builders mean well. It asks where power sits, what it costs to exercise, and whether the system can regulate it when the powerful misbehave.

Applied to Bittensor, that framework tells a story the Charter does not.

Risk sharing

Voskuil's first principle is that security requires distributed humans taking personal risk. Decentralization is not a vibe. It is a property of who bears the cost when the system is attacked, and whether that cost is spread across enough independent actors that no single point of pressure can collapse the whole thing.

Bittensor's validator selection runs through Taostaking. Stake TAO, delegate to a validator, earn yield. The people who control the network are the people who hold the most TAO: founders, early investors, the Opentensor Foundation, and the large holders whose stake shapes validator power. The design is public. The implications are not hidden.

The Charter promises governance distributed across the many. Taostaking produces a different structure. The few hold the stake. The many delegate to them.

Delegation does not distribute risk. It concentrates risk behind a legitimacy layer. A delegate may be honest and serious about the mission, but the delegate remains a pressure point. Coercing the network does not require finding every TAO holder. It requires finding the validators and large stakeholders whose delegated weight controls consensus. The list is short, the names are known, and the addresses are on-chain.

That is risk concentration with a UI, wearing the language of distribution.

Dedicated cost

Voskuil's second principle is that proof must be external to chain history and dedicated. The cost paid to secure the system cannot be repurposed for something else. If the proof is useful in itself, it creates an accumulation advantage. The actor capturing that useful output receives both the protocol reward and the external value of the work, then compounds that advantage into future control.

Bittensor's miners do not waste energy in the cryptodynamic sense. They produce things people might actually buy: model outputs on some subnets, social posts and YouTube content on others, real estate appraisals, sales leads, compute cycles, scraped data. A validator scoring that work is often judging whether the output was useful, and useful is not one thing. Comparing a training subnet against a lead-generation subnet is not like comparing two hashes. Some participants treat that diversity as the point, a market for many kinds of productive labor. Others see it as a bug: emissions flowing toward outputs that cannot be measured on a common scale, scored by validators with no stable method for ranking them against each other. Nobody has convincingly solved that problem.

Proof-of-work mining resets each block. Bittensor's mining accumulates. Winners earn TAO and reputational capital that make future participation cheaper and more influential. Rewards become stake, stake becomes influence over emissions, and influence can be directed toward actors already winning. Useful work compounds in ways random hashing does not. That is precisely what makes the accumulation problem harder.

This does not make Bittensor technically uninteresting. It may make it extremely productive. Productivity is not the same as cryptodynamic security.

Power regulation

Voskuil's third principle is that the system needs an external market capable of regulating power when censorship occurs. In Bitcoin, if a state actor captures 51% of hash rate, users can pay higher fees to incentivize the remaining miners to produce honest blocks. The censor's advantage costs real money and stays contestable because others can respond immediately.

Bittensor has no comparable mechanism. An entity controlling a majority of TAO stake, directly or through delegation, cannot be competed out of that position by a fee market. You cannot pay higher fees to reduce someone else's stake. You cannot incentivize honest validators into majority control when dishonest validators already control consensus. The stake is the power, and the power is persistent.

This is the cryptodynamic failure inherent to proof-of-stake systems. Once a censor acquires majority stake, the system has no external cost mechanism capable of forcing that censor back into competition. The Charter's line about giving power away therefore depends on the willingness of the powerful to voluntarily relinquish what the architecture lets them keep.

That may happen. People can be principled. Foundations can decentralize. Delegates can act with restraint. But hope is not a governance model.

Three principles. Three failures. The architecture cannot deliver cypherpunk-grade censorship resistance because the cryptodynamic properties required for it were absent at genesis, remain absent now, and cannot be added later without changing the system's nature.

That is the cryptoeconomic case. The Nietzschean case sits on top of it.

The pretense

Bittensor's premise is elegant. Miners produce digital commodities. Validators evaluate them. Yuma Consensus rewards validators who agree with the stake-weighted majority. Emissions flow toward the work the network values. In the clean version of the story, no central authority decides what counts. The market for intelligence discovers value continuously.

The word that does the heavy lifting is evaluate.

A Bitcoin miner either found a valid block hash or did not. A Monero transaction either verifies or it does not. A Bittensor miner's output often cannot be checked that way. Was this appraisal accurate? Was this social post worth emitting for? Was this lead qualified? Those questions require scoring functions, benchmark design, validator interpretation, and ongoing human judgment, and they do not reduce to a single yardstick across subnets.

Judgment requires judges, judges require selection, selection requires criteria, and criteria require interpretation by people with enough stake to make their interpretation count. Stake-weighted social consensus is the system, in the same way that gravity is a feature of mass.

Bittensor does not eliminate human judgment. It tokenizes it, weights it by stake, routes emissions through it, and calls the result decentralized intelligence. That may produce useful work. Or not. Either way, it does not remove politics from the machine. It formalizes politics as the machine's reward function.

dTAO made the politics liquid

Dynamic TAO was supposed to solve one of Bittensor's central allocation problems. Before dTAO, subnet emissions depended heavily on validator judgment across the network. Validators decided which subnets deserved weight, and their stake-weighted decisions determined where TAO flowed. The politics were obvious. If validators are the ones assigning value, then validator relationships become the market underneath the market.

dTAO changed the surface. It gave each subnet its own token, usually described as alpha, with its own TAO/alpha liquidity pool. Subnet value became tradeable. Stakers could move TAO into subnet-specific alpha. Emissions could be routed according to market signals rather than direct validator preference.

That sounds like decentralization. It is closer to the truth to say it moved the political fight from explicit validator weights into flows, liquidity, ownership, and market structure.

Under the current flow-based model, Bittensor injects TAO into subnet pools according to sustained net TAO inflows. The docs describe the logic plainly: staking into a subnet counts as inflow, unstaking counts as outflow, negative flows can receive zero emissions, and a long exponential moving average smooths the result. The more sustained capital a subnet attracts, the larger its share of that emission stream.

At the end of each tempo, roughly 360 blocks, alpha accumulated by the subnet is distributed in fixed proportions:

This is the actual machine. Not "AI earns according to merit." Not "the network discovers intelligence." Subnet owners receive a protocol-defined rake. Miners compete for the miner share through validator scoring. Validators and stakers receive the validator share through stake-weighted authority. Emissions follow flows, and flows follow confidence, liquidity, relationships, narratives, and the perceived safety of being aligned with whoever can move the market.

dTAO did not remove gatekeepers. It gave the gatekeepers a more complex market to operate.

The owner class

The 18% owner share is the simplest place to see the problem.

Every subnet has an owner. That owner receives 18% of the subnet's alpha emissions by protocol design. If the subnet attracts sustained TAO inflow, the owner receives a continuing stream of newly emitted alpha. If the owner also holds alpha, stakes into the subnet, influences validators, controls roadmap communication, and participates in the broader social market around the subnet, ownership becomes more than compensation for creating infrastructure. It becomes a standing claim on the subnet's future.

This is not automatically illegitimate. Subnet teams do substantial work. They build incentive mechanisms, maintain miners and validators, ship code, recruit participants, answer support questions, and take reputational risk. A subnet without an accountable owner can become abandoned infrastructure. The owner share exists for a reason.

But it also means Bittensor's supposedly decentralized intelligence markets begin with a proprietary primitive. Every subnet has a throne. Some thrones are occupied by serious builders. Some by extractive teams. Some by people with enough influence elsewhere in the network that ordinary participants cannot evaluate the subnet separately from the power around it.

The result is not a flat market of intelligence. It is a feudal map of subnets, each with its own owner, liquidity pool, validator relationships, miner economy, and political story.

The validator class

Validators remain the priestly class of the subnet economy.

Nietzsche would have recognized the role immediately. In On the Genealogy of Morals, priestly power emerges where value cannot be settled by direct contest. When two warriors fight, one wins. When a valid hash is found, the network accepts it. But when the question becomes "who deserves reward," a class arises to define the terms of deserving.

Bittensor validators occupy that position structurally. They score miners. Their scoring determines the 41% miner share. Their own rewards depend on stake, consensus, and agreement with other validators. Their continued influence depends on delegation. Delegation depends on reputation, yield, social trust, and alignment with large holders.

None of this requires bribery. It requires only repeated interaction among humans whose economic futures depend on one another.

A validator scoring a miner is not merely applying an objective rule. The validator is operating inside a network of relationships: subnet owners who want emission, miners who want weights, stakers who want yield, large holders who can redelegate, founders whose public signals move confidence, and other validators whose consensus determines whether the validator's own weights are rewarded or punished.

That is politics. Sophisticated politics, perhaps. Open-source politics, perhaps. Better than corporate AI procurement, perhaps. Still politics.

The admin key

The base governance layer makes the same point in a different register.

Here the docs and the chain diverge in an instructive way.

Bittensor's public governance docs still describe a bicameral structure. The Triumvirate creates proposals. Triumvirate members are Opentensor Foundation employees. The Senate approves or rejects those proposals. To sit in the Senate, a delegate must control more than 2% of total network stake through delegation or self-stake, and there can be at most twelve seats.

But the live chain no longer looks like those pages. As of this check, the Finney runtime does not expose the SenateMembers pallet that the CLI expects. btcli sudo senate and btcli sudo proposals fail with "Pallet SenateMembers not found." Runtime metadata still includes the Sudo pallet, and the sudo key is set. In October 2025, Subtensor merged a pull request removing the "legacy and non-working governance system" as dead code in preparation for a new governance system.

That is more revealing than the stale documentation. The official story says transition to community governance. The chain says admin authority remains live while the replacement machinery is unfinished. This may be temporary. It may be practical. Young networks often keep emergency levers because something breaks every week. Governance still runs through a privileged control path surrounded by social trust, not through ordinary participants.

The planned governance transition does not eliminate the problem. It changes its costume.

The current direction is still interesting. Stake-lock and conviction proposals can make subnet ownership more contestable. Subnet deregistration can remove abandoned or low-value subnets that waste emissions and attention. Those are real attempts to solve real problems.

But the selection pool still comes from the economic hierarchy. Locked stake is still capital. Subnet ownership is still a privileged role. Deregistration is still a power to decide which markets deserve to live. Making ownership contestable through conviction does not make the network democratic. It makes power more legible, more procedural, and more expensive to fake.

The danger remains the same. Used well, these tools can defend the system against parasites. Used politically, they can discipline rivals. The whole question is who gets to define "low value," "abandoned," or "exploitative" when the same market contains the accuser, the judge, and the beneficiary.

The admission

The revealing thing about the latest wave of Bittensor communications is that the serious dispute has changed.

For a long time, the question was whether critics were misreading the network. Was Bittensor already decentralized in the way it claimed? Were concerns about founder influence, admin authority, validator politics, and emissions discretion just resentment from teams that had lost market favor?

Const's own explanation answers the factual part plainly. "Bittensor is currently not a decentralized protocol in the way Bitcoin is." More specifically, at the core economic incentive layer, Bittensor is "currently not decentralized." The defense is not denial. It is sequencing. Bittensor, in this account, is choosing founder-led speed now so that it can become the "nation-state-resistant bunker" later.

That is an honest argument. It is also an admission.

Const distinguishes ownership from direction. Bittensor, he argues, was not pre-mined, has distributed TAO for more than five years, supports 128 subnet teams, has many validator teams, and lets anyone build, mine, study, or use the technology without an ID card. In that sense, he says, Bittensor is decentralized where it matters: ownership.

But then comes the more important sentence. "Where Bittensor is not decentralized is with core direction." He compares the current phase to Bitcoin's Satoshi period and writes: "I am that person for Bittensor."

There is no need to exaggerate what this means. It proves neither malice nor bad faith. It does not make every discretionary call wrong, and it does not make the eventual decentralization roadmap insincere. It proves something narrower and more important: the critique is not based on confusion. Bittensor's founder now describes the network in substantially the same structural terms its critics have been using. Distributed ownership, permissionless participation, founder-led direction, eventual decentralization.

That turns the Charter into a promissory note.

Veritas, non auctoritas

Erik Cason gives the cleanest contrast in CryptoSovereignty. Hobbes's formula for sovereignty was Auctoritas, non veritas facit legem: legitimacy comes from sovereign command. Cason argues that Bitcoin inverts the sentence. Veritas, non auctoritas facit legem: legitimacy comes from verification.

That is the political philosophy hidden inside Bitcoin's validation rules. A block is valid or it is not. A transaction satisfies the rules or it does not. A signature verifies or it does not. No founder, miner, exchange, foundation, or senate can make an invalid transaction valid by wanting it badly enough. Authority can speak. Nodes verify. Truth wins because the machine gives ordinary participants a cheap way to reject lies.

That is what "code is law" means in its strongest form. Not that every rule is just because it is written in code. Not that software replaces politics everywhere. It means the decisive political act is validation, and validation is available to anyone running the rules. Legitimacy comes from independently checkable truth.

Bittensor, as it currently operates, is almost the inverse. At the block-production layer it has ordinary cryptographic validity. Transactions settle. The chain advances. But the political economy that matters most, the question of who receives emissions and which subnets count as valuable, is not reducible to truth in Cason's sense.

There is no hash for "this subnet adds value." There is no signature check for "this team is extracting." There is no purely cryptographic predicate for "this output deserved the weight," "this validator basket is good for TAO," or "this emission block should be restored on a case-by-case basis." Those are judgments. Judgments require authorities, even when the authorities are called validators, allocators, founders, or the market.

That is the deepest contrast. Bitcoin turns authority into an input that must submit to verification. Bittensor turns verification into an authority problem: who gets to define the benchmark, interpret the output, set the weights, block the emissions, or decide when the system is finally ready to decentralize.

Root Reborn does not dissolve that contrast. It sharpens it. The proposed scoreboard is TAO-denominated yield, which is real information, but it is not the same kind of truth as a valid block. Yield measures the success of an allocation game after the game has already been shaped by liquidity, validator choice, founder roadmaps, subnet narratives, and discretionary intervention. It can discipline allocators. It cannot turn subjective governance into cryptographic law.

This is why Const's comparison to Satoshi is so revealing. Bitcoin's early Satoshi period ended by leaving behind a rule system that did not need Satoshi's ongoing judgment to decide validity. Bittensor's current roadmap asks the community to trust that founder judgment, validator judgment, and market judgment will eventually produce a system where that judgment can be thrown away. Bitcoin made truth prior to authority. Bittensor is still asking authority to deliver truth later.

Emission eligibility as executive power

The same pattern appears in the emission-blocking announcements.

In May, Const announced an interim measure to block emissions on subnets engaged in "active foul play" or offering "no clear path to adding value" to the network. The examples were not sympathetic: long-term burning of all miner emissions, no-code self-mining, dead or abandoned subnets, and TaoFlow exploitation. A healthy network should have a way to deal with that behavior.

But the governance question is not whether the first targets are bad. The question is who holds the instrument.

The announcement said a chain operation to block emissions would be available the following Tuesday. It treated conviction and shorting as future upgrades, alongside the eventual full decentralized governance system. The network introduced a direct emission-blocking tool before the decentralized governance system existed.

A later Discord announcement made the operational cadence even clearer: emissions had been ceased on all subnets with no miner distribution or no code, and this would happen weekly. Dead subnets would be removed. Active ones could be brought back on a case-by-case basis. The process would be "actively managed."

Again, the policy may be defensible. A dead subnet should not drain emissions forever. A no-code self-mining shell is not a public good. But defensible executive action is still executive action. The structure is leadership identifies bad behavior, leadership defines the criteria, a chain operation executes the block, and restoration happens case by case.

The result is discretionary governance with good reasons. It is still discretionary governance.

The distinction matters because every concentrated power begins by solving a real problem. Emergency powers are rarely introduced to protect villains. They are introduced to protect the system from villains. The test is what remains after the first villains are gone.

Root reborn, or the marketization of command

Root Reborn sharpens the point from the other side.

Const's argument for the update is economically coherent. TAO, he says, is not Solana or Ethereum. It is a productive utility token that invests liquidity into 128 subnet companies and receives a root proportion of alpha tokens in return. Today, Root sells that alpha mechanically. The result is sell pressure, distorted price discovery, and what he calls yield leakage. Root is powerful but passive, a dead subnet sitting on one of the most important allocation surfaces in the network.

The proposed fix is to make Root intelligent. Validators would compete to compound subnet yield through allocation and reinvestment. TAO holders would stake to the validators producing the best TAO-denominated returns. Root would become a meta-incentive layer over the subnet economy. Validators would become transparent allocators. The network would stop dumping ownership and start compounding into the subnets that perform.

As capital allocation, that may be better than the status quo. Mechanical selling is dumb. On-chain reinvestment is cleaner than off-chain products that run manually, charge fees, and expose users to operational mistakes. A transparent allocator competition is more legible than an informal market of trusted reinvestment managers.

But notice what kind of solution this is. It does not remove the human layer. It makes the human layer more explicit, more professional, and more financially central. Validators become portfolio managers for the network's productive assets. Their scoreboard is yield. Their power is allocation. Their mandate is to decide which subnet baskets deserve capital.

That may make TAO a more attractive investment. It may improve liquidity. It may reduce immediate sell pressure. It may reward the best subnet teams and punish extractive ones. But it is not the same thing as removing command from the system. It is the marketization of command.

The line from the Root Reborn article that matters most is not the tax claim or the yield arithmetic. It is the promise that the update will let "the validator class direct emissions to subnets providing value." That is exactly the question this essay has been asking. Who defines value? Who directs emissions? What happens when the subnet being judged is a rival, a nuisance, or a team outside the favored social graph?

Raw yield over time is a useful check. It is not a complete governance model. Yield can punish bad allocators, but only after capital has moved, narratives have hardened, and the market has decided which losses matter. It optimizes the network as an investment machine. It does not prove the network has become decentralized in the Bitcoin sense.

The founder problem

Every young network has founders. Every founder has soft power. The question is whether the architecture can survive that fact.

Bittensor has not solved this. It has personalized the problem while pretending the mechanism is impersonal.

When a founder or foundation-linked actor controls enough TAO, enough alpha, enough validator attention, enough informal influence over what the market believes will happen next, or enough access to shape emissions outcomes, the formal title carries less weight than the surface of soft power they already command. Stepping down from one role does not remove market influence. Moving from "game master" to "game player" does not make the game neutral if the former game master still understands the machinery better than anyone else and holds enough economic weight to move it.

This is the part Bittensor's public language struggles to admit. A protocol can be open and still be dominated. A market can be permissionless and still be controlled by actors whose capital, reputation, and relationships make them structurally impossible to ignore. A subnet can be evaluated by validators and still know, in practice, which individuals it cannot afford to offend.

Court politics with dashboards is what the system actually runs on at that level, regardless of what the documentation calls it.

The "still building" defense

One common response is that the network is young. Decentralization takes time. The Charter is aspirational. The mechanisms are being built.

The recent communications make that defense clearer and weaker at the same time. They are clearer because the leadership is no longer pretending the destination has already been reached. They are weaker because the path being described is not a path away from human power yet. It is a path through founder discretion, emission-blocking operations, validator capital allocation, conviction mechanisms, shorting, TaoFlow tuning, and active exclusion of teams deemed extractive.

Those may be practical tools. Some may be necessary. None of them change the cryptodynamic diagnosis.

Voskuil's framework describes properties that must be present in the architecture from genesis. They cannot be introduced later by vibes, roadmaps, or better delegation dashboards. Risk sharing cannot be retrofitted onto a system where stake is already concentrated. Dedicated cost cannot be introduced into a system where the work is inherently useful. External power regulation cannot be added to a consensus mechanism that lacks it by design.

You do not gradually become decentralized at the cryptodynamic level. You either launch with the necessary properties, or you spend the rest of the network's life depending on the restraint of the people who inherited power early.

The Charter can still be inspirational. It can still describe what the community wants to become. But a charter is not a consensus mechanism, and a roadmap is not a cryptoeconomic property.

What the Charter cannot fix

The Charter's line about releasing power is beautiful. It is also an admission. If power must be released, then power is already held. If decentralization depends on the powerful choosing self-limitation, then decentralization is not a property of the architecture. It is a hope about the character of the powerful.

Architecture is not aspiration. A consensus mechanism is not a charter. A token distribution is not repaired by better language. Delegation dashboards do not dissolve concentrated stake. Sortition among top actors does not make the selection pool egalitarian. Emission-cutting does not become neutral because the stated target is exploitation.

The cryptodynamic properties that make power structurally contestable must be present before power concentrates. They cannot be added later by asking the people who benefited from the initial structure to surrender what the structure gave them.

Why it still counts

None of this means Bittensor is useless. That would be the lazy critique.

Bittensor has produced AI work that centralized labs have no incentive to produce in the open. It created a venue where decentralized training, post-training, inference, compute, data, and storage teams could coordinate around a shared incentive layer. It made a generation of builders believe that AI infrastructure did not have to belong to hyperscalers. That accomplishment is not erased by the architecture's failures elsewhere.

But useful work is not the same as decentralized control.

The question is whether Bittensor can produce censorship-resistant AI infrastructure in the strong sense, the kind that cannot be bent by founders, foundations, top validators, major holders, or subnet owners with enough economic weight to pull on the machine. On current architecture, the answer is no.

The verification surface is subjective. Subjective verification requires validators. Validator influence follows stake and delegation. dTAO routes emissions through flows and alpha markets. Owner shares create durable subnet fiefdoms. Governance routes through foundation proposal power and top-stake approval. Future governance proposals rotate authority among the same top economic actors. The system calls this decentralization because many people can participate.

But participation is not control.

The honest diagnosis

Nietzsche wrote Human, All Too Human while stripping away illusions he could no longer afford. The book is not cynical. It is clear.

Bittensor deserves that kind of clarity. Not dismissal of the work. Not mockery of the builders. Not denial of the technical achievements.

But an honest recognition that this network now runs on validator judgment, stake concentration, alpha liquidity, owner emissions, foundation governance, and the informal power of a few people whose position no ordinary participant can match.

The pretense was that intelligence would escape human gatekeepers by becoming a market.

What emerged was a market where the gatekeepers learned to price themselves.

The token is named TAO, after the Daoist concept of the way: harmony, flow, natural order. But the machine that runs on it is closer to Nietzsche than Laozi.

The purpose of power, the Charter says, is to give it away. However, the architecture built a market for keeping it.

---

Disclosure: until recently I was heavily invested in Bittensor, both as a TAO holder and as a member of a subnet team. I no longer hold any TAO, and I never intend to hold it again.