Person. const · Const · unconst · Jake Steeves · Jacob Robert Steeves
Former Google engineer who co-wrote the Bittensor whitepaper with Ala Shaabana, ran the Opentensor Foundation through mainnet, Finney, Revolution and dynamic TAO, and remains the network's most quoted voice and its most contested steward.
As with other commodities, markets could help us efficiently produce machine intelligence. We propose a market where intelligence is priced by other intelligence systems peer-to-peer across the internet.
9 Mar 2020 · in contextEvery quote is verbatim and dated. The reading above each era is written from the cited quotes only. Numbers link to the quote they cite.
The whitepaper and the first README frame Bittensor as an intelligence market and an intermodel benchmark: intelligence is a new commodity that needs a new market, priced by other intelligence systems 002 003 001.
marketThis software is designed to run an intermodel machine intelligence benchmark which measures the performance of other learners via informational significance.
Opening line of the README on his personal unconst/BitTensor GitHub repository, the pre-Opentensor prototype; the README is undated on the page and the year is taken from the repository's history.
As with other commodities, markets could help us efficiently produce machine intelligence. We propose a market where intelligence is priced by other intelligence systems peer-to-peer across the internet.
Opening of the whitepaper abstract; arXiv v1 lists Steeves as co-author with Yuma Rao, Ala Shaabana, Daniel Attevelt and Matthew McAteer, while bittensor.com credits Yuma Rao alone.
A new commodity needs a new type of market.
Whitepaper introduction, arguing that intelligence is becoming a commodity untethered from specific objectives.
The result is a collectively run intelligence market that continually produces newly trained models and pays contributors who create information theoretic value.
Closing sentence of the whitepaper abstract.
On community calls he speaks in Bitcoin's register: Bitcoin is an alignment technology that built the largest supercomputer, and Bittensor bakes the incentive in first to build a neural internet, a global brain for AIs, with no pre-mine and control diffused across humanity 006 007 013. The Finney update adds subnetworks, delegation as liquid democracy and freedom from centralized control 021 020.
neural internetShift From a paper about a market to a movement about alignment and freedom; the brain metaphor peaks here.
We're building a way of aligning individuals and compute so that we can work on these incredibly large problems without being underneath a particular organization which holds the keys.
Community call on The Bittensor Hub, contrasting the network with corporate AI labs.
Bitcoin is aligning individuals through this incentive mechanism. It's an alignment technology, which then allows us to scale to this super internet scale, peer-to-peer, open source, decentralized.
Same call, explaining why Bittensor copies Bitcoin's incentive design.
We like to say that we're building the neural internet, because for the first time ever, we're successfully doing machine learning across thousands of individuals ... We're like the Marshall McEwan global brain, but for AIs.
Same call; the auto transcript renders McLuhan as McEwan, and the ellipsis cuts a clause about individual computers.
Machine intelligence, this stuff is billion-dollar projects. So if we want to connect people together, we have to bake in the incentive first, the value first.
Same call, on why a token is needed before any collaboration can happen.
This is such an expensive, but also such a valuable commodity, we require that commodification layer to allow people to really share this thing.
Same call, describing machine intelligence as a commodity that needs a commodification layer.
If we want something to be aligned with us, we need to make sure that it has control over it and that control is diffused amongst humanity.
Same call, answering a question about AI safety.
What OpenAI did, is they built a company on top of that core model.
Same call, describing OpenAI's business as a company wrapped around one model.
In the core of the technology is competition. But what really will take us up to the high levels of performance will be collaboration in and around that competition.
Same call, on how miners relate to each other.
We didn't do a pre-mine because we wanted every iota of value to flow through the computational aspect of BitTensor.
Same call, on the launch of TAO.
There's the foundation, which is a non-profit guiding the incentive mechanism. Then there's miners running infrastructure. And clients of the system. All three groups have power in BitTensor.
Same call, describing who holds power in the network; no timestamp captured.
Inspired by the efficiency of financial markets, we propose that a market system can be used to effectively produce machine intelligence.
Abstract of the workshop paper he first-authored with Ala Shaabana and Opentensor colleagues for the TSRML workshop at NeurIPS 2022.
There must be a more efficient and objective method to evaluate AI model performance.
Introduction of the same workshop paper, on benchmarks and peer review as the problem Bittensor answers.
Bitcoin created the largest supercomputer in the world, as ... building a censorship-resistant technology.
Panel on AI and Polkadot hosted by The Bittensor Hub; the ellipsis marks a clause lost in the auto transcript.
If we build that into a market, we can leverage the innovation of this open, decentralized, borderless worldwide community.
Same panel, on why intelligence should be traded in a market.
If we don't make sure BitTensor lifts off ... into a decentralized system where regulators really can't interfere, we're compromised.
Same panel, answering a question about regulatory risk.
Finney is about BitTensor's freedom in Web3. Freedom from centralized control.
Friday community update introducing the Finney network.
Subnetworks are basically just running multiple Nakamoto consensus across the same chain so we can scale horizontally.
Same update, explaining subnets before the first ones launched.
Delegation is effectively a liquid democracy where individuals can build organizations and move TAO around.
Same update, introducing delegated staking.
It's also a way for us as a foundation to make revenue ... we've built in a flat rate of 18%.
Same update, on the delegate take and foundation funding.
The enemy gets a name. To Fox Business he describes OpenAI's gravitational pull toward trillion-dollar corporations, an elite that controls AI, walled gardens built by licensing, and insists that Bittensor is not the foundation 026 027 029. A two-word post names the coming upgrade 024.
walled gardenShift The neural internet drops out; anti-OpenAI framing and regulatory resistance come forward.
Bittensor revolution
A two-word post on X, the first public use by him of the name later given to the October 2023 upgrade.
What dystopian future are we fighting against?
Fox Business interview about leaving Google and launching the protocol, framing the stakes.
There's a strong gravitational pull for this technology to construct corporations which are just beyond the trillion dollar range, and that's where OpenAI is focused.
Same interview, on OpenAI's trajectory.
We're creating an elite group of people that are going to control AI and I really strongly believe that that type of power can corrupt.
Same interview, on concentration of AI capability.
I think it's quite obviously anticompetitive and very likely to create a very walled garden for these corporations.
Same interview, responding to tech executives' calls for federal AI licensing.
But Bittensor is not us ... It is simultaneously open but also anonymous, and privacy preserving so people can ... interface with compute and intelligence in an anonymous way.
Same interview, on why regulation of the foundation would not stop the protocol.
With subnets live he talks to builders and investors: Bitcoin's supercomputer via incentives, TAO as light and dark, the Elysium outcome as the real risk, and subnet design as intelligence evaluating intelligence up the ladder toward AGI 032 033 034 036. Proof becomes subnet results 037.
supercomputerShift From protocol philosophy to subnet craft and investor-facing lines; markets return as the way subnets are judged.
It's the most amazing feeling as a builder. Frankenstein, it's alive.
Fortune feature on VC interest in TAO, reacting to centralized AI products validating his thesis; Fortune printed the two sentences as one quote with a dash.
People will come and they will go, and if they buy it because it's a memecoin, they won't be here later. That's fine, because we want people that actually know what we're doing.
Same feature, answering the charge that TAO rides AI hype.
How did we create a supercomputer that is bigger than any government or corporation can create with a centralized entity? The answer is Bitcoin used incentive mechanisms.
Same feature, on the inspiration for Bittensor; Fortune reprinted this line in December 2025.
It's this constant churn, a battle between the light and the dark. That's a lot of why we like Taoist symbolism.
Same feature, describing miners and validators and the origin of the TAO name.
The real issue is not that the AIs are going to take over the world ... It's not the iRobot outcome, it's the Elysium outcome, where humanity and the world is so centralized in terms of power, metastasized by artificial intelligence, that all of us are just but slaves effectively to these massive companies.
Same feature, on what he thinks the actual AI risk is.
This is hard problem in creating a Bittensor subnet and why I always push teams to get very specific and metric focused. You are the intelligence creating the evaluation heuristic for intelligence. The best way to do it is use machine intelligence itself.
Post on X advising subnet builders on incentive design.
The seesaw back and forth between using the asymmetry to create intelligence and then using that intelligence to evaluate even better intelligence is what takes us further up the ladder towards AGI and beyond.
Same post, invoking the P versus NP asymmetry between checking and solving.
Subnet 32 on Bittensor beats all AI detection methods.
Post on X promoting a subnet's benchmark result.
The shorthand is now a Bitcoin-like mining mechanism applied to AI that organises compute, models and data into a fair market 039 040. He steps down as CEO but keeps the same calls 045, then the Covenant exit turns the language personal and defensive: a brother's betrayal, subnets as headless commodities, commitment as time plus stake, and a claim to be the most decentralised AI protocol in existence 050 046 052 053.
miningShift Market talk becomes token-market talk (alpha, dTao, buys and sells); the founder speaks as a holder who reserves the right to trade.
the premined Bittensor clones are on their way lads. give them nothing.
Post on X as competing decentralized AI tokens launched.
Bittensor is an open protocol that applies a Bitcoin-like mining mechanism to AI: we use programmable economic incentives to organize decentralized computing power, models, data, and applications into a fair market.
Interview during his first trip to China, answering a request to introduce Bittensor; published in Chinese and read here in translation.
AI is essentially a computational problem; Bitcoin has already proven that 'incentives + competition' can effectively coordinate distributed resources.
Same interview, on why a Bitcoin design fits AI.
I believe the only impact of the halving on Bittensor is that supply will tighten. However, this will not affect the basic incentive mechanism of the network, which still has significant economic incentives to encourage developers to build on the platform.
Same interview, asked about the December 2025 halving.
It mainly comes from selling inference, selling computing power, selling AutoML (automated machine learning), and selling signals to prediction markets.
Same interview, asked where protocol revenue comes from.
This is a brand new and fair economic platform, where Chinese engineers, builders, and miners can make truly productive contributions.
Same interview, on the message he wanted to bring to Chinese builders.
The headline I most want to see ... is: we have brought this technology to 'millions' of users and truly provided open intelligent services to the world, with the network continuously expanding and operating sustainably.
Same interview, asked for his five-year headline.
Not much will change, I'm still going to be in the same calls and do the same things and continue to write chain code, and make subnets, and run novelty search and call people scammers and suggest changes — just not with the same legal authority (and that's important )
Message to the community on stepping down as Opentensor CEO, as reported by SimplyTao; the original post was not reached.
This will prove to birth the first subnets on Bittensor that run headless and as true commodities.
Post on X the night Covenant AI announced it was leaving subnets 3, 39 and 81, as quoted by The Block.
The outcome of this eventful evening is that Bittensor will invent lock-based subnet ownership.
Same night, same thread, as quoted by The Block.
I do not have the ability to suspend emissions. What i did do is sell some of my alpha holdings on his three subnets, because they were not running, and were on near 100% burn code. ... I don't have any privilege beyond what normal TAO holders have.
Post on X rebutting Samuel Dare's claim that he suspended Covenant's emissions.
Visibility is impossible to avoid in my position. I reserve my right to buy and sell tokens which is what underpins the entire system of dTao.
Same post, on his token sales being read as punitive.
Someone I considered a brother, Samuel Dare, took actions with the intention to cause maximum pain
Opening of his long-form X article 'Response to Covenant'.
Bittensor was built to be resistant to the pitfalls of human error; greed, selfishness, and other sins.
Same article, on why the protocol is designed the way it is.
commitment, which Is measured as the time remaining before tokens can move; time + stake.
Same article, defining the Locked Stake mechanism he proposed as the fix.
we are by far the most decentralized AI protocol In existence. ... Next time we train 1 trillion. - const
Closing of the same article, after Covenant's 72B model training run on subnet 3.
He concedes the point his critics made: Bittensor is not decentralised in the way Bitcoin is, has leaders at its helm, and has chosen speed at the cost of centralisation, while restating the vision as two substrates, mutable AI plus incentive on immovable Bitcoin, with decentralisation as the crowning jewel roughly eighteen months out 054 057 055 059. Show notes for his Affine-era podcast call dTAO the internal machinery of capitalism 060.
two substratesShift From denying control to owning it as a temporary design choice; Bitcoin returns as the ethical anchor rather than the supercomputer.
Bittensor is currently not a decentralized protocol in the way Bitcoin is. It can be, and it will be, but it isn't yet.
Opening of his long-form X article on decentralization, two months after the Covenant dispute.
Bittensor's vision is the unification of two technologies: the fast-paced, mutable substrate of AI plus incentive, and the ethically rooted, immovable substrate of Bitcoin.
Same article, restating the vision.
Bittensor is built on one of the most profound, well-built technologies ever created: Substrate, Polkadot tooling for the creation of blockchains, which is upgradable, connectable, hard-packed with an SDK and cryptography to make it resistant to any group while remaining mutable, extremely fast-running, and decentralized.
Same article, on the chain stack; this passage was quoted by Robert Scoble the same day.
Instead, Bittensor has leaders. There are people at its helm: me, two other engineers, and a very tight-knit group.
Same article, admitting who currently steers the core economic layer.
There is no point in decentralizing a system that doesn't yet deserve decentralization. ... We have decided to move fast and iterate, update, and develop quickly, at the cost of remaining centralized.
Same article, defending the current centralization as a deliberate speed tradeoff.
Bittensor is not complete until thoroughly decentralized, and decentralization is the crowning jewel of what we are building.
Same article, giving roughly eighteen months as the horizon for full decentralization.
the internal machinery of capitalism
Phrase attributed to him in the show notes of This Week in Startups episode 2326, describing how dTAO ranks subnets; transcript not reached.
A brand strategy skeleton turned on the speaker: not what Bittensor's brand is, but what this voice's Bittensor is. Slots with no evidence are left out rather than filled.
What they say it is for, the world it builds, what it stands for.
A peer-to-peer market in which intelligence is priced by other intelligence, built so people can work on very large AI problems without sitting under one organisation that holds the keys.
As with other commodities, markets could help us efficiently produce machine intelligence. We propose a market where intelligence is priced by other intelligence systems peer-to-peer across the internet.
We're building a way of aligning individuals and compute so that we can work on these incredibly large problems without being underneath a particular organization which holds the keys.
Inspired by the efficiency of financial markets, we propose that a market system can be used to effectively produce machine intelligence.
This software is designed to run an intermodel machine intelligence benchmark which measures the performance of other learners via informational significance.
Early on, a neural internet or global brain for AIs; later, the unification of AI plus incentive with Bitcoin's immovable base, serving millions of users and ending fully decentralised.
We like to say that we're building the neural internet, because for the first time ever, we're successfully doing machine learning across thousands of individuals ... We're like the Marshall McEwan global brain, but for AIs.
The headline I most want to see ... is: we have brought this technology to 'millions' of users and truly provided open intelligent services to the world, with the network continuously expanding and operating sustainably.
Bittensor's vision is the unification of two technologies: the fast-paced, mutable substrate of AI plus incentive, and the ethically rooted, immovable substrate of Bitcoin.
Bittensor is not complete until thoroughly decentralized, and decentralization is the crowning jewel of what we are building.
Finney is about BitTensor's freedom in Web3. Freedom from centralized control.
The seesaw back and forth between using the asymmetry to create intelligence and then using that intelligence to evaluate even better intelligence is what takes us further up the ladder towards AGI and beyond.
No pre-mine, control diffused across humanity, resistance to regulators and to greed, and a stated refusal to give premined copies anything; in 2026 he pairs these with an admission that speed has been chosen over decentralisation for now.
We didn't do a pre-mine because we wanted every iota of value to flow through the computational aspect of BitTensor.
If we want something to be aligned with us, we need to make sure that it has control over it and that control is diffused amongst humanity.
If we don't make sure BitTensor lifts off ... into a decentralized system where regulators really can't interfere, we're compromised.
Bittensor was built to be resistant to the pitfalls of human error; greed, selfishness, and other sins.
the premined Bittensor clones are on their way lads. give them nothing.
There is no point in decentralizing a system that doesn't yet deserve decentralization. ... We have decided to move fast and iterate, update, and develop quickly, at the cost of remaining centralized.
In the core of the technology is competition. But what really will take us up to the high levels of performance will be collaboration in and around that competition.
Visibility is impossible to avoid in my position. I reserve my right to buy and sell tokens which is what underpins the entire system of dTao.
Bittensor is currently not a decentralized protocol in the way Bitcoin is. It can be, and it will be, but it isn't yet.
Bittensor is not complete until thoroughly decentralized, and decentralization is the crowning jewel of what we are building.
What is broken that Bittensor answers, in their telling.
The danger is not robots taking over but power metastasising into a few trillion-dollar companies and an elite that controls AI, the Elysium outcome.
The real issue is not that the AIs are going to take over the world ... It's not the iRobot outcome, it's the Elysium outcome, where humanity and the world is so centralized in terms of power, metastasized by artificial intelligence, that all of us are just but slaves effectively to these massive companies.
We're creating an elite group of people that are going to control AI and I really strongly believe that that type of power can corrupt.
What dystopian future are we fighting against?
If we want something to be aligned with us, we need to make sure that it has control over it and that control is diffused amongst humanity.
Intelligence has become a commodity with no market to price it; benchmarks and peer review are inefficient and biased; nobody shares billion-dollar models without an incentive baked in first.
A new commodity needs a new type of market.
There must be a more efficient and objective method to evaluate AI model performance.
Machine intelligence, this stuff is billion-dollar projects. So if we want to connect people together, we have to bake in the incentive first, the value first.
Corporate AI and licensing regimes pull toward walled gardens; if the network does not lift off beyond regulators it is compromised; and in 2026 he names himself and a tight group as the remaining central point.
There's a strong gravitational pull for this technology to construct corporations which are just beyond the trillion dollar range, and that's where OpenAI is focused.
If we don't make sure BitTensor lifts off ... into a decentralized system where regulators really can't interfere, we're compromised.
Instead, Bittensor has leaders. There are people at its helm: me, two other engineers, and a very tight-knit group.
We're creating an elite group of people that are going to control AI and I really strongly believe that that type of power can corrupt.
The tailwinds they cite.
Bitcoin proved incentives can align strangers at internet scale, and calls for federal AI licensing strengthen the case for an open protocol; the Covenant exit became the occasion for lock-based ownership.
Bitcoin is aligning individuals through this incentive mechanism. It's an alignment technology, which then allows us to scale to this super internet scale, peer-to-peer, open source, decentralized.
I think it's quite obviously anticompetitive and very likely to create a very walled garden for these corporations.
The outcome of this eventful evening is that Bittensor will invent lock-based subnet ownership.
commitment, which Is measured as the time remaining before tokens can move; time + stake.
Who they talk to and what they tell them.
He speaks to builders and miners, telling subnet teams to be metric-focused, welcoming Chinese engineers to a fair platform, and telling speculators that memecoin buyers will not stay.
This is hard problem in creating a Bittensor subnet and why I always push teams to get very specific and metric focused. You are the intelligence creating the evaluation heuristic for intelligence. The best way to do it is use machine intelligence itself.
This is a brand new and fair economic platform, where Chinese engineers, builders, and miners can make truly productive contributions.
People will come and they will go, and if they buy it because it's a memecoin, they won't be here later. That's fine, because we want people that actually know what we're doing.
There's the foundation, which is a non-profit guiding the incentive mechanism. Then there's miners running infrastructure. And clients of the system. All three groups have power in BitTensor.
I believe the only impact of the halving on Bittensor is that supply will tighten. However, this will not affect the basic incentive mechanism of the network, which still has significant economic incentives to encourage developers to build on the platform.
Delegation is effectively a liquid democracy where individuals can build organizations and move TAO around.
What they set it against.
OpenAI as a company built on one model and pulled toward the trillion-dollar corporation, premined clones of Bittensor, and, from April 2026, Samuel Dare and Covenant AI; he also argues against decentralising too early.
What OpenAI did, is they built a company on top of that core model.
There's a strong gravitational pull for this technology to construct corporations which are just beyond the trillion dollar range, and that's where OpenAI is focused.
the premined Bittensor clones are on their way lads. give them nothing.
I do not have the ability to suspend emissions. What i did do is sell some of my alpha holdings on his three subnets, because they were not running, and were on near 100% burn code. ... I don't have any privilege beyond what normal TAO holders have.
Someone I considered a brother, Samuel Dare, took actions with the intention to cause maximum pain
There is no point in decentralizing a system that doesn't yet deserve decentralization. ... We have decided to move fast and iterate, update, and develop quickly, at the cost of remaining centralized.
We're building a way of aligning individuals and compute so that we can work on these incredibly large problems without being underneath a particular organization which holds the keys.
I think it's quite obviously anticompetitive and very likely to create a very walled garden for these corporations.
The real issue is not that the AIs are going to take over the world ... It's not the iRobot outcome, it's the Elysium outcome, where humanity and the world is so centralized in terms of power, metastasized by artificial intelligence, that all of us are just but slaves effectively to these massive companies.
Category frame, differentiator, reason to believe.
A market for intelligence and a commodity market first; by 2025 the shorthand is a Bitcoin-like mining mechanism applied to AI, and by 2026 subnets are headless commodities ranked by the machinery of capitalism.
As with other commodities, markets could help us efficiently produce machine intelligence. We propose a market where intelligence is priced by other intelligence systems peer-to-peer across the internet.
This is such an expensive, but also such a valuable commodity, we require that commodification layer to allow people to really share this thing.
Bittensor is an open protocol that applies a Bitcoin-like mining mechanism to AI: we use programmable economic incentives to organize decentralized computing power, models, data, and applications into a fair market.
This will prove to birth the first subnets on Bittensor that run headless and as true commodities.
the internal machinery of capitalism
A new commodity needs a new type of market.
The result is a collectively run intelligence market that continually produces newly trained models and pays contributors who create information theoretic value.
We like to say that we're building the neural internet, because for the first time ever, we're successfully doing machine learning across thousands of individuals ... We're like the Marshall McEwan global brain, but for AIs.
Inspired by the efficiency of financial markets, we propose that a market system can be used to effectively produce machine intelligence.
Competition at the core with collaboration around it, intelligence evaluated by machine intelligence, open and anonymous access, and a claim to be the most decentralised AI protocol in existence.
In the core of the technology is competition. But what really will take us up to the high levels of performance will be collaboration in and around that competition.
This is hard problem in creating a Bittensor subnet and why I always push teams to get very specific and metric focused. You are the intelligence creating the evaluation heuristic for intelligence. The best way to do it is use machine intelligence itself.
But Bittensor is not us ... It is simultaneously open but also anonymous, and privacy preserving so people can ... interface with compute and intelligence in an anonymous way.
we are by far the most decentralized AI protocol In existence. ... Next time we train 1 trillion. - const
If we build that into a market, we can leverage the innovation of this open, decentralized, borderless worldwide community.
Bittensor was built to be resistant to the pitfalls of human error; greed, selfishness, and other sins.
Bitcoin built the largest supercomputer on earth with incentives alone; the same mechanism, plus the checking-versus-solving asymmetry and a hardened Substrate chain, can produce intelligence.
Bitcoin created the largest supercomputer in the world, as ... building a censorship-resistant technology.
How did we create a supercomputer that is bigger than any government or corporation can create with a centralized entity? The answer is Bitcoin used incentive mechanisms.
AI is essentially a computational problem; Bitcoin has already proven that 'incentives + competition' can effectively coordinate distributed resources.
The seesaw back and forth between using the asymmetry to create intelligence and then using that intelligence to evaluate even better intelligence is what takes us further up the ladder towards AGI and beyond.
Bittensor is built on one of the most profound, well-built technologies ever created: Substrate, Polkadot tooling for the creation of blockchains, which is upgradable, connectable, hard-packed with an SDK and cryptography to make it resistant to any group while remaining mutable, extremely fast-running, and decentralized.
Bitcoin is aligning individuals through this incentive mechanism. It's an alignment technology, which then allows us to scale to this super internet scale, peer-to-peer, open source, decentralized.
Machine intelligence, this stuff is billion-dollar projects. So if we want to connect people together, we have to bake in the incentive first, the value first.
The benefit they claim, the feeling they sell, what joining says about you.
Continuously produced, newly trained models with contributors paid for information-theoretic value; revenue from inference, compute, AutoML and prediction signals.
The result is a collectively run intelligence market that continually produces newly trained models and pays contributors who create information theoretic value.
If we build that into a market, we can leverage the innovation of this open, decentralized, borderless worldwide community.
It mainly comes from selling inference, selling computing power, selling AutoML (automated machine learning), and selling signals to prediction markets.
Subnetworks are basically just running multiple Nakamoto consensus across the same chain so we can scale horizontally.
The builder's thrill of seeing the thing come alive.
It's the most amazing feeling as a builder. Frankenstein, it's alive.
Joining marks you as someone who knows what the project is doing, or as a productive contributor on a fair platform, not a memecoin buyer.
People will come and they will go, and if they buy it because it's a memecoin, they won't be here later. That's fine, because we want people that actually know what we're doing.
This is a brand new and fair economic platform, where Chinese engineers, builders, and miners can make truly productive contributions.
The line they repeat and the proof they point to.
The line moves from 'a market where intelligence is priced by other intelligence' to 'the neural internet' to 'freedom from centralized control' to 'a Bitcoin-like mining mechanism to AI' and finally 'the unification of AI plus incentive and Bitcoin'.
As with other commodities, markets could help us efficiently produce machine intelligence. We propose a market where intelligence is priced by other intelligence systems peer-to-peer across the internet.
We like to say that we're building the neural internet, because for the first time ever, we're successfully doing machine learning across thousands of individuals ... We're like the Marshall McEwan global brain, but for AIs.
Finney is about BitTensor's freedom in Web3. Freedom from centralized control.
How did we create a supercomputer that is bigger than any government or corporation can create with a centralized entity? The answer is Bitcoin used incentive mechanisms.
Bittensor is an open protocol that applies a Bitcoin-like mining mechanism to AI: we use programmable economic incentives to organize decentralized computing power, models, data, and applications into a fair market.
Bittensor's vision is the unification of two technologies: the fast-paced, mutable substrate of AI plus incentive, and the ethically rooted, immovable substrate of Bitcoin.
No pre-mine, the 18 percent built-in take, Bitcoin's supercomputer, subnet benchmark wins, a 72B training run with a promise of one trillion next, and the Substrate stack.
We didn't do a pre-mine because we wanted every iota of value to flow through the computational aspect of BitTensor.
It's also a way for us as a foundation to make revenue ... we've built in a flat rate of 18%.
Subnet 32 on Bittensor beats all AI detection methods.
we are by far the most decentralized AI protocol In existence. ... Next time we train 1 trillion. - const
Bittensor is built on one of the most profound, well-built technologies ever created: Substrate, Polkadot tooling for the creation of blockchains, which is upgradable, connectable, hard-packed with an SDK and cryptography to make it resistant to any group while remaining mutable, extremely fast-running, and decentralized.
I believe the only impact of the halving on Bittensor is that supply will tighten. However, this will not affect the basic incentive mechanism of the network, which still has significant economic incentives to encourage developers to build on the platform.
Bitcoin created the largest supercomputer in the world, as ... building a censorship-resistant technology.
It mainly comes from selling inference, selling computing power, selling AutoML (automated machine learning), and selling signals to prediction markets.
Register more than claim.
Terse, combative and self-aware: two-word posts, 'give them nothing', calling people scammers, 'He betrayed us all', signing off '- const'.
Bittensor revolution
the premined Bittensor clones are on their way lads. give them nothing.
Not much will change, I'm still going to be in the same calls and do the same things and continue to write chain code, and make subnets, and run novelty search and call people scammers and suggest changes — just not with the same legal authority (and that's important )
Someone I considered a brother, Samuel Dare, took actions with the intention to cause maximum pain
It's the most amazing feeling as a builder. Frankenstein, it's alive.
What dystopian future are we fighting against?
It's this constant churn, a battle between the light and the dark. That's a lot of why we like Taoist symbolism.
we are by far the most decentralized AI protocol In existence. ... Next time we train 1 trillion. - const
How they describe their own relation to it.
Founder and foundation head who insists 'Bittensor is not us', then a CEO who steps down while keeping the same calls and chain code, then a holder who reserves the right to trade alpha, and finally a self-described leader at the helm of a not-yet-decentralised protocol.
There's the foundation, which is a non-profit guiding the incentive mechanism. Then there's miners running infrastructure. And clients of the system. All three groups have power in BitTensor.
It's also a way for us as a foundation to make revenue ... we've built in a flat rate of 18%.
But Bittensor is not us ... It is simultaneously open but also anonymous, and privacy preserving so people can ... interface with compute and intelligence in an anonymous way.
Not much will change, I'm still going to be in the same calls and do the same things and continue to write chain code, and make subnets, and run novelty search and call people scammers and suggest changes — just not with the same legal authority (and that's important )
I do not have the ability to suspend emissions. What i did do is sell some of my alpha holdings on his three subnets, because they were not running, and were on near 100% burn code. ... I don't have any privilege beyond what normal TAO holders have.
Visibility is impossible to avoid in my position. I reserve my right to buy and sell tokens which is what underpins the entire system of dTao.
Bittensor is currently not a decentralized protocol in the way Bitcoin is. It can be, and it will be, but it isn't yet.
Instead, Bittensor has leaders. There are people at its helm: me, two other engineers, and a very tight-knit group.
What they call things, and refuse to call them.
He names things as he builds them: intermodel benchmark, commodification layer, subnetworks as parallel Nakamoto consensus, delegation as liquid democracy, 'Bittensor revolution', TAO as light and dark, lock-based ownership, commitment as time plus stake, headless commodities.
This software is designed to run an intermodel machine intelligence benchmark which measures the performance of other learners via informational significance.
This is such an expensive, but also such a valuable commodity, we require that commodification layer to allow people to really share this thing.
Subnetworks are basically just running multiple Nakamoto consensus across the same chain so we can scale horizontally.
Delegation is effectively a liquid democracy where individuals can build organizations and move TAO around.
Bittensor revolution
It's this constant churn, a battle between the light and the dark. That's a lot of why we like Taoist symbolism.
The outcome of this eventful evening is that Bittensor will invent lock-based subnet ownership.
commitment, which Is measured as the time remaining before tokens can move; time + stake.
This will prove to birth the first subnets on Bittensor that run headless and as true commodities.
the internal machinery of capitalism
Who they agree with, borrow from, or argue against, with the quote that shows it.
Public appearances on the record, whether or not a quote was taken.
Something wrong or missing? Quotes are corrected as they arrive; a misattribution is removed first and argued after. How to file one · GitHub issue