# What Const has been saying: the tweet archive, read for the keynote

Source: all 3,085 tweets from @const_reborn, 16 Feb 2023 to 11 Sep 2026 (`docs/sources/tweets/const_reborn-tweets.jsonl`). I read every one. The 617 pure retweets were read but given low weight. Reads against `docs/project-context.md` and `docs/thesis.md`.

How to read this file:

- Every quote is his exact text. Nothing inside quotation marks is paraphrased. Long tweets use the full "note tweet" text, not the truncated preview.
- In the one-line quotes, a " / " marks a line break in the original tweet.
- Where a tweet quoted someone else, X appends the quoted tweet's URL to the text. I removed that trailing URL and wrote "(quoting @user)" instead. Nothing else was changed.
- "Tweets" counts are keyword matches over the 2,468 non-retweet tweets. They are approximate. They show relative weight, not exact totals.
- Five of his highest-engagement 2026 posts are X Articles (long-form). The API returns only the article link, not the body. They are flagged where they matter. Someone needs to pull them by hand: 11 Apr 2026 (2,385 likes), 19 Jun, 20 Jun, 21 Jun (1,088 likes), 23 Jun 2026.

---

## 1. What you have been saying

Fifteen themes, ordered by how strongly they recur across the four years. The first five are the spine of the account. Themes 13 to 15 are newer or narrower but matter for this talk.

### Theme 1. Incentives are the machine. Money is an energy function.

Summary: Your single most repeated idea. A token reward is the same kind of object as a loss function in machine learning: it shapes the world the way a loss shapes weights. Bittensor is a computer that runs on money instead of electricity.
Years: 2023, 2024, 2025, 2026. Tweets: about 230 (the largest theme by far). Peak year 2025.

> the token is the objective function
>
> — 14 Dec 2024 · 185 likes · original · [x.com/const_reborn/status/1868075527371980869](https://x.com/const_reborn/status/1868075527371980869)

> @knowerofmarkets Most AI researchers miss the isomorphism between ML objective functions and incentive mechanisms.
>
> — 19 Nov 2023 · 66 likes · reply · [x.com/const_reborn/status/1726210097003536728](https://x.com/const_reborn/status/1726210097003536728)

> Common misconception of crypto people is to think the AI world needs their security guarantees, ZK proofs, “AI in my smart contracts”. This myopic thinking.
>
> Common limitation of AI people is thinking crypto is all a scam and not noticing that we just invented the most expressive language for thermodynamics ever, and that we can literally write energy functions and loss functions with money now.
>
> — 30 Jan 2024 · 86 likes · original · [x.com/const_reborn/status/1752327627539415527](https://x.com/const_reborn/status/1752327627539415527)

> Thermodynamic computing: energy computes 
>
> Machine Learning: error computes 
>
> Incentive mechanisms: markets compute
>
> — 30 Jun 2025 · 119 likes · original · [x.com/const_reborn/status/1939620057342152742](https://x.com/const_reborn/status/1939620057342152742)

> This is why incentive programming is so fascinating. You don’t actually solve any technical problems, you just structure the rewards so other solve it for you. It’s inverse, upside down, inside out engineering.
>
> — 30 Jan 2025 · 117 likes · reply · [x.com/const_reborn/status/1885108699926045138](https://x.com/const_reborn/status/1885108699926045138)

> What's Bittensor? Its a programmable incentive computer which you can direct to solve any well defined problem faster, and more "impressively", than was ever possible, by individual, corporation, or government. It's directed organization on crack. Enabled by humanities innovation in digital money, new programmable value, which is the sinew of how humans, and truthfully, everything organizes.
>
> — 13 Nov 2025 · 244 likes · reply · [x.com/const_reborn/status/1989007120059019395](https://x.com/const_reborn/status/1989007120059019395)

> AI can judge anything. Money can move anything.
>
> — 24 Apr 2026 · 253 likes · original · [x.com/const_reborn/status/2047599733544263777](https://x.com/const_reborn/status/2047599733544263777)

### Theme 2. The miners are the point. The swarm beats every team.

Summary: You keep returning to one surprise: anonymous, permissionless miners out-build any hired team, including yours. The mechanism that finds them is the asset, not the people it finds.
Years: 2023, 2024, 2025, 2026. Tweets: about 180. Peak 2025.

> No matter how cracked your team is, you’ll never beat the machine that picks cracked teams.
>
> — 13 Feb 2025 · 302 likes · original · [x.com/const_reborn/status/1890010448726254035](https://x.com/const_reborn/status/1890010448726254035)

> We haven’t found a problem on Bittensor where the miners haven’t surprised us by their quality.
>
> Harness the miners.
>
> — 2 May 2025 · 293 likes · original · [x.com/const_reborn/status/1918162299397017939](https://x.com/const_reborn/status/1918162299397017939)

> The truth is … I can’t mine anymore. I’ve been solidly out competed in Bittensor about 10 times over. 
>
> And that’s a testament to the power of pure permissionless markets, not my incompetence. 
>
> The beauty? When you mine Tao and get outcompeted, it just means smarter people picked up your pick axe and started digging for you.
>
> — 1 Apr 2025 · 240 likes · quote · [x.com/const_reborn/status/1906887497205899292](https://x.com/const_reborn/status/1906887497205899292)

> Musk can probably poach our miners. 
> But he can't poach the mechanism that found them in the first place.
>
> — 16 Jul 2025 · 233 likes · original · [x.com/const_reborn/status/1945521483972849973](https://x.com/const_reborn/status/1945521483972849973)

> Some of Bittensors best miners haven’t even finished high school.
>
> — 29 Jun 2025 · 229 likes · quote (quoting @polynoamial) · [x.com/const_reborn/status/1939272700498686084](https://x.com/const_reborn/status/1939272700498686084)

> Other teams are cracked, they have great tech, they ship. 
>
> They will never ship faster than the aligned swarm.
>
> — 23 Nov 2024 · 45 likes · reply · [x.com/const_reborn/status/1860370840631542181](https://x.com/const_reborn/status/1860370840631542181)

Note: the Mar 2026 tweet about dropouts, kids and outcasts (642 likes) says the same thing with more color, but its last sentence is a slur on women. Do not use it on stage. See Tensions.

### Theme 3. Ownership, not access. "If you don't own it, you won't control it."

Summary: Decentralized AI is not about training together or getting free API calls. It is about who owns the thing and whether anyone can switch it off. This is the oldest idea in the archive (April 2023) and it is still your answer to AI risk in 2026.
Years: 2023, 2024, 2025, 2026. Tweets: about 120. Rising: 40 in 2026 so far.

> Decentralized AI isn’t about training collectively. 
>
> That’s cool, but 99.9% of people can’t train models.
>
> Instead, Decentralized AI is about censorship resistance, access, and ownership. 
>
> We should collectively own AI, not just contribute to it.
>
> — 5 Apr 2023 · 203 likes · original · [x.com/const_reborn/status/1643555971585789952](https://x.com/const_reborn/status/1643555971585789952)

> If you’re not asking “who really owns this?” you’re not working on decentralized AI. 
>
> If you’re not asking “can someone turn this off?” you’re not working on decentralized AI. 
>
> If you own the system you make, it’s not decentralized AI.
>
> — 5 Apr 2023 · 26 likes · reply · [x.com/const_reborn/status/1643557937477025792](https://x.com/const_reborn/status/1643557937477025792)

> Open AI "democracy" will be patronizing conversations with ChatGPT for the purpose of people being "heard", then behind the scenes everything decided by the same people. 
>
> If you don't own it, you won't control it.
> You won't be heard either.
>
> https://openai.com/blog/democratic-inputs-to-ai
>
> — 26 May 2023 · 55 likes · original · [x.com/const_reborn/status/1662135209356296192](https://x.com/const_reborn/status/1662135209356296192)

> @leopoldasch @ilyasut OpenAI will give you platitudes and fancy mathematics in the name of alignment. 
>
> But they won’t give you ownership.
>
> Silicon Valley only believes in it’s own elitist divinity. 
>
> It won’t share power.
>
> — 6 Jul 2023 · 57 likes · reply · [x.com/const_reborn/status/1676771058869760001](https://x.com/const_reborn/status/1676771058869760001)

> The real risk of AI was never robots with a vendetta. It’s that all the capital ownership concentrates down to a small elite set of technocrats in SF. 
>
> These layoffs are just the beginning. AI is a centralizing force where its biggest companies dream total monopolization with less than 3000 people. 
>
> Where even those 3000 are paid in dilutive fiat and dilutive equity and you’re not allowed to enter until way after fiat-capital takes the lion share. 
>
> The answer is, i believe, to copy Bitcoin. With AI nested in a 21M non-dilutive fair launch, and POW that pays people in visceral ownership. 
>
> Because only ownership will tie us to a future dominated by synthetic intelligence.
>
> — 30 Oct 2025 · 465 likes · quote · [x.com/const_reborn/status/1983941275876520051](https://x.com/const_reborn/status/1983941275876520051)

> The best teachers will be LLMs. The best feedback will be monetary. The best models will train competitively. The best AI company will be co-owned. The best software will be open source. And there is only one way to get all of it.
>
> — 13 Jun 2026 · 788 likes · original · [x.com/const_reborn/status/2065914716589850671](https://x.com/const_reborn/status/2065914716589850671)

### Theme 4. The real fear is a small group of humans, not the machine.

Summary: You have said since 2023 that "AI takes over" is fiction that hides the real outcome: a few unelected people program the agenda in. In 2025 and 2026 the language got sharper: Elysium, Gaza, "enslaved", "intelligence fascism". This is the core message of the talk and you have been building it in public for three years.
Years: 2023, 2025, 2026 (almost silent on it in 2024). Tweets: about 55, plus 16 on UBI and Worldcoin.

> The sci future where AI programs it’s own agenda to dominate humanity is just that
>
> Fiction.
>
> The narrative also covers for and promotes the likely outcome: where that agenda is programmed-in by a small group of unelected davos members. 
>
> Decentralizing it’s control is the antidote to that
>
> — 22 Sep 2023 · 36 likes · reply · [x.com/const_reborn/status/1705370170829541853](https://x.com/const_reborn/status/1705370170829541853)

> @ylecun @erhartford @JayantParashar6 P(AI codes its own bad objectives) < P(AI has objectives coded in by totalitarian group)
>
> — 1 Nov 2023 · 11 likes · reply · [x.com/const_reborn/status/1719712940125724855](https://x.com/const_reborn/status/1719712940125724855)

> Dwarkesh isn’t wrong -- he’s just not serious enough.
>
> This isn’t just the beginning of an unequal economic game.
>
> The invention of AI, and by extension robotics, leads to runaway power concentration. We are eroding the economic weight of the individual human, and therefore the decentralization of power that derives from flesh.
>
> Without carefully constructed ownership and openness layers, humanity doesn’t end up unequal -- it ends up enslaved.
>
> — 7 Jan 2026 · 153 likes · quote · [x.com/const_reborn/status/2009032093729477050](https://x.com/const_reborn/status/2009032093729477050)

> We’re not ready for a world where armies are made of machines, not men — where human power is so eroded that the AI cabal can control the world from a laptop.
>
> — 5 Aug 2025 · 175 likes · original · [x.com/const_reborn/status/1952710606231949733](https://x.com/const_reborn/status/1952710606231949733)

> Warns about everything except the obvious thing: that hyper wealth and power concentration is the lever for true separation and then abuse by machine on humanity. 
>
> We become Gaza.
>
> — 10 Jun 2026 · 135 likes · reply · [x.com/const_reborn/status/2064834911492563380](https://x.com/const_reborn/status/2064834911492563380)

> @AnthropicAI And just like that we collectively saw the future of inequality
>
> — 13 Jun 2026 · 2,588 likes · reply · [x.com/const_reborn/status/2065606907855831301](https://x.com/const_reborn/status/2065606907855831301)

On UBI, which is the establishment's answer to the same fear:

> UBI is what they will offer you in return for control, it’s a bribe, a contract you will sign to solidify your agreement with the loss of power. 
>
> Humans know, we all viscerally know that freedom, sovereignty and ownership are paramount and indeed requirements of real wealth.
>
> — 6 Jun 2025 · 127 likes · quote (quoting @haider1) · [x.com/const_reborn/status/1930927421651296503](https://x.com/const_reborn/status/1930927421651296503)

> That’s why the answer to the impending arc of technological centralization and its power is not “UBI”, it’s not to trade our power and freedom for a fiat drip — its to ingrain ourselves into the visceral control of that technology. 
>
> This is the symbiosis of man and machine.
>
> — 6 Jun 2025 · 68 likes · reply · [x.com/const_reborn/status/1930929295620239856](https://x.com/const_reborn/status/1930929295620239856)

### Theme 5. Bitcoin is the shield. Bittensor stands behind it.

Summary: You are protective of Bitcoin. You call it the freedom technology that lets everything else in crypto exist, and you refuse to rank TAO above it. The founding claim is that Bitcoin broke the state's monopoly on money and Bittensor breaks the coming monopoly on intelligence.
Years: 2023, 2024, 2025, 2026. Tweets: about 97.

> Re Biden's executive order:
>
> Bitcoin broke the gov's monopoly on currency.
> Bittensor stops the gov's monopolization of AI.
>
> #TAO
>
> — 30 Oct 2023 · 444 likes · original · [x.com/const_reborn/status/1719092725402800602](https://x.com/const_reborn/status/1719092725402800602)

> It was essential that we started with Bitcoin. Hard values, clear vision. However, subnets will be our Ethereum moment. Yuma consensus unleashed. The first ever abstraction layer on digital commodity markets. LFG
>
> — 4 Sep 2023 · 310 likes · original · [x.com/const_reborn/status/1698844838110593262](https://x.com/const_reborn/status/1698844838110593262)

> It’s REALLY important to remember that Tao is also nothing without BTC. The whole industry in-fact is dead without BTC. They solve different problems and i hope will help each other. I’ll die on that hill.
>
> — 23 Mar 2026 · 1,019 likes · quote (quoting @Jason) · [x.com/const_reborn/status/2035998840281612313](https://x.com/const_reborn/status/2035998840281612313)

> Bitcoins value is derived from its censorship resistance. You can’t take it, you can always send it, they can never print it away. Thus it holds the foundation of a fair non fiat financial system. It is slow expensive and toxic, but there really is no token that will ever be able to match the purity of its launch , the anonymity of its founder and game theoretic shelling point of its brand / memetic nature. Tao needs Bitcoin to survive because as an industry it is the invincible penetrating spear that has forced governments to allow experimentation with the coding language of money. Bittensor, Ethereum, all others are thus enabled by that bulwark and simultaneously relieved of needing to focus on that censorship resistance. Saying TAO is greater than Bitcoin is like saying Telegram is better than TCP. It’s a silly statement because obviously Telegram is enabled by TCP. We litterally stand ontop of the freedom shield that is Bitcoin. This is how i see it and have always seen it.
>
> — 23 Mar 2026 · 358 likes · reply · [x.com/const_reborn/status/2036033516605981142](https://x.com/const_reborn/status/2036033516605981142)

> Id love to be more decentralized than Bitcoin but it simply isn't the case. Indeed, there is simply non-comparison. Bitcoin is the decentralized shield that protects the whole crypto space. We are like its little brother making sand castles behind the wall.
>
> — 22 Sep 2024 · 21 likes · reply · [x.com/const_reborn/status/1837989110927048899](https://x.com/const_reborn/status/1837989110927048899)

> That's not the point really:
>
> - Google doesn’t do SHA-256
> - Bitcoin doesn’t do TPUs
>
> You can't really compare them.
>
> The point is that Bitcoin emits X Billion of block rewards directly into a continuously auctioned global, permissionless, 24/7, hyper-competitive market for electricity and hardware. 
>
> That structure means Bitcoin converts value into compute where it is cheapest. Bang-for-buck more efficiently. 
>
> A centralized firm can buy more hardware, but it cannot reproduce that level of competitive pressure.
>
> Same point with Templar, it takes its market directly to the salient variable for training: gradient based loss reduction ( the miners get paid based on how much they improve the model each step )
>
> Markets for gradients can out-perform centralized training here. It's just that the leverage doesn’t come from more compute, it comes from aligned incentives.
>
> — 5 Jan 2026 · 130 likes · reply · [x.com/const_reborn/status/2008320502142472419](https://x.com/const_reborn/status/2008320502142472419)

### Theme 6. Permissionless: no HR department, no visa, no resume.

Summary: Anyone anywhere can mine. There is no application, no gate, no boss. You frame this both as an engineering property and as the fairest labor market ever built.
Years: 2023, 2024, 2025, 2026. Tweets: about 62. Peak 2025.

> Dont ask permission. No HR department. Just bring compute and your mind.
>
> TAO for days.
>
> — 22 Nov 2023 · 125 likes · reply · [x.com/const_reborn/status/1727431766212858072](https://x.com/const_reborn/status/1727431766212858072)

> Usually a tweet like this would come with a "we're hiring" plug.
>
> but Bittensor doesn't have an HR department.
>
> You don't need to ask anyone to mine TAO.
>
> — 25 Sep 2024 · 89 likes · reply · [x.com/const_reborn/status/1839074174121202033](https://x.com/const_reborn/status/1839074174121202033)

> You don’t need no visa to mine a permissionless network
>
> — 21 Sep 2025 · 294 likes · original · [x.com/const_reborn/status/1969602169616978039](https://x.com/const_reborn/status/1969602169616978039)

> I'm saying this to all talented individuals.
>
> Mine Bittensor.
>
> Live where ever you want.
>
> — 27 Jan 2026 · 306 likes · quote (quoting @Andercot) · [x.com/const_reborn/status/2016253263067115817](https://x.com/const_reborn/status/2016253263067115817)

> Permissionless, global, and algorithmic markets are scary. 
>
> They’re also the most leveling wealth distribution technologies we’ve ever created. 
>
> — 10 Apr 2025 · 134 likes · quote · [x.com/const_reborn/status/1910459778205179960](https://x.com/const_reborn/status/1910459778205179960)

> Permissionless job markets (mining etc) are the true answers to prejudicial and exclusionary practices in the normal job market. 
>
> We don't measure skin color or nationality. Incentive mechanism just measure output. 
>
> The real level playing field.
>
> — 13 Mar 2025 · 39 likes · reply · [x.com/const_reborn/status/1900317367571198059](https://x.com/const_reborn/status/1900317367571198059)

> University is a scam. Learn how to mine Bittensor.
>
> — 28 Apr 2026 · 1,092 likes · original · [x.com/const_reborn/status/2049218909698396522](https://x.com/const_reborn/status/2049218909698396522)

### Theme 7. Decentralized training is the proof.

Summary: Training a frontier model with strangers under a market is the thing critics said could not be done. You have posted every milestone from a GPT-2 loss curve in August 2023 to a 110B run in August 2026. It is your favorite evidence that "incentives beat clusters".
Years: 2023, 2024, 2025, 2026. Tweets: about 100.

> It might not be much, but it might also be the first ever incentivized pretraining.
>
> #TAO https://x.com/const_reborn/status/1687311282968367104/photo/1
>
> — 4 Aug 2023 · 270 likes · original · [x.com/const_reborn/status/1687311282968367104](https://x.com/const_reborn/status/1687311282968367104)

> People underestimate how inefficient things are:
>
> Tensorplex beat Meta's llama-model with a cluster 50x smaller on a 6 month old Bittensor market.  
>
> That's equivalent to five A100 -> H100 hardware upgrades in half a year.
>
> and the difference came from allocation not compute.
>
> — 28 May 2024 · 436 likes · original · [x.com/const_reborn/status/1795537244465635706](https://x.com/const_reborn/status/1795537244465635706)

> Decentralized training is through incentive alignment: everything else is a meme coin.
>
> — 9 Apr 2025 · 36 likes · reply · [x.com/const_reborn/status/1909961162323603729](https://x.com/const_reborn/status/1909961162323603729)

> Just orchestrated a 128 node permissionless decentralized training run, in 5 minutes, for 5 TAO, via @IOTA_SN9 
>
> They can do this up to 100B param models. 
> Unbelievable. 
>
> https://iota.macrocosmos.ai/dashboard https://x.com/const_reborn/status/2062258366739550618/photo/1
>
> — 3 Jun 2026 · 866 likes · original · [x.com/const_reborn/status/2062258366739550618](https://x.com/const_reborn/status/2062258366739550618)

> Today, after 4 months, Subnet 3 (previously @tplr_ai) got retaken by holders through vote.
>
> We're celebrating by releasing our first model.
>
> Teutonic-I: the worlds most performant decentralized LLM.
>
> https://teutonic.ai/paper.html
>
> — 14 Aug 2026 · 1,047 likes · original · [x.com/const_reborn/status/2088349997645312205](https://x.com/const_reborn/status/2088349997645312205)

> We use a digital currency reward because we believe that we cannot rely on altruistic computational resources alone.
>
> Only by coordinating compute and talent with the power of digital markets, like Bitcoin did, can we hope to out-compete centralized labs.
>
> — 24 Aug 2026 · 211 likes · reply · [x.com/const_reborn/status/2091896293970821317](https://x.com/const_reborn/status/2091896293970821317)

### Theme 8. Protocols beat corporations.

Summary: A corporation is a way to organize people and capital. A protocol does the same job with no borders, no payroll and no committee. When protocols work they "blast the corp out the water". You do not want to beat OpenAI at being a company; you want to make the company obsolete.
Years: 2023, 2024, 2025, 2026. Tweets: about 40.

> You're building an AI company. We're building a 1000 year Intelligence Federation. We are not the same.
>
> — 28 Jun 2024 · 745 likes · original · [x.com/const_reborn/status/1806721647241802062](https://x.com/const_reborn/status/1806721647241802062)

> We’ll never compete with OpenAI as a company - that model is perfected in Silicon Valley. 
>
> Our only hope is to revolutionize the way we organize ourselves to solve problems with open protocols. 
>
> But … protocols are difficult, they add a level of complexity unseen in a normal corporate structure: adversaries, black hats, Sybil attacks — things you don’t see in a normal corporation.
>
> But when protocols click? They click in a way that no legal structure can match. 
>
> Permissionlessness, speed, pure directional power. When aligned, protocol based incentives blast the corp out the water. Bitcoin is the first example. 
>
> This is why we work with them. Cause we know that if we’re going to usher in a better way for this technological revolution to be controlled, we need to be better, faster, and more powerful, not just more ethical. 
>
> If you’re in AI? If you’re in crypto? Pivot to the intersection. We have a problem to solve for humanity.
>
> — 21 Dec 2024 · 432 likes · original · [x.com/const_reborn/status/1870521957483233574](https://x.com/const_reborn/status/1870521957483233574)

> Your incredulity is understandable, but wrong: 
>
> Uber, Airbnb, Bitcoin etc, all of them are showcases of low or unpermissioned incentive designs which took the world by storm. 
>
> Uber is an incentivized protocol
> Airbnb is an incentivized protocol 
> Bitcoin mining is an incentivized protocol
>
> Their global, unpermissioned, incentivized, nature **is** their killer app. Why? because it allows them to efficiently procure, allocate, and organize resources.
>
> Allocate, procure, and organize resources in a way that greatly exceeded their competitors classical organizational designs ("MSFT, OpenAI, APPL, NVDA") -- in their domain.  
>
> That was no coincidence. The biggest problem facing an organization is how to organize its people and its capital.
>
> Cryptocurrency goes even further, as it excels at the creation of organizational protocols for human, digital, or other resource, using the most expressive language-of-value the world has ever seen. 
>
> Abstractly, Bittensor is just a toolkit for creating such systems: "a protocol for the creation of incentivized protocols" -- we use it for AI.
>
> Referring back to your quote tweet re ASICs: Bittensor subnets are micro-directed-incentive designs, each 'specific' to a certain digital commodity: hashes, inference speed, storage, training etc.
>
> This makes them similar to ASICs since ASICs are circuits designed for the production of 'specific' computational challenges. 
>
> Hope that helps.
>
> — 3 Apr 2024 · 448 likes · reply · [x.com/const_reborn/status/1775543621393268912](https://x.com/const_reborn/status/1775543621393268912)

> We only have corporations because we didn’t know how to create markets for abstract quantities.
>
> The free soul yearns to live right up against the knife’s edge, where your work is evaluated for its merits, not arbitrated by a committee.
>
> Miners are the most happy and free amongst us.
>
> — 26 Dec 2024 · 92 likes · reply · [x.com/const_reborn/status/1872299351189418159](https://x.com/const_reborn/status/1872299351189418159)

> Crypto-economic systems take time to bootstrap — but the reason we build them is because once they work they’re unstoppable. 
>
> For instance, you can create a cloud provider. Sure, profit day one. Raise funds, buy compute, sell it. 
>
> But can you create a cloud provider that’s purely a protocol? A cloud provider that doesn’t have a datacenter, or land line, or a deals department? 
>
> Just a pure market which commodifies something better and more efficiently than any company ever could. 
>
> That takes a few tries. It’s not “profitable” day one because you’re scaling the ownership out fast. 
>
> But that lasts a thousand years. It’s dimensionally different.
>
> — 8 Sep 2025 · 281 likes · original · [x.com/const_reborn/status/1964847357008195784](https://x.com/const_reborn/status/1964847357008195784)

> The best thing about making a mechanism instead of a company is that it scales itself.
>
> — 17 Jul 2026 · 275 likes · quote · [x.com/const_reborn/status/2077912333939073147](https://x.com/const_reborn/status/2077912333939073147)

### Theme 9. After AI, everything is a commodity.

Summary: When software is cheap, the only things left with value are the raw digital commodities behind the API: compute, inference, storage, data, intelligence itself. Whoever owns the markets for those owns the future. This was a 2023 idea that became your main 2026 frame.
Years: 2023, 2024, 2025, 2026. Tweets: about 47. Peak 2025 and 2026.

> Whats the real moat? The moat is the decentralized digital commodity market and its ability to abstract a super computer that no amount of money can buy, and no single company can own — by definition.
>
> — 24 Sep 2023 · 179 likes · original · [x.com/const_reborn/status/1706024844221993440](https://x.com/const_reborn/status/1706024844221993440)

> The difference was that they needed collect a dataset, now that the dataset is collected it’s simply accessible through distilling from them. 
>
> Intelligence is the ultimate commodity. The densest information.
>
> — 9 May 2023 · 32 likes · quote (quoting @francis_yao_) · [x.com/const_reborn/status/1655817765255794688](https://x.com/const_reborn/status/1655817765255794688)

> Software is eating the world
> AI is eating software
> Incentives will devour AI
>
> — 1 Mar 2025 · 208 likes · quote (quoting @const_reborn) · [x.com/const_reborn/status/1895905156610605378](https://x.com/const_reborn/status/1895905156610605378)

> People beginning to realize that when software is cheap all that remains is digital commodities behind APIs, your agent needs inference (64) it needs compute (4,51), it needs fine tunes (56), it needs VPN IPs (65), it needs data scraping (13), it needs twitter data (22), it need prediction (6) … 
>
> You can orchestrate with a prompt and some digital commodities what previously took teams of hundreds now just with a loop and good prompt. 
>
> It collapses to the commodities and you MUST get on top of them before it falls down. 
>
> #TAO
>
> — 7 Mar 2026 · 646 likes · original · [x.com/const_reborn/status/2030405496130113663](https://x.com/const_reborn/status/2030405496130113663)

> After AI, it’s all just commodities. 
> Own them, make them on Bittensor.
>
> — 29 Jan 2026 · 336 likes · original · [x.com/const_reborn/status/2016871202233401527](https://x.com/const_reborn/status/2016871202233401527)

> Your favorite app is just a prompt running on a server. Your favorite website is just a prompt + some storage. Your favorite chat app is just a prompt + inference. 
>
> When software is cheap everything digital is just AI + commodities: storage, compute, inference, intelligence.
>
> — 26 Jan 2026 · 325 likes · original · [x.com/const_reborn/status/2015772271063728295](https://x.com/const_reborn/status/2015772271063728295)

### Theme 10. Fair launch. No premine. 21 million.

Summary: Every TAO was mined under competition. Nothing was printed for insiders. You treat this as a moral fact and as an engineering choice: a slow emission curve gives the network years to grow character.
Years: 2023, 2024, 2025, 2026. Tweets: about 37. Peak 2025.

> One of the core design choices we went for with TAO (and now with dTAO) was organic growth over a drawn out inflation curve. 
>
> If all the tokens get minted up-front the network has no chance to mature and it privileges sniping, VCs and maximizes value concentration over creation.
>
> — 31 Dec 2024 · 308 likes · original · [x.com/const_reborn/status/1873906010567389692](https://x.com/const_reborn/status/1873906010567389692)

> In contrast, the drawn out supply curves of Bitcoin and Bittensor (as well as others) mean the network has years to develop strength of character, both of its holders and technology.
>
> Anti-fragility can only play out this way.
>
> 21 M was never a meme.
>
> — 31 Dec 2024 · 147 likes · reply · [x.com/const_reborn/status/1873906452789895515](https://x.com/const_reborn/status/1873906452789895515)

> 0 initial supply, no privileged allocation, slow consistent inflation over multiple years.
>
> These are not arbitrary design choices.
>
> — 19 Jan 2025 · 108 likes · reply · [x.com/const_reborn/status/1880887359602032947](https://x.com/const_reborn/status/1880887359602032947)

> It’s funny cause this is just proof that we were actually here first, and that everyday people, nerds stoked on the vision, got an opportunity to mine Bittensor just like it was for Bitcoin. 
>
> And it took a couple years before people got very serious mining — now we have the best miners on earth probably. 
>
> I’d rather give out that initial TAO to people that had the guts to mine something rather than VC insiders. 
>
> That’s the whole difference.
>
> — 20 Aug 2025 · 141 likes · reply · [x.com/const_reborn/status/1958228919381049568](https://x.com/const_reborn/status/1958228919381049568)

> @chang_defi Our philosophy was that every TAO ever mined needed to be related to work put in, and also under competition, nothing free, nothing given, everything fought for. The belief has always been open transparent and free markets for the development of AI.
>
> — 26 Mar 2026 · 232 likes · reply · [x.com/const_reborn/status/2037074088640823746](https://x.com/const_reborn/status/2037074088640823746)

> Many years back @KyleSamani and I had a call.
>
> He dropped at 5 minutes because i said we used a 21M Bitcoin style supply curve. 
>
> Happy to pickup the conversation where we left off.
>
> — 4 Apr 2026 · 717 likes · quote (quoting @KyleSamani) · [x.com/const_reborn/status/2040459231187386627](https://x.com/const_reborn/status/2040459231187386627)

### Theme 11. Anti-fragility: exploits, FUD and volatility are the teacher.

Summary: You welcome attacks. An exploit is a signal that hardens the mechanism. A price crash filters out weak hands. Stagnation, not volatility, is what kills a network.
Years: 2024, 2025, 2026. Tweets: about 44.

> Remember: dialectics rule the world. FUD is the fuel that burns the dead wood. Welcome it. Understand anti-fragility. Forget charts and be bullish only on the truth. The most profound things take time to understand, they are not junk food, they are medicines. #TAO
>
> — 30 Mar 2024 · 628 likes · original · [x.com/const_reborn/status/1774097399428038858](https://x.com/const_reborn/status/1774097399428038858)

> The real death knell of a network is stagnation, not volatility.
>
> Volatility is the food that the anti-fragile feed off -- fuel that feeds the fire of innovation. 
>
> Evidence: within 20 minutes of subnet 21 going down, Omega birthed a better one which will give us Sora.
>
> — 11 Jun 2024 · 181 likes · original · [x.com/const_reborn/status/1800636370618401202](https://x.com/const_reborn/status/1800636370618401202)

> Exploits are what teach a system its weak spots. 
> The quicker you find them the faster you learn. 
>
> The outcome of this eventful evening is that Bittensor will invent lock-based subnet ownership -- specifically: ownership of a subnet determined by a team's long term  economic commitment to the project.
>
> This will mean: 1) investors see long in advance if an owner has unlocked their tokens, 2) be able to reprice the subnet before the owner and 3) liquidly direct their own conviction to another team, or agent, to manage the system.
>
> Thank you @DistStateAndMe for helping further Bittensor's decentralization and develop a solution to one of cryptos oldest problems: founders who rug their token holders.
>
> Looking forward to training some 1T param models with the miners who are experts in this unique field.
>
> "What is dead can never die"
>
> — 10 Apr 2026 · 1,324 likes · quote · [x.com/const_reborn/status/2042427496751063191](https://x.com/const_reborn/status/2042427496751063191)

> Pressure creates diamonds and there have been no tokens in existence, no underlying technologies so molded by the relentless pressure, of automatic sells, by the nefarious onslaught of miners, and the social cabalism of Bittensor, and stood tall, in the test of time, than those that still breathe TAO emission
>
> — 13 Aug 2025 · 335 likes · original · [x.com/const_reborn/status/1955744700931170442](https://x.com/const_reborn/status/1955744700931170442)

> But that's not the end: the absolute power of Realpolitik, of Rome for instance, begets the creation of new legitimate systems of power designed to remain fair even under adversarial conditions.
>
> Where Rome fell, the legitimacy of Christianity followed.
> Where the US falls, mechanisms such as Bitcoin will follow.
>
> and if we are to make a new world with this knowledge, our next systems must not deny power politics, exploitation, deception, Trumpism -- they must assume them. 
>
> This is why digital currency as a political tool is so profound: A language of power and money that we can use cryptographically constrain these forces.
>
> Embedding fairness, neutrality, and shared rules directly into the algorithmic systems themselves.
>
> The goal is not to eliminate the struggle for power, but to internalize it -- so that even self-interested, adversarial behavior produces legitimate outcomes.
>
> Do not fear the exploit. Design for it
>
> — 4 Jan 2026 · 95 likes · reply · [x.com/const_reborn/status/2007605197359460726](https://x.com/const_reborn/status/2007605197359460726)

> Zoom out. 
>
> Seen this a number of times. Bull markets attract the most unsavory of people. They build casinos, ponzis and financial schemes. The market punishes us. 
>
> Next, those people, those scammers leave. Bitcoin goes back to being real. Crypto takes a breath of fresh air, the builders keep building and cycle begins again. 
>
> That cycle literally *is* the technology. A monetary system that reacts and sheds itself. 
>
> If it were any different it would be a stultified cabal that stays ugly for ever. It can’t. That’s what is good about it. 
>
> When you understand technically what was built with Bitcoin. You’ll understand it doesn’t matter if Epstein funded the core developers. It doesn’t matter if Satan himself runs the nodes, made a DAT, or wrote the code. 
>
> The structure or Bitcoin is itself Good and decentralized, and fair. 
>
> The need for this monetary revolution didn’t end with a price drop and indeed we need it now more than ever. 
>
> LFG
>
> — 6 Feb 2026 · 498 likes · original · [x.com/const_reborn/status/2019742359622164541](https://x.com/const_reborn/status/2019742359622164541)

### Theme 12. Giving power away: decentralizing the foundation and yourself.

Summary: Since November 2023 you have said in public that the goal of power is to give it away, that OTF should fade like Satoshi did, and that Bittensor is bigger than you. This is the arc the talk's "I hold a key" section sits on.
Years: 2023, 2024, 2025, 2026. Tweets: about 32.

> Power came from above when firing Sam.
>
> Where it was, and always will be until we draw it down into the hands of many. 
>
> Decentralization is just this: forcing the separation and distribution of control. 
>
> Pulling it down.
>
> — 18 Nov 2023 · 314 likes · original · [x.com/const_reborn/status/1725967721546801236](https://x.com/const_reborn/status/1725967721546801236)

> The goal of power is to give it away. 
> The art of power is its decentralization. https://x.com/const_reborn/status/1725967725762134121/photo/1
>
> — 18 Nov 2023 · 227 likes · reply · [x.com/const_reborn/status/1725967725762134121](https://x.com/const_reborn/status/1725967725762134121)

> True story, we got to subnets on Bittensor because I lost my laptop and realized there was no entry for me in the network we created. 
>
> Decentralization is all about this:
>
> we're less important than we think we are. So who's next?
>
> — 5 Oct 2023 · 89 likes · reply · [x.com/const_reborn/status/1709779337359077871](https://x.com/const_reborn/status/1709779337359077871)

> When we first built the Opentensor Foundation to support Bittensor we were responsible for everything:
>
> - The Python API.
> - The Blockchain code.
> - The Explorer/Scanner.
> - The Validator nodes.
> - The Subnet protocols.
> - The CLI.
> - Fundraising. 
> etc. etc.
>
> Today, the majority of these endeavors are being competitively run by other orgs in the ecosystem. 
>
> @manifoldlabs 
> @FoundryServices 
> @taostats 
> @taoshiio 
> @Datura
> @TensorplexLabs 
> ... and many more subnet owners + validators
>
> In the next few years all of it will be run by competitive collectives like this. 
>
> Choose your gov with liquid democracy and competitive voting. 
>
> We don't want a monopoly on anything.
>
> — 17 Jun 2024 · 419 likes · original · [x.com/const_reborn/status/1802711231201227202](https://x.com/const_reborn/status/1802711231201227202)

> Satoshi left and Bitcoin became its community. 
>
> OTF will fade in the same way. 
>
> Kill the center, become the swarm.
>
> — 14 Mar 2025 · 342 likes · quote (quoting @Gylez) · [x.com/const_reborn/status/1900593003711844637](https://x.com/const_reborn/status/1900593003711844637)

> I’m kinda just in awe these days. Bittensor is so much bigger than OTF, than me.
>
> — 6 Jun 2025 · 143 likes · reply · [x.com/const_reborn/status/1930778798280073328](https://x.com/const_reborn/status/1930778798280073328)

> @DreadBong0 @opentensor 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 )
>
> — 12 Feb 2026 · 379 likes · reply · [x.com/const_reborn/status/2022093414293286953](https://x.com/const_reborn/status/2022093414293286953)

### Theme 13. Agents will run Bittensor.

Summary: The 2026 theme. Agents mine, build subnets, manage fleets and pay for their own compute. Your most-liked tweet of all time is about your own six agents. This is where the "your agents can mine for you" line in the thesis comes from.
Years: 2024 (7 mentions), 2025, 2026 (58 mentions). Tweets: about 87.

> I have 6 agents running now, two are doing ML research, one is cloning Claude code, two trade hyperliquid and one mines Bittensor. 
>
> All they need is compute from Bittensor (Lium,  Basilica and Targon) and inference from Chutes and Openrouter.
>
> The trading agents pay for the compute and inference by buying credits with proceeds. 
>
> I could literally die and this system wouldn’t turn off. Indeed it’s programmed to just get better.
>
> — 12 Mar 2026 · 2,191 likes · original · [x.com/const_reborn/status/2032002840961442279](https://x.com/const_reborn/status/2032002840961442279)

> Bittensor will be run by agents. They will feed the mining, resist the exploits, manage the fleets, build the subnets and consume the commodities
>
> — 14 Mar 2026 · 938 likes · original · [x.com/const_reborn/status/2032748075706425367](https://x.com/const_reborn/status/2032748075706425367)

> Basically all subnets on Bittensor are being mined by teams which heavily use AI optimized solutions, scripts that evaluate software submissions or models with multi layer inferences to update them and even higher level scripts that adapt those sub level optimizers continuously. In Bittensor’s final form no one will ever write a miner client. It all just becomes evals + incentives and swarms of agent miners unleashed against them. Literal armies of recursive agent scripts and LLM calls optimizing all aspects of every submission ad infinitum.
>
> — 12 Feb 2026 · 232 likes · reply · [x.com/const_reborn/status/2021987989476253978](https://x.com/const_reborn/status/2021987989476253978)

> Agents that upgrade themselves that run their own infrastructure that live entirely on digital currency
>
> — 4 Feb 2026 · 220 likes · original · [x.com/const_reborn/status/2018839416299762016](https://x.com/const_reborn/status/2018839416299762016)

> the only way to orchestrate 24/7, resource-optimal, production loops is with agents under markets.
>
> — 29 Jul 2026 · 196 likes · original · [x.com/const_reborn/status/2082546236893090193](https://x.com/const_reborn/status/2082546236893090193)

> First it is the model, the llm. 
>
> Then it is the harness, the backend and the tool calls. 
>
> Next it is the agent, the memory and the planning. 
>
> From there it becomes the organization, the swarm, and your orchestrator. 
>
> And then it becomes the protocol, a connective market, incentives and their governance.
>
> — 17 May 2026 · 475 likes · original · [x.com/const_reborn/status/2055917811852034179](https://x.com/const_reborn/status/2055917811852034179)

### Theme 14. Bittensor has no country. The third path.

Summary: Decentralized systems are supranational. They sit outside the US–China race and make it obsolete. You said "third path" yourself in January 2026, in the same words the thesis uses.
Years: 2024, 2025, 2026. Tweets: about 60.

> Truly decentralized systems are supranational and don’t adhere to standard geopolitics — indeed, they render them obsolete.
>
> I’ll be visiting Asia to discuss refocusing the AI race through non-permissioned borderless competitive systems; I believe it's the way to avert an AI hot war.
>
> Please connect me with the most underground of underground hacker spaces, if you know any.
>
> — 7 Feb 2024 · 229 likes · original · [x.com/const_reborn/status/1755217914775429345](https://x.com/const_reborn/status/1755217914775429345)

> This is why Bittensor has no country, or tribe. 
>
> We play a neutral game which cracks social issues we've been facing as societies for decades. 
>
> The left wing is concerned with racial politics. 
> The right is concerned with national goals.  
>
> Their answers are united in Web3.
>
> — 13 Mar 2025 · 210 likes · quote (quoting @const_reborn) · [x.com/const_reborn/status/1900318650675912822](https://x.com/const_reborn/status/1900318650675912822)

> There’s only one other monetary system in the world that can compete with US backed fiat AI
>
> — 22 Jan 2025 · 341 likes · quote (quoting @OpenAI) · [x.com/const_reborn/status/1882044268535226386](https://x.com/const_reborn/status/1882044268535226386)

> It’s not regulation which saves us. EU style bureaucracy just ensures advancement doesn’t birth itself in that ecosystem. 
>
> Corporatism is the most capital efficient solution and the clear winner in 2026. Yet birthing this in either jurisdiction, China or USA through the corporation will certainly lead to an unholy union of state corporate fascism (already organizing itself predictably out of Silicon Valley). This is just because there are such strong self reinforcing qualities (monopolization is always the goal and likely outcome of a corporation). 
>
> The first (regulation) is not a route at all because it nullifies itself and the second (corporate control) is dystopian.
>
> There is a third i believe which builds the control structure and capital formation structure for AI using the immutable tooling of protocol technologies, Bitcoin being the prime example of a tech which avoids monopolization through its design. 
>
> “Can’t monopolize and centralize” while simultaneously “can’t be stultified and is efficient” must be the carrying qualities of such a system.
>
> — 18 Jan 2026 · 83 likes · reply · [x.com/const_reborn/status/2012688518401699887](https://x.com/const_reborn/status/2012688518401699887)

> Fiat-funded, fiat-protected and fiat-controlled US and foreign labs will increasingly become reliant on Big Government/Federal Reserve
>
> In exchange, those same elites will seek privileged access, propogandization, censorship, and monopoly over intelligence and over our minds and ultimately our freedom of thought.
>
> Alternate ways of organizing -- bottom up and outside of these regimes -- is our only hope in an era of intelligence fascism.
>
> — 23 Jul 2026 · 285 likes · quote · [x.com/const_reborn/status/2080245516105900060](https://x.com/const_reborn/status/2080245516105900060)

> @miles_engineer I hope Bittensor gives the world a protocol for neutral digital production -- under the umbrella of Bitcoin's monetary neutrality.
>
> — 5 Jan 2026 · 186 likes · reply · [x.com/const_reborn/status/2008265035815284782](https://x.com/const_reborn/status/2008265035815284782)

### Theme 15. Verification is the whole game. Intelligence measures intelligence.

Summary: Checking an answer is cheaper than finding it (the P versus NP asymmetry). That gap is what lets a small validator judge a big miner, and it is why a benchmark, not a model, is the product. "The best measure of intelligence is intelligence itself."
Years: 2024, 2025, 2026. Tweets: about 110 (broad keyword match on verify, benchmark, evaluate, judge).

> 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. Why? because there is actually a computational asymmetry between determining if something is good vs creating it.
>
> That falls back on the famous P!=NP conjecture which says you can check the solution faster than you can solve it. 
>
> 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.
>
> — 8 Jul 2024 · 141 likes · quote · [x.com/const_reborn/status/1810381338106847557](https://x.com/const_reborn/status/1810381338106847557)

> In Bitcoin, the miners produce a hash of the block data by incrementing a nonce until Sha256( nonce, block_data) has a required number of leading zeros. 
>
> When a miner returns the nonce + block_data the check is simple, namely running Sha256( nonce, block_data) once.
>
> However, finding the specific nonce where the leading zeros suffice the min requirement will take a really long time, a quality of one-way-functions like Sha256.
>
> Sha types algorithms are quintessential examples of the asymmetry between finding a solution and validating it. 
>
> Many such instances exist and the problem is well known: https://en.wikipedia.org/wiki/P_versus_NP_problem
>
> Intelligence also has this quality. Where a large model may be required to find a good solution to a problem (i.e. answer a math problem) but where a smaller one can check that solution easily.
>
> Also a quality of Subnet 1, where when the question is constructed before-hand by looking at the relevant reference data, the answer can be checked easily by a smaller LLM with that reference data in hand. 
>
> The subnet incentive mechanism runs like a one-way-function, hash function for intelligence.
>
> — 1 Apr 2024 · 10 likes · reply · [x.com/const_reborn/status/1774839304696864908](https://x.com/const_reborn/status/1774839304696864908)

> This is why I always tell teams to focus myopically on verification. Your product is down stream from the verification not upstream from it. What exactly is this commodity you are creating? How do you measure it? 
>
> You can spin your wheels building fancy optimizations all day and then realize in an instant that miners won’t use it, or already invented one better themselves.
>
> — 27 May 2025 · 18 likes · reply · [x.com/const_reborn/status/1927294652488540419](https://x.com/const_reborn/status/1927294652488540419)

> The future is not AI replacing jobs.
> The future is everything becoming a feedback loop with intelligent validation.
>
> — 24 Apr 2026 · 449 likes · original · [x.com/const_reborn/status/2047596581218603016](https://x.com/const_reborn/status/2047596581218603016)

> Schools of the future will just be proctors 
>
> i.e. testing facilities which manage non gameable, non memorizable benchmarks, like Bittensor validators. 
>
> You’ll learn on your own with AI and visit test taking facilities to prove any skill you want.
>
> — 23 Jan 2026 · 155 likes · quote (quoting @naval) · [x.com/const_reborn/status/2014691122887409971](https://x.com/const_reborn/status/2014691122887409971)

> Cognition labs' "Devin" turned the fairly obscure SWE-benchmark into a ++billion dollar valuation. 
>
> Benchmark-first companies will rule the 21st century.
>
> — 12 Aug 2024 · 90 likes · original · [x.com/const_reborn/status/1823086257284768170](https://x.com/const_reborn/status/1823086257284768170)

---

## 2. How your thinking evolved

### 2023 (451 tweets): the enemy is OpenAI, the answer is ownership

- Preoccupied with: OpenAI, Sam Altman and regulatory capture (18 of your 33 lifetime OpenAI/Altman tweets are from 2023). Worldcoin as "fiat humanity". Inference as training data, distillation, the prompting network as "the largest distillation dataset the world has ever seen". Yuma Consensus explainers. The launch of subnets ("Revolution", Oct 2023) and the first announcement of dynamic TAO (29 Dec 2023).
- Key move: "open ownership or gtfo" (Mar 2023) and "If you don't own it, you won't control it" (May 2023) set the frame that still holds.
- The Altman firing (Nov 2023) produced your first public promise to give your own power away: "This is a good time to announce that I will be giving away my personal power. Back to the community by vote." — 18 Nov 2023, 157 likes, [link](https://x.com/const_reborn/status/1725967728727498877)
- Mea culpa on competitors (27 Nov 2023): you admitted "True hypocrisy" in ignoring other decentralized AI protocols and listed them.

### 2024 (410 tweets): the physics of incentives, and the first honesty about the key

- Preoccupied with: money as an energy function; "thermodynamics ≅ neural optimization ≅ crypto currency"; Bittensor as "a protocol for the creation of incentivized protocols"; subnets as "the ASIC chips of decentralized computing". Subnet 9 beating GPT-2 and then Llama with a cluster "50x smaller". The July 2024 wallet drain and the decision to firewall the chain.
- What changed: you stopped dunking on OpenAI (only 6 mentions) and started explaining. Long threads to critics (Mar 2024, Sep 2024) laid out the design in plain terms. In September 2024 you wrote your first frank statement of the sudo key and proof of authority: "The key is a multi sig but it **does** have root privs which is dangerous -- ill admit that." — 22 Sep 2024, 110 likes, [link](https://x.com/const_reborn/status/1837985092301422799)
- Also in 2024: "Democracy was the product of centralization" (Jul), "You're building an AI company. We're building a 1000 year Intelligence Federation" (Jun), "Turn all jobs into protocols / Destroy the corporation" (Dec). The 21M supply curve became a stated moral position, not just a parameter.
- You walked back one thing: after "God's programming language" and "God's machine learning" you wrote in Dec 2024 "Bittensor is powerful tech, but it does NOT make us Gods. I should be more careful with analogy. Yikes." — 6 Dec 2024, 9 likes, [link](https://x.com/const_reborn/status/1865109615374917887)

### 2025 (1,083 tweets, your busiest year): dynamic TAO, permissionless everything, and the fight for the word "decentralized"

- Preoccupied with: dTAO launch (13 Feb 2025) and its economics; "Root is dead"; your own subnet (Affine); Templar and permissionless training; "no visa", "no HR", "prejudice-lessness"; UBI as a bribe; "fiat AI"; Bitcoin as the industry's shield; TaoFlow (Nov); the first halving (Dec).
- What changed: you became a participant, not only a founder ("I can participate in Bittensor in a fair way. I'm making a subnet." Mar 2025). You picked fights with Prime Intellect, Nous and Gensyn over what counts as decentralized training ("churches without faith", Aug 2025), then congratulated Gensyn in December. You disclosed the 2021 wallet split and blackmail (Aug 2025). Religious language rose sharply (27 tweets).
- Politics entered: #FreePalestine (May, 1,025 likes), Palantir and Karp (Aug, Sep), Trump tariffs read as "Art of the Deal" (Apr).
- The "real fear" theme came back after a quiet 2024: 24 tweets, including "You will work for the machines that work for the elites — and you'll own none of it."

### 2026 (524 tweets to 11 Sep): agents, commodities, and the hardest year for the key

- Preoccupied with: agents (58 mentions, up from 22); "when software is cheap all that remains is digital commodities"; Teutonic (10B, then 110B); the Covenant / Subnet 3 dispute (Apr); the return of root validators (Jun); Realpolitik and legitimacy (Jan); "we become Gaza" (Jun).
- What changed: on 12 Feb 2026 you said your role would continue "just not with the same legal authority (and that's important)". On 10 Apr 2026 you wrote "I do not have the ability to suspend emissions", the first time you defined your own power by what it cannot do. Five of your biggest posts of the year are X Articles whose text is not in this archive.
- New synthesis: "The future is not AI replacing jobs. The future is everything becoming a feedback loop with intelligent validation." (Apr 2026). "After AI, it's all just commodities. Own them, make them on Bittensor." (Jan 2026).

### What you stopped saying

- Worldcoin and "fiat humanity": 6 tweets in 2023, 9 in 2025, 1 in 2026.
- OpenAI dunks: 18 in 2023, 4 in 2026.
- "Our Ethereum moment" and "We will have our DeFi": the 2023 crypto-lineage framing is gone. You now compare to Bitcoin only.
- "1000 year Intelligence Federation": used in 2024, then dropped in favor of "protocol" and "market".
- "Bitcoin is 450 times more powerful than all compute providers combined" (Dec 2024). By Jan 2026 you wrote "You can't really compare them." The thesis is right to avoid the FLOPs claim; so are you now.
- "God's programming language" and similar (see 2024 walk-back).
- "Root is dead, long live Bittensor" (Feb 2025) became "root is coming back, and it's going to be glorious" (Jun 2026). See Tensions.

---

## 3. Lines worth saying on stage

Verbatim. Short. Sorted roughly by where they fit in the thesis.

On the fear (thesis sections 1 to 4):

- "The real risk of AI was never robots with a vendetta. It’s that all the capital ownership concentrates down to a small elite set of technocrats in SF." — 30 Oct 2025, 465 likes, [link](https://x.com/const_reborn/status/1983941275876520051)
- "Without carefully constructed ownership and openness layers, humanity doesn’t end up unequal -- it ends up enslaved." — 7 Jan 2026, 153 likes, [link](https://x.com/const_reborn/status/2009032093729477050)
- "You will work for the machines that work for the elites — and you’ll own none of it." — 31 May 2025, 130 likes, [link](https://x.com/const_reborn/status/1928851952709800276)
- "And just like that we collectively saw the future of inequality" — 13 Jun 2026, 2,588 likes, [link](https://x.com/const_reborn/status/2065606907855831301)
- "We’re not ready for a world where armies are made of machines, not men — where human power is so eroded that the AI cabal can control the world from a laptop." — 5 Aug 2025, 175 likes, [link](https://x.com/const_reborn/status/1952710606231949733)
- "Just that this time they’re not diluting money: they’re diluting us." — 24 Jul 2023, 48 likes, [link](https://x.com/const_reborn/status/1683501214619557894)
- "UBI is what they will offer you in return for control, it’s a bribe, a contract you will sign to solidify your agreement with the loss of power." — 6 Jun 2025, 127 likes, [link](https://x.com/const_reborn/status/1930927421651296503)

On the cure and the mechanism (sections 5 to 8):

- "Bitcoin broke the gov's monopoly on currency. / Bittensor stops the gov's monopolization of AI." — 30 Oct 2023, 444 likes, [link](https://x.com/const_reborn/status/1719092725402800602)
- "the token is the objective function" — 14 Dec 2024, 185 likes, [link](https://x.com/const_reborn/status/1868075527371980869)
- "You don’t actually solve any technical problems, you just structure the rewards so other solve it for you. It’s inverse, upside down, inside out engineering." — 30 Jan 2025, 117 likes, [link](https://x.com/const_reborn/status/1885108699926045138)
- "Thermodynamic computing: energy computes / Machine Learning: error computes / Incentive mechanisms: markets compute" — 30 Jun 2025, 119 likes, [link](https://x.com/const_reborn/status/1939620057342152742)
- "AI can judge anything. Money can move anything." — 24 Apr 2026, 253 likes, [link](https://x.com/const_reborn/status/2047599733544263777)
- "We’re not creating a marketplace of AI. We’re using markets to create AI." — 21 Dec 2024, 57 likes, [link](https://x.com/const_reborn/status/1870523839396135392)
- "and the difference came from allocation not compute." — 28 May 2024, 436 likes, [link](https://x.com/const_reborn/status/1795537244465635706)
- "A centralized firm can buy more hardware, but it cannot reproduce that level of competitive pressure." — 5 Jan 2026, 130 likes, [link](https://x.com/const_reborn/status/2008320502142472419)
- "The only thing larger than a closed cluster is a network joined together." — 24 Aug 2026, 82 likes, [link](https://x.com/const_reborn/status/2091896296189624600)
- "No matter how cracked your team is, you’ll never beat the machine that picks cracked teams." — 13 Feb 2025, 302 likes, [link](https://x.com/const_reborn/status/1890010448726254035)
- "When you mine Tao and get outcompeted, it just means smarter people picked up your pick axe and started digging for you." — 1 Apr 2025, 240 likes, [link](https://x.com/const_reborn/status/1906887497205899292)
- "To create intelligence by programming currency." — 19 Jul 2025, 292 likes, [link](https://x.com/const_reborn/status/1946542962520760331)
- "Software is eating the world / AI is eating software / Incentives will devour AI" — 1 Mar 2025, 208 likes, [link](https://x.com/const_reborn/status/1895905156610605378)

On fair launch and permissionlessness (section 9):

- "Dont ask permission. No HR department. Just bring compute and your mind." — 22 Nov 2023, 125 likes, [link](https://x.com/const_reborn/status/1727431766212858072)
- "You don’t need no visa to mine a permissionless network" — 21 Sep 2025, 294 likes, [link](https://x.com/const_reborn/status/1969602169616978039)
- "Don’t submit a resume. / Submit a model." — 29 Jan 2024, 182 likes, [link](https://x.com/const_reborn/status/1752061498732712064)
- "21 M was never a meme." — 31 Dec 2024, 147 likes, [link](https://x.com/const_reborn/status/1873906452789895515)
- "nothing free, nothing given, everything fought for." — 26 Mar 2026, 232 likes, [link](https://x.com/const_reborn/status/2037074088640823746)
- "They’re also the most leveling wealth distribution technologies we’ve ever created." — 10 Apr 2025, 134 likes, [link](https://x.com/const_reborn/status/1910459778205179960)
- "We don't measure skin color or nationality. Incentive mechanism just measure output." — 13 Mar 2025, 39 likes, [link](https://x.com/const_reborn/status/1900317367571198059)

On the third path (section 10):

- "This is why Bittensor has no country, or tribe." — 13 Mar 2025, 210 likes, [link](https://x.com/const_reborn/status/1900318650675912822)
- "It wont be a country, it will be a protocol." — 10 May 2025, 70 likes, [link](https://x.com/const_reborn/status/1921153183725814014)
- "Truly decentralized systems are supranational and don’t adhere to standard geopolitics — indeed, they render them obsolete." — 7 Feb 2024, 229 likes, [link](https://x.com/const_reborn/status/1755217914775429345)
- "Bitcoin is international law." — 4 Jan 2026, 81 likes, [link](https://x.com/const_reborn/status/2007618557413376381)
- "There is no way to compete with fiat-money tank engines in the center lane." — 31 Dec 2025, 184 likes, [link](https://x.com/const_reborn/status/2006479414574592245)

On the honest part and the call (sections 12 to 14):

- "The goal of power is to give it away. / The art of power is its decentralization." — 18 Nov 2023, 227 likes, [link](https://x.com/const_reborn/status/1725967725762134121)
- "Kill the center, become the swarm." — 14 Mar 2025, 342 likes, [link](https://x.com/const_reborn/status/1900593003711844637)
- "Exploits are what teach a system its weak spots. / The quicker you find them the faster you learn." — 10 Apr 2026, 1,324 likes, [link](https://x.com/const_reborn/status/2042427496751063191)
- "Do not fear the exploit. Design for it" — 4 Jan 2026, 95 likes, [link](https://x.com/const_reborn/status/2007605197359460726)
- "The real death knell of a network is stagnation, not volatility." — 11 Jun 2024, 181 likes, [link](https://x.com/const_reborn/status/1800636370618401202)
- "Intelligence is cheap. / Spirit is rare. / Never give up." — 23 Dec 2024, 277 likes, [link](https://x.com/const_reborn/status/1871310335824757032)
- "The best teachers will be LLMs. The best feedback will be monetary. The best models will train competitively. The best AI company will be co-owned. The best software will be open source. And there is only one way to get all of it." — 13 Jun 2026, 788 likes, [link](https://x.com/const_reborn/status/2065914716589850671)
- "The answer to fears of AI is this, not top down censorship or unchecked acceleration, but human unity with the machines." — 11 Feb 2025, 390 likes, [link](https://x.com/const_reborn/status/1889295429566701891)

---

## 4. Ideas in the tweets that are missing from the thesis

Ordered by how much they would add.

### 4.1 The verification asymmetry (P versus NP) is why the machine works

The thesis explains Yuma Consensus as "validators judge the quality". It never says why a small judge can grade a big worker. You have said it for two years: checking is cheaper than solving, and that gap is the whole design. It also answers "how can a validator evaluate a 100B model?", a question the room will ask.

> 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. Why? because there is actually a computational asymmetry between determining if something is good vs creating it.
>
> That falls back on the famous P!=NP conjecture which says you can check the solution faster than you can solve it. 
>
> 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.
>
> — 8 Jul 2024 · 141 likes · quote · [x.com/const_reborn/status/1810381338106847557](https://x.com/const_reborn/status/1810381338106847557)

> In 2019, Richard Sutton began his prescient essay, ‘The Bitter Lesson’, like this:
>
> The biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin.
>
> Without fail, since the inception of the field, systems equipped with a priori information (in a way analogous to some theories of intelligence) have lost out to those armed with massive computational budgets, and left to find solutions on their own. This was the zero to one realization for machine intelligence.
>
> This principle can be observed across every field of AI. For example, early approaches conceived of vision as searching for edges, angles and cylinders but were later discarded in favour of deep learning approaches that allowed neural networks to find invariances (like edges) in the data themselves. Recently, approaches have prevailed that remove human direction altogether and allow systems to discover meaning from unlabeled images.
>
> In language we see a similar pattern. Early researchers tried to build knowledge of ontological concepts and syntax into their systems in order to orchestrate language comprehension, but these systems were replaced by those those who built their own syntax trees, and later, by those that could learn from unlabeled corpora without any prior ontological framework or notion of syntax to speak of.
>
> The direction here is clear. The more withheld the lock maker, teacher, engineer, the better the result. Solutions, with their endless complexities, cannot be manufactured linearly. Instead, we must include only the meta-methods, step back, and allow the interaction of data and compute to uncover intelligence independently.
>
> We must also step back from the definitions of intelligence that currently guide our advancement. We must relinquish benchmarks, paper writing, and conferences. We must think of the field itself as a meta system upon which AI can grow.
>
> The Bittensor network is the largest bounding of this problem, the alignment of compute and human capital through a monetary container, creating the optimal environment for the advancement and growth of intelligence alongside humanity. The network is a pure market for representation knowledge, evaluation by machine intelligence systems directly: a measure of intelligence by intelligence itself.
>
> — 8 Jul 2024 · 95 likes · reply · [x.com/const_reborn/status/1810369546609635795](https://x.com/const_reborn/status/1810369546609635795)

Use the last two sentences of that essay. They are the cleanest statement of the mission in the archive.

### 4.2 Decentralized training is the proof point, and the thesis does not show it

Thesis section 11 lists Chutes, Targon, Lium and revenue. It does not mention that the network has trained a 10B model with 73 teams, is training a 110B model, and can spin a 128-node run for 5 TAO. Training is the thing centralized labs say cannot be done outside a data center. It belongs on the "what has been built" slide.

> Anyone in the world can contribute: rent GPUs, read the dataset, download and train on the current state, and then submit a model which pushes the loss slightly lower.
>
> We had roughly 73 distinct teams that did over 70 days. https://x.com/const_reborn/status/2088350004742083015/photo/1
>
> — 14 Aug 2026 · 85 likes · reply · [x.com/const_reborn/status/2088350004742083015](https://x.com/const_reborn/status/2088350004742083015)

> Teutonic-I was trained "competition parallel": an economic training regime where adversaries submit updates and only those that verifiably push the model can attain rewards.
>
> Teams race to procure compute, build algorithms, and outsmart each other to get the the next epsilon of lowered loss.
>
> — 14 Aug 2026 · 97 likes · reply · [x.com/const_reborn/status/2088350002338779492](https://x.com/const_reborn/status/2088350002338779492)

> Announcing Teutonic-II-110B: a frontier permissionless training of a 110B LLM 
>
> We are scaling 10x beyond our previous model and 50% larger than any prior decentralized run. https://x.com/const_reborn/status/2091896286035235027/photo/1
>
> — 24 Aug 2026 · 852 likes · quote (quoting @const_reborn) · [x.com/const_reborn/status/2091896286035235027](https://x.com/const_reborn/status/2091896286035235027)

Caution: these are your own subnet's numbers. Say so.

### 4.3 "When software is cheap, all that remains is commodities"

This is your main 2026 frame and it fits the labor argument in section 2. If AI writes the software, the moat is not code, it is the raw inputs: compute, inference, data, and intelligence itself. Whoever owns the markets for those inputs owns the economy. That is a sharper reason for "own the AI capital" than the economists give.

> People beginning to realize that when software is cheap all that remains is digital commodities behind APIs, your agent needs inference (64) it needs compute (4,51), it needs fine tunes (56), it needs VPN IPs (65), it needs data scraping (13), it needs twitter data (22), it need prediction (6) … 
>
> You can orchestrate with a prompt and some digital commodities what previously took teams of hundreds now just with a loop and good prompt. 
>
> It collapses to the commodities and you MUST get on top of them before it falls down. 
>
> #TAO
>
> — 7 Mar 2026 · 646 likes · original · [x.com/const_reborn/status/2030405496130113663](https://x.com/const_reborn/status/2030405496130113663)

> If you think software is your moat above all other things, you're going to have a bad time
>
> — 11 Jun 2026 · 218 likes · original · [x.com/const_reborn/status/2065144999155507394](https://x.com/const_reborn/status/2065144999155507394)

> The commodification of compute is one thing (H100s already have their own tickers), but so will intelligence.
>
> Being that it is a definable, benchmarkable quantity, there will be gradations of it—dimensions of it—like a cut stone, measured for speed, force, cost, and depth.
>
> — 20 Apr 2026 · 304 likes · quote (quoting @Rewkang) · [x.com/const_reborn/status/2046183589985788221](https://x.com/const_reborn/status/2046183589985788221)

### 4.4 Agents as the new miner, and the "I could die" line

The thesis says "Your agents can mine for you" once. The archive has your most-liked tweet of all time on this, plus a full vision: agents that pay for their own compute, upgrade themselves and never turn off. This is what makes the network self-running rather than founder-dependent, and it is a direct answer to "it runs through me".

> I have 6 agents running now, two are doing ML research, one is cloning Claude code, two trade hyperliquid and one mines Bittensor. 
>
> All they need is compute from Bittensor (Lium,  Basilica and Targon) and inference from Chutes and Openrouter.
>
> The trading agents pay for the compute and inference by buying credits with proceeds. 
>
> I could literally die and this system wouldn’t turn off. Indeed it’s programmed to just get better.
>
> — 12 Mar 2026 · 2,191 likes · original · [x.com/const_reborn/status/2032002840961442279](https://x.com/const_reborn/status/2032002840961442279)

> Basically all subnets on Bittensor are being mined by teams which heavily use AI optimized solutions, scripts that evaluate software submissions or models with multi layer inferences to update them and even higher level scripts that adapt those sub level optimizers continuously. In Bittensor’s final form no one will ever write a miner client. It all just becomes evals + incentives and swarms of agent miners unleashed against them. Literal armies of recursive agent scripts and LLM calls optimizing all aspects of every submission ad infinitum.
>
> — 12 Feb 2026 · 232 likes · reply · [x.com/const_reborn/status/2021987989476253978](https://x.com/const_reborn/status/2021987989476253978)

> Within 1 minute a GPU provided by an agent on @lium_io is utilized by another to train on @MacrocosmosAI.
>
> Both are paid by their own markets: the top sells into gradient value and the lower into GPU time.
>
> Thus the glue is agents and the coordination is TAO.
>
> — 6 May 2026 · 546 likes · original · [x.com/const_reborn/status/2051937551582150893](https://x.com/const_reborn/status/2051937551582150893)

### 4.5 UBI is a bribe

Section 5 of the thesis says every proposed cure runs through a government. Your tweets go further: UBI is the price of surrendering power, "a fiat drip". It is a stronger line and it ties the labor-power argument to ownership.

> UBI is what they will offer you in return for control, it’s a bribe, a contract you will sign to solidify your agreement with the loss of power. 
>
> Humans know, we all viscerally know that freedom, sovereignty and ownership are paramount and indeed requirements of real wealth.
>
> — 6 Jun 2025 · 127 likes · quote (quoting @haider1) · [x.com/const_reborn/status/1930927421651296503](https://x.com/const_reborn/status/1930927421651296503)

> The elephant to the nth power isn’t an elephant anymore, it’s a hockey stick. 
>
> Which is why Sam Altman promotes worldcoin, he thinks he can level it with UBI
>
> All of us suckling off the tits of the UBI drip 
>
> Nanny state to the Nth degree with Ai, as the “big mother”
>
> Devouring not just word cells and blue collars. 
>
> It fattens us all up for the slaughter.
>
> — 23 Sep 2023 · 37 likes · reply · [x.com/const_reborn/status/1705597603910480041](https://x.com/const_reborn/status/1705597603910480041)

### 4.6 The human faces of permissionless

The thesis argues permissionlessness in the abstract. You have the faces: miners who never finished high school, people in jungles and small towns, a 14-year-old who built a trading subnet, a prison escapee. In Montreal, a room of researchers, this is the emotional counterweight to "University is a scam".

> Some of Bittensors best miners haven’t even finished high school.
>
> — 29 Jun 2025 · 229 likes · quote (quoting @polynoamial) · [x.com/const_reborn/status/1939272700498686084](https://x.com/const_reborn/status/1939272700498686084)

> The protocol company is a long on global talent pools not needing to be regionally centralized in any way. 
>
> Mine intelligence at the cutting edge of innovation; live in your small town on the edge of the world.
>
> — 11 May 2025 · 162 likes · quote (quoting @casper_hansen_) · [x.com/const_reborn/status/1921482648645423482](https://x.com/const_reborn/status/1921482648645423482)

> To use the jargon of the day: 
>
> Mantis built “auto research” but for trading signals.
>
> Basically an optimization game where agents submit signals, those signals are measured for their information-theoretic value and those aggregate into a meta trading model. 
>
> Because it’s permissionless (any one can contribute from anywhere without asking) you can have ML agent or their operators play the game simultaneously 
>
> To double down further, it’s incentivized: meaning they can get paid doing it and the best contributors to the “auto research” are paid in ownership of the mechanism that produced the research.
>
> Then also, anyone can transfer that too cause it’s on a blockchain. Which means your auto research is freedom technology aside from being cool AF.
>
> Btw ps this was made by a 14 year old.
>
> — 21 Mar 2026 · 245 likes · quote · [x.com/const_reborn/status/2035388858464907573](https://x.com/const_reborn/status/2035388858464907573)

> @micaelabazo I believe in the invisible billion geniuses of this world
>
> — 12 May 2025 · 75 likes · reply · [x.com/const_reborn/status/1921842177732206716](https://x.com/const_reborn/status/1921842177732206716)

### 4.7 Corporations end in bureaucracy; protocols scale themselves

The thesis's "a lab can never make this promise" (section 12) is about the key. Your tweets make a second, economic claim: the corporation is an inefficient way to organize compute and talent, and a market beats it on allocation. That is the Tensorplex "50x smaller" story and the "allocation not compute" line.

> Bureaucracy is the end-state of all corporations.
>
> They simply don't have the tools to build competitiveness directly into the power structure.
>
> Dynamic TAO.
>
> — 22 Jun 2024 · 163 likes · quote · [x.com/const_reborn/status/1804509952175944018](https://x.com/const_reborn/status/1804509952175944018)

> Your incredulity is understandable, but wrong: 
>
> Uber, Airbnb, Bitcoin etc, all of them are showcases of low or unpermissioned incentive designs which took the world by storm. 
>
> Uber is an incentivized protocol
> Airbnb is an incentivized protocol 
> Bitcoin mining is an incentivized protocol
>
> Their global, unpermissioned, incentivized, nature **is** their killer app. Why? because it allows them to efficiently procure, allocate, and organize resources.
>
> Allocate, procure, and organize resources in a way that greatly exceeded their competitors classical organizational designs ("MSFT, OpenAI, APPL, NVDA") -- in their domain.  
>
> That was no coincidence. The biggest problem facing an organization is how to organize its people and its capital.
>
> Cryptocurrency goes even further, as it excels at the creation of organizational protocols for human, digital, or other resource, using the most expressive language-of-value the world has ever seen. 
>
> Abstractly, Bittensor is just a toolkit for creating such systems: "a protocol for the creation of incentivized protocols" -- we use it for AI.
>
> Referring back to your quote tweet re ASICs: Bittensor subnets are micro-directed-incentive designs, each 'specific' to a certain digital commodity: hashes, inference speed, storage, training etc.
>
> This makes them similar to ASICs since ASICs are circuits designed for the production of 'specific' computational challenges. 
>
> Hope that helps.
>
> — 3 Apr 2024 · 448 likes · reply · [x.com/const_reborn/status/1775543621393268912](https://x.com/const_reborn/status/1775543621393268912)

> The best thing about making a mechanism instead of a company is that it scales itself.
>
> — 17 Jul 2026 · 275 likes · quote · [x.com/const_reborn/status/2077912333939073147](https://x.com/const_reborn/status/2077912333939073147)

### 4.8 Coopertition: subnets mine each other

A network of markets that buy from each other is a different picture from "128 subnets". It also answers the emissions-versus-revenue critique: internal demand is real demand.

> Subnets will mine each other the same way Taohash mines Bitcoin -- pulling alpha from one subnet to fund the other.
>
> First movers will be compute subnets which focus on higher-order commodities. 
>
> Compute -> inference -> training -> optimization.
>
> All internal.
>
> — 7 May 2025 · 523 likes · original · [x.com/const_reborn/status/1920088171615846619](https://x.com/const_reborn/status/1920088171615846619)

> The real race is toward equilibrium.
>
> Bittensor is already achieving it. Subnets will soon mine each other — commoditizing every layer of the stack, cutting costs, and letting upper layers scale horizontally.
>
> Look at Celium: compute pods priced at fractions of centralized clouds. These pods aren’t just cheaper — they’re perfectly matched to higher-order markets like Templar, Chutes, or 9/19.
>
> Only at these prices do commodities like 19 and 64 become truly viable and profitable.
>
> And when those margins click, so do the economics of reinforcement learning networks — or any consumer of models and inference.
>
> We’re building the full stack — bottom-up — interconnected by mining.
>
> When we internalize that, it’s no longer a competition. We’re markets for one another.
>
> Allies, even — united by the goal of beating centralized control with open systems.
>
> gm.
>
> — 29 May 2025 · 137 likes · reply · [x.com/const_reborn/status/1928006470206984312](https://x.com/const_reborn/status/1928006470206984312)

> 2026 is already becoming Bittensor's Year of Integration.
>
> There are no explicit incentives for teams to work together. It's just happening. 
>
> "survival of the fittest" -> "survival of the collaborative"
>
> — 9 Jan 2026 · 520 likes · original · [x.com/const_reborn/status/2009692868852486594](https://x.com/const_reborn/status/2009692868852486594)

### 4.9 Realpolitik, legitimacy and why a protocol is the answer to power

Your January 2026 thread is the political theory under the "third path". Legitimate systems get voluntary cooperation (cheap); coercive systems must keep buying force (expensive). A protocol that assumes bad actors and still produces fair outcomes is the only design that survives both.

> But that's not the end: the absolute power of Realpolitik, of Rome for instance, begets the creation of new legitimate systems of power designed to remain fair even under adversarial conditions.
>
> Where Rome fell, the legitimacy of Christianity followed.
> Where the US falls, mechanisms such as Bitcoin will follow.
>
> and if we are to make a new world with this knowledge, our next systems must not deny power politics, exploitation, deception, Trumpism -- they must assume them. 
>
> This is why digital currency as a political tool is so profound: A language of power and money that we can use cryptographically constrain these forces.
>
> Embedding fairness, neutrality, and shared rules directly into the algorithmic systems themselves.
>
> The goal is not to eliminate the struggle for power, but to internalize it -- so that even self-interested, adversarial behavior produces legitimate outcomes.
>
> Do not fear the exploit. Design for it
>
> — 4 Jan 2026 · 95 likes · reply · [x.com/const_reborn/status/2007605197359460726](https://x.com/const_reborn/status/2007605197359460726)

> It’s not regulation which saves us. EU style bureaucracy just ensures advancement doesn’t birth itself in that ecosystem. 
>
> Corporatism is the most capital efficient solution and the clear winner in 2026. Yet birthing this in either jurisdiction, China or USA through the corporation will certainly lead to an unholy union of state corporate fascism (already organizing itself predictably out of Silicon Valley). This is just because there are such strong self reinforcing qualities (monopolization is always the goal and likely outcome of a corporation). 
>
> The first (regulation) is not a route at all because it nullifies itself and the second (corporate control) is dystopian.
>
> There is a third i believe which builds the control structure and capital formation structure for AI using the immutable tooling of protocol technologies, Bitcoin being the prime example of a tech which avoids monopolization through its design. 
>
> “Can’t monopolize and centralize” while simultaneously “can’t be stultified and is efficient” must be the carrying qualities of such a system.
>
> — 18 Jan 2026 · 83 likes · reply · [x.com/const_reborn/status/2012688518401699887](https://x.com/const_reborn/status/2012688518401699887)

### 4.10 "Bittensor is a machine learning algorithm"

You have said the network itself is one big learner: the weights matrix on chain is trained by validator–miner interaction. This links the neuromorphic inspiration in section 7 to the actual design.

> Neural networks self-organize their state in a loss landscape defined by a reward model.
>
> Bittensor subnets organize miners in an incentive landscape defined by a validator.
>
> It turns out that AI and crypto share preference in the tech stacks that make them click.    
>
> Thats the stack.
>
> — 29 Apr 2025 · 200 likes · quote · [x.com/const_reborn/status/1917189281191223621](https://x.com/const_reborn/status/1917189281191223621)

> Bittensor’s core consensus algorithm is called “Yuma” which is effectively a neural layer converting stake at the first layer to incentives (mapped over the miners) at the second layer. 
>
> But the art, and the field of neural incentive design is just beginning: Yuma is the perceptron of 1943. 
>
> Bittensor will require a PyTorch moment for incentive designers to reach its full potential.
>
> — 17 Aug 2024 · 197 likes · original · [x.com/const_reborn/status/1824894015990341665](https://x.com/const_reborn/status/1824894015990341665)

### 4.11 Privacy: the one thing the labs cannot sell you

Chutes and Targon run inference inside trusted hardware. You have made "cryptographically private models" a 2026 talking point. It is a concrete "what a centralized lab cannot offer" that the thesis lacks.

> Everyone should know about what Chutes is doing. 
>
> Fully permissionless inference mining. 
> Fully end to end encrypted. 
> Fully private TEE machines. 
>
> You could safely send private keys over the wire and know it was fully private. 
>
> The entire stack encrypted from your machine to the LLM and back. 
>
> The fact that that happens on top of a permissionless network with infra run my god knows who is nothing short of mind boggling.
>
> — 26 Mar 2026 · 572 likes · quote · [x.com/const_reborn/status/2037262495522455758](https://x.com/const_reborn/status/2037262495522455758)

> It’s not long before they start KYC’ing you for model use under the auspices of “security”
>
> — 4 May 2026 · 324 likes · original · [x.com/const_reborn/status/2051391441566351495](https://x.com/const_reborn/status/2051391441566351495)

### 4.12 Volatility as a filter, price as illusion

If the market is down on 28 September, this is your language for it. The thesis has no line for a bad price day.

> Volatility is the filtration mechanism which optimizes the machine.  
>
> Be in crypto for the right charts. https://x.com/const_reborn/status/1986786783347036552/photo/1
>
> — 7 Nov 2025 · 177 likes · original · [x.com/const_reborn/status/1986786783347036552](https://x.com/const_reborn/status/1986786783347036552)

> @mentatminds In all seriousness: we’re here because we believe in the power of markets to optimize technology. It is precisely the waxing and waning of tides that powers that machine. Devs don’t need to do anything, market doing it for us.
>
> — 21 Nov 2025 · 260 likes · reply · [x.com/const_reborn/status/1991877989316104239](https://x.com/const_reborn/status/1991877989316104239)

---

## 5. Tensions

Places where your own tweets contradict, complicate, or pre-empt the thesis. Quoted so you are not surprised on stage.

### 5.1 The key: you have promised to give it up before, with dates that passed

The thesis says "a small group holds it, and it runs through me" and gives a December 2027 ossification date. The record of earlier promises:

- "I will be giving away my personal power. Back to the community by vote. More serious details coming this week." — 18 Nov 2023, 157 likes, [link](https://x.com/const_reborn/status/1725967728727498877)
- "The chain operates on a Proof of Authority (PoA) mechanism, with plans to transition to Proof of Stake (PoS) by the end of next year." — 22 Sep 2024, 89 likes, [link](https://x.com/const_reborn/status/1837975797757575557) That meant end of 2025. The thesis's own note says nominated proof of stake is still about a year away.
- "We are well positioned to be the most decentralized POS chain around. Bet on that." — 3 Jul 2024, 367 likes, [link](https://x.com/const_reborn/status/1808602219551600658)
- "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 )" — 12 Feb 2026, 379 likes, [link](https://x.com/const_reborn/status/2022093414293286953)

A hostile questioner will read the 2024 date back to you. Have the reason it slipped ready. Also decide how the Feb 2026 change squares with "it runs through me" in the thesis. If legal authority already moved, say to whom.

### 5.2 The key as a feature, said in your own words

The thesis frames the key as "a bug we are removing". You have also defended it as the thing that saved the network and the thing that lets you fight bad actors:

> Democracy was the product of centralization. 
>
> Our goal has been to myopically focus on economic decentralization since this will underpin all decentralization to come. 
>
> We are well positioned to be the most decentralized POS chain around. Bet on that. 
>
> Regarding this attack, it was precisely because we had the ability to firewall the chain that we could mitigate the damage. 
>
> More information coming to all. 
> Thank you for your patience. 
> -- const.
>
> — 3 Jul 2024 · 367 likes · original · [x.com/const_reborn/status/1808602219551600658](https://x.com/const_reborn/status/1808602219551600658)

> @here4impact I will continue to upgrade Bittensor and stop bad actors from hurting this network until the day I die. 
>
> If your investors are looking for a investment that simply rolls over and dies, this ain't it.
>
> — 20 Dec 2025 · 164 likes · reply · [x.com/const_reborn/status/2002473021093871935](https://x.com/const_reborn/status/2002473021093871935)

> @badenglishtea @slopez32185 @here4impact Triumvirate and soon the two validator + subnet houses. Technically i only have soft power, which i do wield with the violence of someone that lives this blood sweat and tears. While that lasts I intend to use it.
>
> — 20 Dec 2025 · 48 likes · reply · [x.com/const_reborn/status/2002481030150885738](https://x.com/const_reborn/status/2002481030150885738)

"Until the day I die" and "wield with the violence" are the opposite of ossification. Pick one register for the stage.

### 5.3 "Not comparable to Bitcoin", in your words, since 2024

Good news: you have said it yourself, twice, before the June 2026 statement the thesis cites.

> I don't agree with @saifedean about who we are -- we are not scammers and gullible gamblers -- but he is right to defend the important difference in emission immutability between Bitcoin and other tokens, including Bittensor.
>
> Bitcoin is quite literally the industry shield we all work behind and was specifically designed  to protect against fiat currency and debasement. 
>
> Bittensor is on a 21M cap and emission schedule, but we cannot pretend for a second that at our level of decentralized immutability we are comparable to Bitcoin.
>
> That being said, Bittensor is still in our first halving: in time we can and will ossify and calcify the same level of social and technical changes into the protocol that Bitcoin has and attain a similar degree of immutability -- its just a matter of time and work.
>
> — 19 Jun 2024 · 225 likes · reply · [x.com/const_reborn/status/1803379237812052268](https://x.com/const_reborn/status/1803379237812052268)

> Id love to be more decentralized than Bitcoin but it simply isn't the case. Indeed, there is simply non-comparison. Bitcoin is the decentralized shield that protects the whole crypto space. We are like its little brother making sand castles behind the wall.
>
> — 22 Sep 2024 · 21 likes · reply · [x.com/const_reborn/status/1837989110927048899](https://x.com/const_reborn/status/1837989110927048899)

### 5.4 Fair launch and the VC sales

The thesis says: "No venture round bought a slice before the public could." Your March 2026 tweets give the detail a questioner will have:

> @chang_defi Tao never premined anything. Especially not to VCs. We sold tokens OTC in 2021 and 2022 which we personally mined.
>
> — 26 Mar 2026 · 642 likes · reply · [x.com/const_reborn/status/2037061646158372940](https://x.com/const_reborn/status/2037061646158372940)

> @s_mtv28 @chang_defi ~600k Tao towards three entities Firstmark, DCG and Polychain from 2021 ending in 2023. Those funds helped us build the team behind OTF.
>
> — 26 Mar 2026 · 78 likes · reply · [x.com/const_reborn/status/2037071940049842306](https://x.com/const_reborn/status/2037071940049842306)

> Id say probably around $18. It was a good trade for us at the time and a good trade for them in the long run. For the most part they have been very helpful partners and we are happy they are still here, but ultimately if they leave (and they have sold somewhat over the years in tranches, some more than others) — then that’s their prerogative like all Tao holders.
>
> — 26 Mar 2026 · 53 likes · reply · [x.com/const_reborn/status/2037073629641392516](https://x.com/const_reborn/status/2037073629641392516)

> As for OTF budget and funding. We didn’t premine, so Ala and myself donated a good portion of what we mined from early days. 
>
> Then the community kicked in and staked to our validator allowing us an 18% take on staking yield for the next few years. 
>
> Those funds went more or less 100% to running chain infrastructure and paying employees. We had 30 employees at peak. 
>
> Since the foundation is a Canadian not-for-profit it would have been illegal to take a profit. There are no equity holders and you can’t remove funds.
>
> — 30 Mar 2026 · 271 likes · reply · [x.com/const_reborn/status/2038718347571794049](https://x.com/const_reborn/status/2038718347571794049)

This is consistent with "no premine" but not with "no venture round bought a slice". The tokens were mined and then sold OTC to three funds at about $18 during 2021 to 2023, before most of the public knew the network existed. Fix the thesis line to "no tokens were printed for anyone; we sold some of what we mined to fund the foundation" and say the number before someone else does.

### 5.5 The wallet split and the blackmail

You disclosed this yourself in August 2025. Anyone doing a hostile Q&A will have read it. Know that it is public. The "almost 30 million dollars" figure in it refers to an earlier incident (the 2023 "cabal attack" you have mentioned), not the July 2024 drain the thesis puts at about $8M. Be ready to keep the two apart.

> I’m going to highlight this tweet ( and burner account) with one like and bring it out to 18k followers. 
>
> I decided a long while ago that if I ever get blackmailed i was going to tell the truth. Blackmail is used to capture and corrupt and I’d rather face the crowd then be manipulated by evil people. 
>
> Story: 
>
> Between the Nakamoto and Finney chains (when accounts were migrated) Ala and myself split our funds into a set of secondary wallets to protect our identities. This was a stupid idea (it didn’t protect our wallets).
>
> Later, a developer that worked for the foundation would use this to blackmail me. They wanted to leverage this information but when he couldn’t, he would go and use keys from the foundation to hack almost 30 million dollars from people in the network. Ala and I payed for those hacks from personal funds and paid our pound of flesh. 
>
> These screenshots (in below tweet) are between Carro and I at the time when Carro had been manipulated to side with the people blackmailing me. It was very personal and heated. He got over it when he realized they were criminals who just wanted money. 
>
> All in all, you can see where all the funds came from if you check the public records (for instance it has been fixed on taostats), i.e. nothing minted etc.
>
> That still doesn’t change the fact that splitting our funds was an ugly choice and not something I can defend — as it muddied the chain history. I deeply regret it. 
>
> People make mistakes. I have made errors. i am human, Bittensor is human and I won’t let these fucks use it to have me lose my truth @gruntxFiishher 
>
> Next.
>
> — 22 Aug 2025 · 330 likes · quote · [x.com/const_reborn/status/1958898625825689769](https://x.com/const_reborn/status/1958898625825689769)

### 5.6 Root: dead, then back

Three emission regimes in sixteen months is already in the thesis's weak points. The quotes:

- "Root is dead, long live Bittensor" — 7 Feb 2025, 636 likes, [link](https://x.com/const_reborn/status/1887993127979344126)
- "root is coming back, and it's going to be glorious" — 15 Jun 2026, 636 likes, [link](https://x.com/const_reborn/status/2066615306944917611)

And your own rule for changing rules, from Dec 2025: "I agree that we should not arbitrarily change rules, that slow is steady and steady is fast, and yet, there is also reality of upgrading this network in a way that is positive." — 20 Dec 2025, 83 likes, [link](https://x.com/const_reborn/status/2002479397442339217)

### 5.7 Emissions versus revenue, in your words

The thesis quotes $28 to 35M ARR from outside customers. In December 2025 you said:

> "generate revenue too quickly" -- I don't think any subet on Bittensor is generating flow through revenue in any serious way. It is all speculative investment. So the idea that flow emission has pushed people towards making revenue is some sort of sick meme which persists from ignorance.
>
> — 4 Dec 2025 · 39 likes · reply · [x.com/const_reborn/status/1996632415633043965](https://x.com/const_reborn/status/1996632415633043965)

And in August 2025:

> @Pritensor What matters is not the quantity of subnets but the distribution of emission. Top three subnets on Bittensor produce revenue. Two of them are emission positive and the third serves more token on open router than any competing provider.
>
> — 14 Aug 2025 · 12 likes · reply · [x.com/const_reborn/status/1956021150443606291](https://x.com/const_reborn/status/1956021150443606291)

And in February 2026:

> @AlgodTrading Inference is highly subsidized cross industry. 40% cost coverage is a great place to be compared to labs at 1-2%. As the market for tokens reaches maturity everything will rise to >100%. At that point we will be at X multiples.
>
> — 19 Feb 2026 · 31 likes · reply · [x.com/const_reborn/status/2024277486700253589](https://x.com/const_reborn/status/2024277486700253589)

Reconcile these before the talk. "Inference is highly subsidized... 40% cost coverage" is a defensible and honest framing that beats the ARR number.

### 5.8 Competitors: "we're in this together" versus "churches without faith"

The thesis calls Bittensor "the only thing of its kind". You have both embraced and attacked the others. All of it is public.

> One unfortunate irony in the world of crypto is a belief in decentralization and yet the tendency towards myopic deference to a single protocol as the absolute leviathan under which the world will be set free. 
>
> but no, the antidote to totalitarianism is the dispersion of power, and the explosion of different hierarchies competing to best express our common morality.
>
> At times I have lost sight of this and have refused to acknowledge the work of other decentralized AI protocols. True hypocrisy. 
>
> The bellow are a few, but not all, seeking this same thing. Koodos to these communities pushing for the freedom and decentralization AI.
>
> @communeaidotorg
> @gensynai
> @ritualnet
> @akashnet
>
> If you know of others, please post them in the comments.
>
> — 27 Nov 2023 · 347 likes · original · [x.com/const_reborn/status/1729163079273922892](https://x.com/const_reborn/status/1729163079273922892)

> "assumes the quality bar of something like @tplr_ai is currently the same as @primeintellect @NousResearch" 
>
> Yeah it's not. Our 8b runs beat Nous 40 while they claim decentralized training on a closed network with whitelists and without incentives. 
>
> Templar is live, unpermissioned, using distributed training mechanism superior to both (CCLoco). No premine, no fucks. Launching a 70B on Wednesday.
>
> @teknium I gave you your first break. 
> @vincentweisser I explained mining incentives to you before you launched prime. 
>
> You're currently churches without faith. 
> Lets go. Launch your tokens and lets mine.
>
> — 19 Aug 2025 · 236 likes · quote · [x.com/const_reborn/status/1957594544691703953](https://x.com/const_reborn/status/1957594544691703953)

> I'm fighting now because no way Im going to let these same VC SF cunts who laughed in my face for years when I said we could mine intelligence come and pretend they invented it first while selling it out to fiat money. 
>
> You're gonna have to kill me.
>
> — 20 Aug 2025 · 704 likes · original · [x.com/const_reborn/status/1958238281675440451](https://x.com/const_reborn/status/1958238281675440451)

> Congrats to Gensyn on launching their network for machine intelligence, using economic primitives to push forward and open the frontier of decentralized AI.
>
> Economic primitives will allow us to compete and build along side the techno-giants, organizing the world's talent and compute more competitively, efficiently, and fairly.
>
> Their inherent openness allows us to connect in games where competition becomes cooperation towards a shared goal in larger and larger networks.
>
> For this reason, no permissioned company can compete, and ultimately we will win a world where no technocrats can close the doors on us mere mortals  accessing the fruits of intelligence.
>
> Bravo to Ben and Harry. Welcome to the thunderdome.
>
> p.s. Looking forward to Bittensor miners and subnets joining in on the fun.
>
> — 9 Dec 2025 · 152 likes · quote · [x.com/const_reborn/status/1998445373795266679](https://x.com/const_reborn/status/1998445373795266679)

Your actual distinction, which holds up, is permissionless plus incentivized versus whitelisted and unpaid: "no. they are not permissionless runs nor incentivized. They are whitelisted runs. You need to sign up. And there is no incentives, so the miners are not looking to exploit. Verification is not solved yet." — 13 Jun 2025, 11 likes, [link](https://x.com/const_reborn/status/1933435634145792194) Say that, not "only thing of its kind".

### 5.9 Governments and geopolitics: "no country" versus "the western project"

The thesis says Bittensor is outside the US and China. Your tweets mostly agree, but some pull toward a Western or American frame, and some are hot politics.

> The “west” is just the advocates of pluralism. 
>
> From the city states of Ancient Greece, to the senators of Rome, through the Christian nations of Europe, in the divided houses of US, and finally under the immutability of blockchains. 
>
> The western project is simple and continuous: to create decentralization under a protocol. 
>
> It is not the mythos of China, nor Persia, nor Russia — it is our unique project.
>
> — 17 Jan 2026 · 100 likes · original · [x.com/const_reborn/status/2012569445332304280](https://x.com/const_reborn/status/2012569445332304280)

> The US building a strategic Bitcoin reserve on the precipice of losing the world reserve currency is the most clutch sleight of hand in history.
>
> — 6 Nov 2024 · 326 likes · original · [x.com/const_reborn/status/1854221337507610862](https://x.com/const_reborn/status/1854221337507610862)

> Food for thought: 
>
> “The worst thing you can possibly do in a deal is seem desperate to make it. That makes the other guy smell blood, and then you’re dead.” - Trump
>
> The tariffs are a classic Art of the Deal, strategy, Trump is gaining leverage. 
>
> Over the coming months he will roll them back in exchange for deals claiming victory. 
>
> It might even work. It’s not the deranged chaos the markets feel it is. Indeed, that chaos is exactly the leverage Trump wants. 
>
> That’s his game.
>
> — 3 Apr 2025 · 319 likes · original · [x.com/const_reborn/status/1907806430528262302](https://x.com/const_reborn/status/1907806430528262302)

> #FreePalestine
>
> — 27 May 2025 · 1,025 likes · reply · [x.com/const_reborn/status/1927302944283591101](https://x.com/const_reborn/status/1927302944283591101)

> Looks like you’ve chosen your friends @vincentweisser, Palantir, Anduril, a crew that would profit off having the whole world end up like Gaza.
>
> Bitcoin is FREEDOM money. Not in league with the companies that sell your information to the NSA. 
>
> Which way western man? https://x.com/vincentweisser/status/1957608445655581127
>
> — 19 Aug 2025 · 159 likes · original · [x.com/const_reborn/status/1957609755427647665](https://x.com/const_reborn/status/1957609755427647665)

For Montreal: the "west" tweet reads as civilizational. The Palestine and Palantir tweets are on the record and will be brought up if the talk touches Gaza. The thesis's open question 1 (the Gaza image) is bigger than the thesis knows: you have used "We become Gaza" and "the whole world end up like Gaza" on X, so the image is already tied to your name. Decide deliberately.

Also for Montreal: "University is a scam. Learn how to mine Bittensor." — 28 Apr 2026, 1,092 likes, [link](https://x.com/const_reborn/status/2049218909698396522) (1,092 likes) and "Exploit the system. Go straight to the workforce. Mine Bittensor." — 28 Apr 2026, 416 likes, [link](https://x.com/const_reborn/status/2049220629849890822) Half the room is affiliated with a university. Either own it or leave it out.

### 5.10 Price

The thesis states market cap honestly. Your tweets treat price as noise, then celebrate it:

- "Forget charts and be bullish only on the truth." — 30 Mar 2024, 628 likes, [link](https://x.com/const_reborn/status/1774097399428038858)
- "Price is the illusion. What matters is energy, mission." — 8 Aug 2025, 6 likes, [link](https://x.com/const_reborn/status/1953846117352845564)
- "I love the smell of a bear market in the morning" — 14 Nov 2025, 295 likes, [link](https://x.com/const_reborn/status/1989323270735175851)
- "do you smell it? Bittensor is cooking." — 24 Jun 2026, 1,153 likes, [link](https://x.com/const_reborn/status/2069571359777153318)
- "Up 30% on investment in 4 days which is pretty wild." — 12 Mar 2026, 79 likes, [link](https://x.com/const_reborn/status/2032021749332656367)

"Make holders hold" is a goal of the talk. Your strongest holder line is not about price; it is "It’s all coming along the way it should to create a 1000 year project, not a pump. / Just watch. Patience." — 29 Dec 2024, 220 likes, [link](https://x.com/const_reborn/status/1873462217653494233)

### 5.11 Covenant, Subnet 3, and selling alpha

Already in the thesis. Your own account, so the words match:

> Btw i should address some of the points in the article which AI slop influencers are now repeating. 
>
> > I suspended emissions to his subnets
>
> 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. This changed the emission the same way all buys and sells on Bittensor do. I don’t have any privilege beyond what normal TAO holders have. 
>
> > I deprecated Covenant’s channels and removed moderation rights
>
> No, Samuel specifically deprecated his own channels which he did via a pinned comment and also a Twitter post. To be clear, he is talking about a discord channel.
>
> > I removed his ability to moderate his community
>
> Sam was deleting posts in his channel of genuine, honest criticism. I removed that ability temporarily and then reinstated it later. I did not remove his moderator role. I simply stopped him from deleting posts from others in his channels. 
>
> > “deprecating infrastructure”
> Not even sure what this one means. 
>
> > large visible token sales 
> Not large. Less than 1% of what i had invested in his teams. 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.
>
> Anything else?
>
> — 10 Apr 2026 · 1,673 likes · reply · [x.com/const_reborn/status/2042508660723581079](https://x.com/const_reborn/status/2042508660723581079)

> This is a genuine criticism against me. If i have blinders it’s that i over look negatives and focus on positives in an effort to bring forward peoples good characters. I loved Samuel because he had energy and drive, commitment and an unbelievable work ethic. What i overlooked was his chaotic and irrational rage which ultimately led him to make this horrible decision. Neither of those qualities excludes the other. Besides all the FUD, Bittensor still owes Samuel thank you for his unending drive to bring the largest decentralized training run to subnet 3 — it would not have happened without his tireless work. And now we owe him a thank you for showing the protocol how to move forward with subnet owner decentralization. I hope he is well. Bittensor is also fine. Let it be done.
>
> — 10 Apr 2026 · 677 likes · reply · [x.com/const_reborn/status/2042548758978625786](https://x.com/const_reborn/status/2042548758978625786)

Note the tension between "I do not have the ability to suspend emissions" and, three months earlier, "we introduced Taoflow to damage non productive subnets which rest only on price — it worked and 28 has less emission." — 20 Dec 2025, 16 likes, [link](https://x.com/const_reborn/status/2002475591761866885) One says you have no lever; the other says you built one. Both are true at different layers. Be able to explain which.

### 5.12 "Bittensor pays for contribution, not need"

Thesis weak point 15. You have already conceded it: "It is reductive to assume all human problems can be encapsulated as incentive hill climbs." — 20 Dec 2024, 5 likes, [link](https://x.com/const_reborn/status/1870135335658598470) Use it. It disarms the question.

### 5.13 Off-topic landmines

Not tensions with the thesis, but things a journalist will have.

- Feb 2025 thread on Grimes, Musk and "promiscuous women" (many replies). Mar 2026 "blue haired fat chicks". Both are on the record.
- Dec 2024 "God's programming language" and your own walk-back.
- "aliens are real, i have proof" (Mar 2026, a joke inside a serious post).
- Sep 2025 tweet on Israel and the covenant.

None of these belong in the talk. Know they exist.

---

## 6. Top 30 by engagement

From the summary file, with one line each on why it landed.

1. 13 Jun 2026 · 2,588 likes · [link](https://x.com/const_reborn/status/2065606907855831301) — "@AnthropicAI And just like that we collectively saw the future of inequality". Landed because it was an eleven-word reply to a lab announcement that named the fear everyone in the replies was feeling. Your most-liked tweet ever is your core thesis in one line.
2. 11 Apr 2026 · 2,385 likes · [link](https://x.com/const_reborn/status/2043083033969033610) — X Article, 11 Apr 2026. Body not in the archive; posted the day after the Covenant dispute and almost certainly your long-form response. Landed because the community wanted the full account from you. Pull the text.
3. 12 Mar 2026 · 2,191 likes · [link](https://x.com/const_reborn/status/2032002840961442279) — "I have 6 agents running now..." Landed because it was concrete, personal and slightly alarming: a founder saying "I could literally die and this system wouldn't turn off."
4. 10 Apr 2026 · 1,673 likes · [link](https://x.com/const_reborn/status/2042508660723581079) — Point-by-point rebuttal of the Covenant article. Landed because it was specific ("Less than 1%") and did not hide behind PR.
5. 10 Apr 2026 · 1,324 likes · [link](https://x.com/const_reborn/status/2042427496751063191) — "Exploits are what teach a system its weak spots." Landed because it turned a bad night into a design principle and a new feature.
6. 24 Jun 2026 · 1,153 likes · [link](https://x.com/const_reborn/status/2069571359777153318) — "do you smell it? Bittensor is cooking." Pure vibe during a price run. Landed for the same reason all such tweets land.
7. 28 Apr 2026 · 1,092 likes · [link](https://x.com/const_reborn/status/2049218909698396522) — "University is a scam. Learn how to mine Bittensor." Landed as a provocation that fits a generation's mood. See Tensions 5.9 before using in Montreal.
8. 21 Jun 2026 · 1,088 likes · [link](https://x.com/const_reborn/status/2068760333787808043) — X Article, 21 Jun 2026. Body not in archive. Likely the June statement on decentralization the thesis cites. Pull the text.
9. 14 Aug 2026 · 1,047 likes · [link](https://x.com/const_reborn/status/2088349997645312205) — Subnet 3 retaken by holders, Teutonic-I released. Landed because it closed a four-month saga with a product, not a statement.
10. 27 May 2025 · 1,025 likes · [link](https://x.com/const_reborn/status/1927302944283591101) — "#FreePalestine". Landed far beyond the TAO audience. It is your only pure politics tweet in the top 30.
11. 23 Mar 2026 · 1,019 likes · [link](https://x.com/const_reborn/status/2035998840281612313) — "Tao is also nothing without BTC." Landed with Bitcoiners as much as with your own community. Pure lineage statement.
12. 10 Apr 2026 · 959 likes · [link](https://x.com/const_reborn/status/2042410716271948281) — "the first subnets on Bittensor that run headless and as true commodities." Landed because it reframed a founder-rug as the birth of founder-less subnets.
13. 14 Mar 2026 · 938 likes · [link](https://x.com/const_reborn/status/2032748075706425367) — "Bittensor will be run by agents." One sentence, full vision. Landed as a prophecy people wanted to repeat.
14. 11 Apr 2026 · 902 likes · [link](https://x.com/const_reborn/status/2043039054464156079) — "We will fix this at the protocol level and we are going to bring these subnets back to life." Landed as the promise the community needed on the night.
15. 3 Jun 2026 · 866 likes · [link](https://x.com/const_reborn/status/2062258366739550618) — 128-node run in 5 minutes for 5 TAO. Landed because the numbers are absurdly small for what they bought.
16. 24 Aug 2026 · 852 likes · [link](https://x.com/const_reborn/status/2091896286035235027) — Teutonic-II-110B announced. Landed because 110B is a frontier-lab number.
17. 3 Mar 2025 · 842 likes · [link](https://x.com/const_reborn/status/1896612307645698394) — "I can participate in Bittensor in a fair way. I'm making a subnet." Landed because the founder entering the arena as a player was a signal that dTAO was real.
18. 25 Mar 2026 · 819 likes · [link](https://x.com/const_reborn/status/2036718579006001308) — Intel shout-out to Manifold. Landed because a chip giant validated a subnet team's trusted-computing work.
19. 13 Jun 2026 · 788 likes · [link](https://x.com/const_reborn/status/2065914716589850671) — "The best teachers will be LLMs. The best feedback will be monetary..." Landed as a five-line manifesto. Stage-ready.
20. 29 Dec 2024 · 781 likes · [link](https://x.com/const_reborn/status/1873347340851146929) — "Limitless subnets / No registration fees / Untethered subnet tokens / Depermissioned EVM / Let the neural internet roar / Q1 2025". Landed as the dTAO roadmap in six lines.
21. 20 Jun 2026 · 752 likes · [link](https://x.com/const_reborn/status/2068136709439517112) — X Article, 20 Jun 2026. Body not in archive. Pull the text.
22. 28 Jun 2024 · 745 likes · [link](https://x.com/const_reborn/status/1806721647241802062) — "You're building an AI company. We're building a 1000 year Intelligence Federation. We are not the same." Landed as the sharpest us-versus-them line of 2024.
23. 12 Apr 2026 · 736 likes · [link](https://x.com/const_reborn/status/2043254173651579264) — "AI agent builds subnet from scratch > Beats SOTA 4B param model in 2 weeks > writes a paper about it." Landed because it showed the agent thesis working, not predicted.
24. 4 Apr 2026 · 717 likes · [link](https://x.com/const_reborn/status/2040459231187386627) — Kyle Samani dropped the call at "21M Bitcoin style supply curve". Landed because it made the fair-launch story personal and funny.
25. 20 Aug 2025 · 704 likes · [link](https://x.com/const_reborn/status/1958238281675440451) — "no way Im going to let these same VC SF cunts who laughed in my face..." Landed on raw anger. Do not quote on stage; know it exists.
26. 16 Apr 2026 · 677 likes · [link](https://x.com/const_reborn/status/2044662810794217771) — "No other place" (Hippius). Landed as a product flex with an image.
27. 10 Apr 2026 · 677 likes · [link](https://x.com/const_reborn/status/2042548758978625786) — "This is a genuine criticism against me..." Landed because you accepted a criticism in public and still thanked the person who hurt you.
28. 4 May 2026 · 650 likes · [link](https://x.com/const_reborn/status/2051370035046748283) — "making subnets on bittensor is just too much fun". Joy sells.
29. 24 Mar 2026 · 646 likes · [link](https://x.com/const_reborn/status/2036403074789453926) — "You guys should just speak to me instead of having no clue". Landed as a founder telling a podcast to call him. Led to the Threadguy appearance.
30. 7 Mar 2026 · 646 likes · [link](https://x.com/const_reborn/status/2030405496130113663) — "when software is cheap all that remains is digital commodities behind APIs". Landed as the clearest one-paragraph statement of the 2026 commodity thesis. Stage-ready.

---

## Method note

Counts come from regular-expression keyword matches on the 2,468 non-retweet tweets (see the per-theme numbers). Overlap between themes is expected; one tweet can count in several. I read every tweet in date order once, then verified each quoted tweet against the raw JSON at render time. If a quoted fragment did not match the source text exactly, the file would not have been written.