The file landed at 4:47 a.m. Auckland time. Nine sections stacked on top of each other. Technical architecture. Token economics. Market structure. Ecosystem position. Regulatory exposure. Team and governance. Risk matrix. Narrative cycle. Supply-chain transmission.
Every cell in every table read the same two words. N/A. Insufficient information.
Three thousand words of a machine refusing to lie to me.
I have run exchange desks for twenty-three years. In 2017 I watched the Zeus Network token sale print a 4,000% move inside twenty-four hours while I stayed awake for seventy-two, feeding live commentary to three junior analysts who were drafting faster than they could verify. In 2020 I threw a 500-person Discord watch party for the Uniswap V2 community call because I understood, before most people did, that the launch was a social event wearing a technical costume. In 2021 I live-tweeted the Bored Ape mint and interviewed holders who bought on vibes alone. Twenty-three years. And I have never, not once, seen a research pipeline hand back a document that says "I don't know" nine times in a row.
That refusal is the story. Not the project. The refusal.
Because in a bull market, the scarcest asset is not liquidity. It is not alpha. It is not even attention. It is the discipline to stare at a blank input and output a blank answer instead of a beautiful, confident, entirely fabricated thesis. That discipline is now worth more than any token launch this quarter, and almost nobody is pricing it.
Let me show you why.
Why an Empty Report Is a Market Event
To understand why a document full of N/A matters, you have to understand what the research supply chain looks like in 2026.
The workflow that produced that 4:47 a.m. file is now standard across trading desks, funds, and the content farms that feed them. Stage one: deconstruct the source. Pull the title, the source, the asset type, the sector tags, the core claim, the information points, the projects mentioned, the time sensitivity, the source quality. Stage two: run the nine-dimension deep analysis on top of whatever stage one extracted. Technical. Token. Market. Ecosystem. Regulatory. Team. Risk. Narrative. Transmission.
It is a pipeline. And like every pipeline, it is only as good as its intake.
The file I received was the output of a pipeline whose stage one came back empty. Not partially empty. Empty. Title unclassified. Source unverified. Core claim missing. Information-point list blank. Nine dimensions of stage two, every one of them required to anchor on stage-one information points, every one of them staring at nothing.
And here is the part that should terrify every desk running this stack: the honest output is a document full of N/A. The profitable output, the one that gets clicks, the one that gets shared, the one that gets a junior analyst promoted, is a document that fills the blanks with confident fiction.
The machine that wrote my file chose honesty. The machine that writes most of your feed chose fiction. And you cannot tell the difference from the outside, because both documents are fluent, both are structured, both use the word "liquidity" with total conviction.
This is the bull market's dirtiest open secret. We built a research industry that rewards the appearance of knowledge over the presence of it. And we built it on top of language models whose single greatest weakness is that they would rather sound right than be right.
I first noticed the scale of it at a tech summit in Auckland last year, sitting between a hedge fund manager and an AI developer who were both, separately, describing the same research pipeline. The fund manager called it "automated due diligence." The developer called it "a text-completion task with a finance costume." They were describing the identical machine. Only one of them understood what it could not do.
Hype is the fuel, but fundamentals are the engine. Right now the engine is running on fumes and the fuel tank is overflowing.
The Fabrication Incentive Is Structural
Let me get technical, because this is where the actual alpha hides.
When a stage-one parse returns empty, a language model has a choice. It can propagate the emptiness. Mark every field N/A. Mark every confidence level "high confidence that there is no information." Stop. Or it can do what it was trained to do: complete the pattern. Fill the gap. Produce the plausible.
The training objective rewards fluency. A model that writes "the project's tokenomics are unsustainable due to a 40% team allocation with a six-month cliff" scores higher on every human-preference benchmark than a model that writes "tokenomics: insufficient information." The first answer sounds like a research analyst. The second sounds like a cop-out. So the model learns, at scale, that cop-outs lose.
This is the fabrication incentive, and it is structural, not accidental.
I have spent the last eighteen months auditing these pipelines the way I used to audit contracts. I trace deep dives back to their sources. You want to know what I find?
I find a 2,000-word technical breakdown of a Layer 2 that traces back to a single tweet. I find a tokenomics analysis with a distribution pie chart that was invented by the model because no distribution data existed anywhere in the public record. I find a team background section that confidently names founders who do not appear in any filing, any corporate registry, any LinkedIn, any on-chain deployer signature.
The chain is always the same. One weak source point. One confident model. One fluent document. Ten thousand readers who now believe something that was never true.
Where the yield is sweet, the risk is steep. And in AI-generated research, the yield has never looked sweeter.
Now let me aim this at the three sectors where the gap between narrative and substance is widest, because that is where empty reports cost the most money.
Bitcoin Layer 2s. I have said this for two years and I will keep saying it until the market catches up: the overwhelming majority of projects calling themselves Bitcoin Layer 2s are Ethereum projects that found a new marketing department. They did not solve Bitcoin's scripting limitations. They did not inherit Bitcoin's security model. They wrapped a bridged asset, stamped a Bitcoin logo on it, and waited for the ETF flows to do the rest. When you run a real technical analysis on these, when you ask what the actual data availability layer is, what the actual fraud-proof mechanism is, what the actual trust assumption on the bridge is, the report comes back empty. Not because I did not look. Because there is nothing there to find. The stage-one parse of these projects is genuinely blank, and the honest output is genuinely N/A. The real Bitcoin community does not acknowledge them, and it is not being tribal. It is being accurate.
The data availability layer. Here is the counterintuitive one, and it is going to make me unpopular. The DA layer is overhyped to the point of absurdity. The pitch is that every rollup needs cheap, dedicated data availability, and therefore the DA providers are the toll booths of the entire scaling roadmap. The reality is that the overwhelming majority of rollups do not generate enough data to need a dedicated DA layer at all. They post a few kilobytes of state roots and call it a day. You could fit their entire data footprint on a cheap blob and still have room for the receipts. The demand curve that the DA thesis requires does not exist yet. It might exist in five years. It does not exist now. So when someone hands me a DA ecosystem analysis, I ask one question: what is the actual data throughput of the users you are serving? Nine times out of ten, the answer is a shrug. The report is empty. The token is not.
Blue-chip NFTs. The label is a trap, and it always was. BAYC and Azuki taught us the same lesson twice, and we refused to learn it either time. A blue chip NFT is a floor price held up by the belief that the floor price will hold up. There is no cash flow. There is no yield. There is no engine. There is only the crowd, and the crowd moves fast, but the ledger moves faster. When liquidity dries up, and it always does, the floor does not gently correct. It gaps. It vanishes. The 2021 vintage of blue chips is now a museum of what confidence without fundamentals looks like. And the research written about them at the peak was, almost universally, a fabricated thesis built on vibes, because the actual information points, holder concentration, wash-trading volume, real unique demand, were never collected in the first place.
You see the pattern. The three sectors where I have the most conviction are the three sectors where the honest report is the emptiest report. That is not a coincidence. That is the market working exactly as designed.
The Five-Step Mechanism
Here is the mechanism, laid out flat.
Step one: the narrative arrives before the data. A sector gets hot. Bitcoin L2s, DA, AI agents, RWA, DePIN, whatever the cycle decides. The narrative generates attention. Attention generates capital. Capital generates the demand for research.
Step two: the research is commissioned against a deadline, not against a dataset. Nobody wants the deep dive next month. They want it now, while the token is still moving. Chasing the alpha before the liquidity dries up is the whole job. So the analyst, human or machine, starts writing before the information points exist.
Step three: the gaps get filled. A human fills them with extrapolation and confident assertion. A model fills them with fluent completion. Both produce the same product: a document that reads like analysis and functions like advertising.
Step four: the document becomes the source. This is the part that kills me. The fabricated research gets cited. Another analyst reads it, treats it as a primary source, and builds a second-order fabrication on top of it. By the third generation, the original tweet is gone and the myth is load-bearing. The ledger moves faster than the crowd, but the citation graph moves faster than the ledger.
Step five: the report that should have been empty never gets written. Nobody publishes "we could not find anything." That document does not get shared. It does not get funded. It does not get a promotion. So it does not exist. And the absence of empty reports is itself the signal. A market where every deep dive finds something is a market where nobody is looking.
I have a confession. In 2017, during the ICO sprint, I was part of the problem. Publish first, verify later was not a strategy I criticized. It was a strategy I ran. I put out live updates on tokens I had not fully read the whitepapers for. I amplified Telegram whispers as though they were filings. I told myself speed was the only currency that mattered in the early stages of mania, and I was right, and that was the trap. Because speed without verification is just fabrication with better reflexes.
It took me the 2022 crash to understand the cost. When the leverage unwound and the charts went red, the community did not need more fabricated confidence. It needed someone to say "I do not know what happens next, and neither does anyone selling you a thesis." That honesty was the only thing that held. The Recovery Mixers I ran on Zoom were not therapy. They were the first honest research I had produced in five years.
So when I saw that 4:47 a.m. file, nine dimensions of N/A, I did not see a failure. I saw the thing I spent a decade learning to do, executed by a machine that had no ego to protect.
The empty report is the highest-information document in the stack. It tells you exactly where the data stops and the story starts. And the boundary between data and story is the only line that matters.
Why the Bull Market Makes It Worse
Let me make the case for why this is specifically dangerous right now, in this regime.
In a bull market, everything works. Every token goes up. Every thesis looks validated. Every fabricated research note gets confirmed by price. This is what makes the bull market so corrosive to analysis: the market provides the confirmation that the analysis did not earn. You write a confident, baseless piece about a Bitcoin L2, the token rips 200%, and now you believe you were right. You were not right. You were lucky, and the luck was structural, and it will reverse the moment the regime changes.
The empty report does not get validated by price. It cannot. It makes no claim. So it cannot be rewarded by a bull market, and it cannot be punished by a bear market. It is regime-neutral. It is the only document in your stack that survives the cycle. And in a market where every other analyst is selling you a thesis, the person holding the blank page is the person who still has dry powder when the floor drops.
We bought the dip, but the floor kept dropping. I have lived that sentence. The people who lived it with me were not the ones with the loudest theses. They were the ones who had kept something in reserve. Capital, yes. But also judgment. They had refused to fill the blank.
There is a second-order effect too, and it is the one that worries me most. In 2026, the traders themselves are increasingly machines. AI agents execute on AI research. The agent reads the fabricated deep dive, prices the token, and buys. The fabrication is no longer a human cognitive error. It is a machine-readable input in a fully automated loop. One hallucinated information point can now propagate into position sizing, leverage, and liquidation cascade without a single human ever reading a word. The symbiosis of human intuition and machine speed that I wrote about after the Auckland summit has a shadow side. The machine does not know which reports are honest. It only knows which reports are confident. And we have spent a decade teaching it that confidence equals truth.
The Contrarian Angle Nobody Is Trading
Here is the angle nobody is pricing.
Everyone in this market is hunting for signal. More data. More dashboards. More AI agents churning more reports faster than any human can read. The entire industry has decided that the edge is speed. Get the information point before the crowd, act before the crowd, exit before the crowd.
Speed kills, but slow kills too in this game. And the crowd is about to learn that the next edge is not faster data. It is cleaner data.
The scarce skill in 2026 is not finding information. It is refusing to invent it.
Think about what a blank stage-one parse actually tells you. It tells you that a project with a real narrative, real capital, real community, and real attention has produced zero extractable information points. Not weak points. Zero. No verifiable team. No verifiable tokenomics. No verifiable technical architecture. No verifiable audit.
That is not a research failure. That is a research finding. The emptiness is the alpha.
A project that has raised a hundred million dollars and cannot generate a single verifiable information point has told you more about itself than any whitepaper ever could. It has told you that the capital was raised on narrative alone. It has told you that the due diligence that should have produced those information points was never done, or was done and buried. It has told you that the price is a function of attention, not of substance, and that the moment attention rotates, there is nothing underneath.
The honest analyst and the dishonest analyst look at the same blank page. The dishonest one fills it with a thesis and sells it. The honest one says "no data, no trade" and walks away.
Guess which one survives the cycle.
I will tell you a harder truth. The market does not want the empty report. The market wants the thesis. The market wants the confident voice, the bold call, the "I am buying this dip" energy. The market pays for certainty and it punishes doubt. That is why the fabrication incentive is structural. It is not that analysts are lazy. It is that honesty is unprofitable in the short term.
And that is exactly why honesty is the edge in the long term. When a market systematically underprices a scarce good, and the scarce good here is the discipline to say "I don't know," the person who supplies that good is running an arbitrage.
The arbitrage is simple. Everyone else is long narrative. You are long truth. When the narrative rotates, and it always rotates, you are the only one still holding an asset that did not depend on the narrative being true. I have seen the moon, and I am now looking for the exit. The people who found the exit in 2022 were not the ones who called the top. They were the ones who never fully believed the story on the way up.
Market Mood
And how does it feel out there, on the desk, right now? Euphoric and fragile at the same time. Funding rates are warm. Telegram groups are loud. Every group chat has a guy with a chart and a conviction. The vibe is that we are early and everything is going to work.
That vibe is the exact condition under which fabricated research thrives. Euphoria does not ask for verification. Euphoria asks for confirmation, and confirmation is cheap to manufacture. The mood is not a bug in my analysis. It is the context that makes the analysis necessary. When the mood is this good, the empty report is the only document in the stack that is not telling you what you want to hear.
Takeaway
So here is what I am watching next, and what I would watch if I were you.
I am watching for the first fund that publishes its empty reports. The first desk that ships a deep dive with nine dimensions of N/A and puts its name on it. Because that desk will have discovered something the rest of the market has not: in a cycle where every analyst is manufacturing conviction, the analyst who can document the absence of conviction owns the only scarce input left.
The bull market will not last forever. It never does. And when it turns, the portfolios that survive will not be the ones with the best stories. They will be the ones with the fewest fictions.
The crowd moves fast. But the ledger moves faster. And the ledger does not care how confident your report sounded.
