A research deliverable crossed my desk this Tuesday. Forty-one pages. Nine analytical dimensions โ technical, tokenomics, market structure, ecosystem position, regulatory posture, team, risk, narrative, supply-chain transmission. Roughly sixty tables. Ruled margins, styled headers, a closing disclaimer in the standard cadence.
Information points: zero.
Not thin. Not provisional. Zero. Every cell read N/A. Every conclusion read "unable to evaluate, insufficient input." Every risk flag read "undetermined." Someone had spent real hours on the scaffolding, and the document contained no claim about the world capable of being true or false.
It was the most honest thing I read this quarter. And the fact that it needed to exist โ that a nine-dimension framework was invoked, populated, formatted, and delivered with an empty payload โ is a cleaner read on this market than anything else in my inbox.
Skepticism isn't a mood. It's an input cost. Most of the industry is currently paying it in installments and calling the receipt research.
How Crypto Built a Framework Economy
The template didn't come from nowhere. It came from venture diligence.
Between 2019 and 2021 crypto research bifurcated. One branch was fund-side โ memos written to justify a position to an investment committee, structured around market size, wedge, moat, token design, unlock schedule. The other was publication-side โ long-form posts written to build an audience, structured around narrative and timing. Both branches converged on the same artifact. The checklist.
By DeFi Summer in 2020 I was running the same arithmetic most analysts were: Aave and Uniswap composability, TVL expansion, yield-farming loops. The data was live and readable on-chain. What I noticed then, and what has compounded since, is that the format of analysis hardened faster than its substance. TVL became a headline rather than a decomposition. "Composability" became a word you typed, not a graph you drew.
Three accelerants followed.
Standardization. Funds wanted comparable deals, and comparability requires identical fields. Identical fields produce identical documents. A 2026 research note and a 2021 research note share roughly eighty percent of their structure and maybe twenty percent of their epistemics.
Regulatory pressure. After 2022 and 2023 the compliance function, not the conviction function, set the shape of the deliverable. Once your note has to survive legal review, you get disclaimers, dimension coverage, and hedged conclusions. You get structure that protects the author rather than informs the reader. The SEC's regulation-by-enforcement didn't merely chill token issuance. It reshaped what is sayable inside a research document, which reshaped what gets written.
Generative tooling. By 2025 an analyst could stand up a nine-dimension framework in an afternoon. By 2026 an agent could do it in ninety seconds, populate every field with structurally plausible text, and format the output to house style. The marginal cost of a framework went to approximately zero. The marginal cost of an information point did not.
When the cost of the container collapses and the cost of the contents holds, you get exactly what economics predicts: containers everywhere, contents scarce. The market filled with documents that look like analysis the way a stock photo looks like a family.
Liquidity doesn't read the container. But the people who allocate liquidity do. That's the problem.
Anatomy of an Empty Report
Dimension count is a proxy for rigor, and it always was. Nine dimensions is not an accident. Neither is seven, nor twelve. These numbers descend from the diligence checklist of growth-stage funds, which descends from a private-equity template, which descends from an audit standard. None of them descend from crypto. Nobody has ever established that nine questions predict crypto outcomes. The number propagated because it feels exhaustive, and because a committee cannot easily reject a memo that covered everything.
A framework's job is not to find the truth. A framework's job is to make the author unfireable. Nine dimensions means nine places to point when the position goes to zero. We covered the regulatory dimension โ it was flagged as undetermined. Undetermined is not a finding. It's an indemnity.
Watch how each dimension behaves under an empty payload.
The technical section asks for innovation, maturity, security assumptions, performance. All four read N/A. The presence of the category does work โ it signals the author thought about code. What it never does is tell you the sequencer runs in a single AWS region in Frankfurt. That's a fact, and facts require someone to read the deployment config.
The tokenomics section asks for team allocation, investor allocation, community allocation, unlock schedule. Empty version: four rows of N/A. Real version: cliff at month twelve, linear over thirty-six, eighteen percent to a foundation behind a multisig you can't inspect. Only one of those versions is useful. The useful one costs an hour on Etherscan and a willingness to say something the project's team will resent.
The market section asks for cycle position, price impact, sentiment, funding. Empty reads N/A across the board. Which is bizarre, because market data is the cheapest data in this industry. Funding rates are public. Open interest is public. Order-book depth is public. An empty market section doesn't mean the data was unavailable. It means nobody pulled it, or the pull wasn't wired into the pipeline.
That's the tell. In an empty framework, the sections that go dark are diagnostic. Technical goes dark because it's expensive. Tokenomics goes dark because it's adversarial. Market goes dark because the pipeline is broken.
The regulatory section asks for a Howey analysis: money invested, common enterprise, expectation of profit, efforts of others. Four elements, all readable as N/A. Here is what that misses. The live regulatory question in 2026 is not whether a token is a security. It's who can hold it, where it can be custodied, which broker-dealer will touch it, and what the transfer agent does at the margin. Those are operational questions with operational answers. A Howey table โ populated or empty โ answers none of them.
The team section asks for technical capability, industry experience, stability. The useful version involves noticing that three of five core contributors' GitHub keys stopped signing in the same week. That is the highest-signal observable event in crypto, and it is free to check.
The risk section is where the empty framework becomes almost elegant. Six categories โ technical, market, operational, regulatory, competitive, narrative โ every cell reading N/A, composite rating reading "unable to assign." That is a risk matrix that transferred all risk back to the reader while appearing to absorb it. It is, functionally, a disclaimer with columns.
The narrative section asks for the story, its heat cycle, the expectation gap. Narrative is the one input that requires no permissions, no data vendor, no archive node. You observe narrative by reading. An empty narrative section means the analyst didn't read.
And supply-chain transmission โ upstream, midstream, downstream โ most clearly reveals the template's origin. That section exists because hardware supply chains matter in semiconductors and because it sounds sophisticated when applied to crypto. What it usually produces is a diagram with three boxes and arrows naming mining rigs, exchanges, and "DeFi" in a fixed order regardless of what the project actually does.
I have written all of these sections. Between 2017 and 2018 I audited more than fifty whitepapers for a boutique advisory shop in Vancouver while concurrently running three small-cap utility token launches across Southeast Asia. What took me years to admit is that the audit was frequently a longer document than the whitepaper and just as unfalsifiable. I produced eighty percent of those audits from template. The outputs that actually killed deals were usually one paragraph long โ a token with no sink, a treasury key held by one person, a vesting schedule that front-ran its own liquidity.
Eighty percent of those projects lacked a viable liquidity model. Not a flawed one. No model at all. Speculative FOMO standing in for market structure. And the deliverable documenting this was forty pages of structure wrapped around a one-sentence finding.
So the uncomfortable question: if a one-paragraph finding does the work and a forty-page framework does not, why does the industry keep buying forty pages?
Because the buyer isn't buying information. The buyer is buying coverage.
The Cost Structure of Producing Nothing
An information point costs, at minimum, a data pull, an archive query, a contract read, a founder call containing a question the founder does not want to answer, and a willingness to write a sentence that can be proven wrong. Call it two to eight hours per genuine point โ and note that the expensive part isn't labor. It's reputational exposure. A wrong claim is a liability. An N/A is not.
A framework dimension costs a heading, four sub-headings, and a table. In 2021, call it fifteen minutes, most of it formatting. In 2026, with an agent drafting to house style, call it four minutes and a review pass.
The incentive gradient is enormous and it points somewhere specific. Produce dimensions. Consume information. Net exposure zero. Net appearance: comprehensive.
And the market pays for appearance, because the measurement apparatus is broken. How does a crypto research product get evaluated today? Engagement. Citations. Followers. Reposts. The number of times a fund partner forwards it to a group chat. Not one of those metrics correlates with predictive accuracy, and several correlate negatively. The most-framed, most-comprehensive, most-hedged documents travel furthest, because they are the safest objects to forward. Forwarding a hedged document costs you nothing. Forwarding a call costs you something.
A note reading "unable to evaluate, insufficient input" earns zero engagement and raises zero follow-up. A note reading "structural bid, target up forty percent" gets screenshotted. The second has a fifty percent chance of being wrong. The first has a zero percent chance of being anything. The industry rewards the second, consistently, which is why the first is rare.
The deeper problem: measuring research by engagement imports the attention economy's core pathology into capital allocation. Attention is zero-sum and reflexive. Information is not. A market that allocates attention efficiently can still allocate capital terribly, because attention and capital do not share an objective function.
Liquidity doesn't allocate to information. Liquidity allocates to attention. That sentence explains more of the last three years than any tokenomics model I have built.
What Institutions Actually Consume
I spent most of 2024 modeling spot Bitcoin ETF flows against traditional equity fund flows. The conclusion I published then was that institutional capital was acting as a volatility dampener rather than a speculation driver, and that Bitcoin's price action was decoupling from the altcoin cycle as a result.
The second-order lesson, which I didn't publish, concerns research demand. Institutions do not consume frameworks. They consume position sizing, borrow costs, custody mechanics, settlement windows, and liquidity depth at the size they need to trade. A pension consultant does not care about your nine dimensions. A pension consultant cares whether a fifty-million-dollar clip moves the tape, whether the prime broker takes the other side, and whether the auditor signs the NAV.
That's a different document. Shorter than a framework. Harder to write than a framework, because every claim is checkable by someone with money. You cannot write N/A into a liquidity-depth table and expect a trading desk to accept it. The desk will pull the book and find out.
There's the structural gap. Crypto-native research is optimized for an audience that cannot verify. Institutional research demand is generated by an audience that can and will. The empty framework clears precisely because its consumers lack verification capability. The moment verification capability arrives โ a desk, an auditor, a risk officer โ the framework stops settling.
Which makes the empty framework a bearish signal about the retail-facing information market and a bullish signal about the institutional-facing one. The same twenty pages that work on Crypto Twitter are dead on arrival in a diligence committee. Something has to give. Given the flow direction since 2024, the thing that gives is the retail-facing format.
The Machine-Readable Squeeze
In 2026 I designed a simulation in which autonomous agents held blockchain wallets and executed micro-transactions, then modeled what that does to liquidity velocity. The uncontroversial finding: agent-to-agent settlement raises transaction frequency by an order of magnitude, stressing fee markets and changing the optimal block-space auction.
The finding that got me into arguments was about research consumption. Agents don't read. Agents parse.
This matters more than it sounds. Human readers apply charity. When a human opens a document with nine dimensions and zero filled fields, they interpolate โ the analyst ran out of time, the data was messy, a follow-up is coming. Humans read N/A as a soft state.
An agent reads N/A as null. Null propagates. A pipeline ingesting ten thousand research documents and hitting a null regulatory field doesn't infer undetermined regulatory posture. It computes a missing value and either drops the record or imputes from the mean, which is worse. Machine-readable research punishes rhetorical padding and rewards determinism. You cannot hedge to a parser.
Call it the determinism squeeze. For thirty years research writing has been optimized for human forgiveness โ the ability to say something without committing to it. That optimization is about to be repriced. When agents become primary consumers of market commentary, a document's value becomes a function of how much of it is machine-actionable. A filled field with a defendable number outperforms nine empty dimensions, because only one of them survives contact with a parsing layer.
This is the real disintermediation, and it isn't the one anyone is watching. They're watching agents trade. They should be watching agents read.
The Empty Framework Is Correct
Here is where I break with my own camp.
The standard skeptic's take on an empty report is that it represents failed diligence and that the fix is more data, more dimensions, tighter templates. I think that's backwards, and I think it repeats the framework's original error.
The empty report's actual failure is that it was produced at all. A framework returning nine N/As has said something true: the analysis was not executable. The correct output in that state is one page and one sentence โ "no information point available, no position can be evaluated." Everything past that sentence is structure doing the job of substance, and structure doing the job of substance is indistinguishable from marketing.
I've started treating an honest N/A as a higher-quality signal than a confident three-star rating. A three-star rating on a matrix is a number someone chose. It drifts between analysts, drifts between Tuesdays, and anchors the reader to false precision. An N/A is a fact about the state of the world. It's falsifiable, it's honest, and it is nearly free to produce โ which is precisely why it's rare.
The blind spot in the skeptic camp is assuming more dimensions produce more insight. Past a threshold โ I'd put it around four โ added dimensions produce more places to hide. Every dimension you add is a new field that can be marked undetermined, and undetermined fields are how a document that says nothing passes for a document that says everything.
Skepticism isn't the exhaustive checklist. Skepticism is the willingness to write the one sentence that kills the deal, and then sign it.
Takeaway
The positioning question for the rest of this cycle: you are long attention and short information. When does that trade reverse?
It reverses when verification gets cheap. Watch for desks that pull their own books, agents that parse instead of read, and auditors who will not accept a null. Watch the fill rate โ not the number of research documents, but the share of their fields holding a falsifiable number. That ratio is the cleanest available measure of how much of this bull market is priced on structure versus how much is priced on knowledge.
My guess is the ratio is falling. And when it turns, it will turn fast โ because it always does, at the exact moment attention and liquidity stop agreeing with each other.