The report was eleven pages long. Inter typeface. Generous margins. Nine sections, four comparison tables, a Howey test matrix, and a risk heat map rendered in five shades of amber.
It had been commissioned by a fund that wired eight figures into the project two weeks earlier.
Six of the nine sections read, verbatim: "N/A — insufficient information." The other three cited the project's own documentation back at itself, reformatted as findings.
I have been reading crypto research for twenty-five years. I have never watched a document that expensive say that little.
So I did what I did in 2017, three weeks before an ICO launch, when I found an integer overflow sitting inside a transfer function that nobody had bothered to open: I stopped reading the prose and started reading the scaffolding. The framework was the product. The analysis was decoration.
We audited the silence between the lines of code. That silence now has an invoice attached.
The Industrialization of Looking
Between 2017 and 2021, crypto research was a person, a block explorer, and an opinion. By 2023 it had become a pipeline.
Stage one extracts "information points" from a source — funding round, headcount, token supply, audit firm, chain, backers. Stage two pushes that list through nine standardized dimensions: technical, tokenomics, market, ecosystem position, regulatory, team and governance, risk, narrative, value-chain transmission. Each dimension owns sub-tables. Each sub-table owns fields.
Each field owns a default.
And the default is N/A.
This is not a scandal. It is an architecture, and it was built on purpose. The nine-dimension framework was never designed to produce knowledge. It was designed to be completable. A fund needs an artifact to attach to a position — an internal PDF that says the diligence happened. A listing committee needs something to reference. A distribution partner needs a link in a Telegram. The shape of rigor is fungible; the substance is not.
So we got shape. A lot of it.
The acceleration happened after November 2022. The FTX collapse vaporized the assumption that a smart person's word was collateral, and the entire industry responded by demanding paperwork. Institutional entry in 2024 and 2025 poured gas on it — suddenly every conversation needed a document, because allocators answer to compliance, and compliance answers to paper. That part is rational. Where it went wrong is that the industry standardized the form of diligence long before it standardized the verification of it.
Here is the honest nuance, and I will defend it: templates are not inherently fraud. When the primary source is itself a primary document — an SEC rulemaking, a MiCA technical standard, an ETF prospectus — the template is genuinely useful. I built my own regulatory synthesis workflow on exactly that logic in early 2025, because the input was dense, dated, and authoritative, and the value I added was speed of translation. A good template on a primary document is leverage. A good template on a narrative is a forgery with better typography.
And most crypto research is run on narratives.
The Fill Rate
I spent four weeks opening research packets from the previous fourteen months — the ones that circulate in group chats, the ones with the watermarks. Different logos, same skeleton.
Here is the metric nobody publishes: the fill rate. What percentage of a report's structured cells contain a number, a hash, a dated on-chain action, or a named and verifiable counterparty — as opposed to an adjective, a paraphrase of the project's own blog, or N/A.
I scored 31 documents. Median fill rate: 22%.
Eight of them scored below 12%. Those eight were the longest. The correlation is inverse and it is brutal: the emptier the analysis, the more elaborate the container. Page count is camouflage. A risk heat map with five amber shades looks like work the way a stage set looks like a house.
Let me get concrete about the texture, because this is where the smell lives.
The Howey tables. Every packet has one. Four rows — investment of money, common enterprise, expectation of profit, efforts of others — plus a verdict column. In most of the packets I read, three of the four cells held a hedge and the verdict cell read "nuanced." That is not legal analysis. That is a posture toward legal risk, wearing a table as a costume.
The risk matrices. Six categories, five severity gradations, a probability column. I found one asset rated "high" on technical risk by three houses and "low" by two — in the same quarter, off the same publicly available data. The standard deviation of severity assignment across analysts is larger than the standard deviation across assets. The matrix is not measuring the token. It is measuring the analyst's position in it.
The team sections. Fully populated, almost without exception. A LinkedIn scrape, a conference photo, a "previously at" line. This is the one dimension where information is cheap and abundant, so it is the one dimension that always scores well — and it is also the dimension with the least predictive power over whether the thing is still running in eighteen months. The sections that fill themselves are the sections that tell you nothing.
The competitive landscape. Two real cells from the project's own deck, six cells of N/A. I once compared three reports covering the same sector and found the market-share column populated with three mutually contradictory numbers, none of them sourced, all of them footnote-free.
The narrative and heat-cycle row. This one is my favorite. It gets filled with vibes — "modularity," "AI agents," "RWA" — and the sustainability assessment is a sentence, and the sentence is almost always optimistic, because a fund that paid for a report is not paying for a bear case. Nobody commissions a funeral.
The value-chain transmission diagram. Arrows pointing from N/A to N/A.
Then there is the pipeline failure itself, which is the part that should terrify anyone who trusts these documents. When the upstream extraction returns empty — bad OCR, a paywalled doc, an article that was itself a summary of a summary — the downstream stage does not halt. It cannot halt. It has a schema to satisfy. It emits N/A field by field, section by section, and the output still renders as a finished product, because a report with nine sections and four tables looks finished regardless of what is inside it.
I ran that experiment on a chain of three published articles that all described the same protocol upgrade. By the time I opened the files, the third one described a mechanism the first one had never claimed, a mechanism the protocol team later told me did not exist. Every dimension had been filled. Confidently. The pipeline had no fact to work from — only a shape to complete.
I have seen the shape of this before, from the other side. That ICO had an audit report. Twelve elegant pages. It said the contract had been reviewed. What it did not say — what no formatting choice could reveal — was that the reviewer had never opened transfer, the only function that mattered. I found the overflow in a weekend. The report had been circulating for a month.
Same energy. Different decade. The instrument of reassurance was the instrument of blindness.
Which brings me to where I stop being angry at the N/A cells and start being angry at something else.
The Honest Report Problem
Counter-intuitive, and I will take the downvote: the N/A reports are not the worst reports. They are the least dishonest ones.
A document that says "insufficient information" in six of nine sections has told you something true and actionable. It told you the team section is thin because there is no public team. It told you the tokenomics table is empty because the vesting schedule has never been published. It told you the risk matrix is blank because nothing has been deployed yet. That is a finding. A bearish one, delivered in the driest font available, and almost everybody scrolls past it because it does not look like work.
The dangerous report is the fully-populated one. Every cell filled. Team graded A-minus. Tokenomics "well-structured." Risk "manageable." Those cells are populated because the analyst had a narrative to protect, or a relationship, or an allocation. Manufactured completeness reads as competence. N/A reads as laziness. We have the incentives exactly inverted.
I watched this play out at the funding layer too. Grant committees scoring submissions on completeness rather than on whether anything in them was ever verified — which is precisely why the RetroPGF model remains the only public-goods mechanism I have seen that structurally resists it. The reviewers there spend their own allocation votes on code they can read. Nine sections of N/A earns nothing from a badge-holder. They open the diff. That is the entire difference between a report and a review: whether the reader is capable of checking the claim.
Templates scale because templates are never checked.
I will make one more prediction, because the pattern is already visible. The next generation of marketing is not a whitepaper — it is a research report written about itself, seed-funded by the treasury, formatted to look independent, and quoted by the ecosystem's own ambassadors. We are watching this happen right now with the chain-deployment wars. The decks compare OP Stack against ZK Stack on technical merits, dimension by dimension, with N/A in the cells that would require a shipping date. But the actual contest is not technical. It is which team can convince more projects to deploy first. That is a distribution question. It will never appear in the technical dimension of the template, because the template has no field for it.
And Uniswap V4 gets the same treatment. Every packet I have read calls hooks "programmable Lego." Not one of them scored the cell that matters: how many teams can actually ship an audited hook, and who audits a hook that the pool owner can swap out at will. Complexity has a fee schedule. The template does not have a row for the fee.
What To Watch
Next time you open a research packet, run one pass. Search the document for four strings: a contract address, a block height, a bytecode hash, a dated on-chain transaction.
Find none of them — and you are not holding analysis. You are holding a mood board with a Howey table stapled to it.
The frameworks will keep getting more elaborate, not less. We are in a bull phase. Check sizes are up, LP interest is up, and demand for the feeling of diligence has never been higher. Every incentive points toward bigger templates and thinner fill rates.
The question is who blinks first: the allocators who stop paying for shape, or the analysts who keep selling it.
The code was always the report. Everything else is margin.
