There is a document circulating in my group chats this week that contains zero findings, zero price targets, zero token allocations, and zero project names โ and it is, by a comfortable margin, the most technically honest piece of due diligence I have read all quarter.
Nine sections. Nine headers. Every field carrying the same two characters: N/A.
Under technology: no TPS, no trust model, no audit status, no repository. Under tokenomics: no supply cap, no unlock cliff, no emission curve. Under team: nothing. Under the risk matrix: six category rows, six empty columns, and one sentence beneath them โ "Cannot be rated. Any rating without a subject would be fabrication."
The analyst refused to write the report. Then did something rarer. They wrote down, dimension by dimension, exactly which inputs would have been required to produce it. Not a rant. A specification. A build list for a memo that never got built, plus three named recovery paths for the missing data.
I have written more than 400 due diligence memos in twelve years. I have never written one that good. Most of mine have numbers in them. That is not the same thing.
Context
Crypto research is not a research business. It is a content business wearing a lab coat.
The pipeline is standardized now. Stage one parses: extract claims, identify projects, note the author's stance, timestamp the signal. Stage two analyzes: technology, token, market, ecosystem, regulation, team, risk, narrative, transmission. Nine dimensions, because nine dimensions sound like a framework and five sound like a blog post.
The pipeline has a structural flaw nobody writes about, because writing about it costs you the pipeline. Stage two cannot produce output from stage one's blank input โ but stage two still has a deadline, a client, and a reputation that reads "publishes weekly."
So the analyst does the thing the industry trained them to do. They fill the table. Market cap is public, so the market section fills itself. Team can be inferred from LinkedIn. Risk matrices can be generalized from category templates. Within an hour, an empty input becomes a confident 1,800-word note with a "moderate risk" rating and a "monitor for developments" conclusion. Nobody is lying, exactly. Everybody is publishing.
I did the same thing in 2017, in reverse. Sophomore year at NYU, ETHDenver, $3,000 of summer savings into three ICOs promising revolutionary AI tokens. I skipped the part where I read the GitHub. I read the landing page instead, because the landing page had a countdown timer and the repository had three commits and a README.
When Ethereum Classic hard-forked and volatility tore through my positions, I sold at the bottom. The fork wasn't the failure. The fork wasn't even the event. The fork wasn't anything except the moment the gap between what I had read and what existed became expensive. I filled every blank in my own head with the project's own marketing, and paid retail for the privilege.
That is the trade this analyst declined.
The Core
Let me dissect the refusal, because the refusal is doing more work than it looks like.
The nine-dimension scaffold is modular. That is its virtue and its trap. A blank input doesn't block the framework โ it invites it, because each dimension can be generated from public priors. Price is on CoinGecko. Team is on LinkedIn. Regulation is a jurisdiction lookup. Narrative is a Twitter search. An analyst who has run the framework forty times can produce plausible output on a subject they have never studied, and the output will read exactly like the output on a subject they have studied for a year.
That is the actual mechanism behind most of the research you read. It is not fraud. It is autocomplete.
Here is the cost structure, which almost nobody tabulates:
| Dimension | Public data density | Fabrication cost | Who pays | |---|---|---|---| | Market | High | Low, immediate | The reader's entry price | | Technology | Low | High | Developers who inherit the doc's assumptions | | Tokenomics | Medium | Very high | LPs and lock-up buyers | | Team | Medium | High | Retail, at exit liquidity | | Risk | None | Extreme | Everyone, eventually |
Read that table from the bottom up and you have the last four years.
In 2020, during DeFi Summer, I was in a University of Pennsylvania student group manually tracing Yearn Finance vault strategies โ $50,000 in simulated yield across three protocols, spreadsheet open, slippage recalculated by hand. The gurus in the Discord were publishing strategy threads hourly. I found a slippage accounting discrepancy in one protocol's vault math. When I raised it, I was called a noob by people who had never opened the contract. Yield is a sedative; volatility is the needle. The discrepancy was real. One protocol reaped its users, quietly, on a number that never made a headline.
That is the tokenomics row. The fabrication cost there is very high and the payment date is deferred. Deferred payments are how you build a career as a bull.
In 2021, at NFT NYC, I sat with Axie Infinity players who had lost their savings to a phishing site mimicking the official launcher. Not a protocol bug. A signature spoofing attack behind a pixel-perfect clone. I traced the contract interaction logs myself and wrote a blunt thread naming the operational negligence while stopping short of calling it a hack, because it wasn't one. The thread cost me nothing. The players had already paid.
That is the team row and the risk row, and they are the same row.
By 2022, when Terra collapsed, I stopped pretending the framework was the point. I hosted a weekly mixer in Manhattan โ developers, traders, whoever needed somewhere to be angry โ and let the anecdotes do the analytical work. Cold hands dissect the heat of a hype cycle, but cold hands had not been dissecting anything. The heat was the data. Every person who told me they had done their research had done exactly what I did in Denver in 2017: converted a blank table into a confident position, then quoted the position back to themselves as evidence.
And in 2025 I watched the same mechanism run at machine speed. An AI trading agent advertising 500% APY. A social audit with five developers. The decision logs โ the entire value proposition โ generated off-chain by a shell script, dressed in an interface designed to look like inference. It was obvious within forty minutes. It was obvious to us because we had seen the blank before and knew what fills it. Assets don't read your thread. They don't get the memo, the retraction, or the "we have updated our view" footnote four months later. The project shut down before mass adoption, which is the only kind of win that counts in this work, and it counted for nothing because nobody outside the five of us noticed.
Here is the part of the refusal memo that deserves more attention than the refusal itself.
The report's nine empty dimensions are not neutral. They are an argument. A framework that can be filled from priors is a framework that manufactures priors. The analyst who wrote N/A in every field was not being humble. They were demonstrating, with a seventeen-row status table and a recovery protocol, that the scaffold can distinguish "unknown" from "unfalsifiable" โ and that the industry's entire output has been collapsing the two for four years without a single client complaint.
That is the finding. The finding is that there was no finding, and that a finding was expected anyway. Every N/A is a shadow.
The Contrarian Angle
The bulls of always-publish have a stronger case than my side likes to admit, and I want to state it properly before dismissing it.
Silence is a position. It is not a neutral one. If every rigorous analyst refuses to publish on thin inputs, the field is not elevated โ it is vacated. The influencer with 400,000 followers and no framework fills the gap by Thursday. The refusal memo's rigor is invisible to the retail reader who needed a call, and the confident wrong answer is visible, shareable, and actionable. A null result is socially useless, which is the same as commercially useless, which is the same as losing the argument by default.
There is a second, sharper point aimed at my own side: the refusal memo is itself a product. A very good one. Its structure โ nine dimensions, a status table, a three-path recovery plan, a one-line integrity claim โ performs the exact function of a credit rating. It signals rigor without transacting in evidence. The scaffold does the credibility work the data was supposed to do, and it does it while saying nothing at all.
I have watched that move before. I have made that move before. Every analyst who has ever written "further research required" as an entire conclusion has made it. The refusal is honest about the input and quietly flattering about the method, and we keep bundling those two things together because the bundling is what sells.
The uncomfortable synthesis is not that the refusal was wrong or that the publishers are right. It is that the market cannot currently price the difference. We audit the code, but we mourn the users. We have no instrument for measuring the value of a report that declined to exist. Which means the honest analyst is structurally underwater, permanently, and the only people who survive are the ones with something else paying their rent.
The Takeaway
Watch the refusals. Not the memos โ the people. Over the next two quarters, the signal will not be which protocols publish the loudest technical roadmap. It will be which analysts go quiet on which subject, when, and whether they say why.
A blank table is the cleanest thing in this industry. The question worth carrying into the next cycle is not what a report says. It is who was willing to submit one with nine columns of N/A and their own name at the top โ and who, handed the same empty input, chose to fill it in.