I spent an hour this week reading a crypto research report. It had nine sections. Fourteen tables. More than sixty data fields. Every substantive value in it read "N/A."
Not estimated. Not redacted. Not pending. "N/A — insufficient information." The technical assessment returned nothing. The token economics returned nothing. The regulatory analysis, the team evaluation, the risk matrix, the narrative forecast — nothing. A document with the full architecture of an intelligence product and the information content of a blank page.
I have audited smart contracts for nineteen years. I have read a lot of bad analysis. This was the first time I read a document engineered — deliberately, structurally — to look like analysis while refusing to contain any. That is not a bug. That is a feature. And the feature is spreading across the bear market like rot in a load-bearing wall.
The format is familiar because it is standardized. Nine modules: technical, tokenomics, market, ecosystem position, regulatory, team and governance, risk, narrative, supply-chain transmission. It is a chassis. Anyone can bolt a project name onto it. The output looks like diligence. The output reads like diligence. The output is not diligence, because diligence is the residue of actually looking at something.
Where does the chassis come from? Partly from sell-side research desks, which need repeatable formats to cover dozens of assets with three analysts. Partly from data aggregators, which turned "coverage" into a product and "coverage" into a checkbox. Partly — increasingly — from language models, which were trained on the format and reproduce its shape without its content. The template is the single most reproduced artifact in crypto research. It is more reproduced than any whitepaper, any audit, any post-mortem. And most of it is empty.
In a bull market, nobody reads the N/A. Prices go up, and the report is decoration. In a bear market, readers actually open the file, and the emptiness becomes visible. This is the context we are in now. Survival has replaced speculation as the operative question. "Is my asset safe?" is the only question left. And the template — the thing that was supposed to answer that question — returns "N/A."
That returns us to first principles. What is a research report for? It is not a wrapper. It is a measurement. If a measurement produces no number, it is not a measurement. It is a form. Forms do not fail loudly. They fail by silence.
The schema is worth decompiling. Reversing the stack to find the original intent. Fourteen risk flags. A Howey-test table. A supply schedule. A transmission graph. Each of these is a stand-in for a specific piece of evidence. The template does not ask "what is true?" It asks "what category of thing should be here?" And a category can always be filled with "N/A" without contradicting anything. That is the structural flaw. The format has no error correction for absence.
Consider the technical section. A real technical review reads the bytecode. It verifies the compiler version, checks for the exact diff between deployments, traces admin keys to their multisig signers. In 2017 I did this on 0x's v0.9.9 exchange contract. I found three unsigned-integer overflow paths in fillOrder. That took six weeks of reading assembly against the test suite. The template does not read assembly. It cannot. It has a field labeled "security assumptions," and I can write "N/A" in that field faster than the contract can execute a single CALL. The template marks the security box. The box is checked by the act of writing in it, not by the act of verifying anything.
Consider tokenomics. A real supply structure reconciles the genesis allocation against current on-chain state. You sum the team vesting contract, the treasury multisig, the foundation wallet. You check whether the unlock schedule in the docs matches the cliff in the code. Most of the time it does not. I have seen a "six-month cliff" that vested daily from block one. The template has a row for team unlock. The template cannot see the block. Truth is not consensus; truth is verifiable code. The template does not verify code. It verifies that a row exists.
The regulatory section is the most revealing. It carries a Howey table — investment of money, common enterprise, expectation of profit, from the efforts of others. Four fields. Four "N/As." But the Howey test is not a table. It is an argument. A court asks whether the network is sufficiently decentralized that no promoter's efforts drive the return. You answer that by reading the governance contract, the upgrade authority, the treasury spend history. I wrote about this after the NFT metadata crisis — I traced forty percent of "decentralized" collections to a handful of centralized IPFS pinning services. The legal conclusion followed from the infrastructure. The template removed the infrastructure and kept the conclusion. Now the conclusion is empty, because the evidence that produced it was never collected.
Note the abstraction leak. The format presents categories as if collecting them produced analysis. It does not. The categories are a map. The map is not the territory. Abstraction layers hide complexity, but not error — and this one hides complexity while importing error, because a filled category feels like a verified one. A row labeled "audited" reads as safe. A row labeled "N/A" reads as pending. Neither row has touched a byte.
The market section is where the template lies loudest, because market data is the easiest data to actually collect and the template still refuses to. A real market read pulls the order book, the funding rate, the open interest, the DEX/CEX spread. In the current regime, funding is the signal: when perps trade at a persistent discount and open interest declines while spot holds, you are watching leverage exit, not conviction. That is a measurable thing. The template has a row for "market sentiment" and a row for "price impact," and both are "N/A." It had access to public APIs. It had access to every block explorer. It returned nothing.
The ecosystem-position module is subtler. It wants a transmission graph — upstream infrastructure, midstream protocol, downstream application. In a healthy report, that graph is annotated with dependencies you can verify. Which oracles feed the protocol's liquidation engine. Which bridge secures its collateral. Which sequencer orders its blocks. The graph in this report is three boxes labeled "N/A," connected by arrows. The arrows are structural. They simulate the shape of a supply chain without naming a single counterparty. A reader cannot act on it. A reader cannot even argue with it. You cannot falsify a blank.
The risk matrix is the most dangerous section, because it is the one a reader most wants to trust. Six risk classes. Five columns — level, probability, impact, mitigation. Every cell empty. But the presence of the matrix implies a process. It implies someone enumerated threats. It implies someone assigned likelihoods. None of that happened. The matrix is a promise the document does not keep. In crypto, an unstated risk is not a neutral fact. It is a liability with a due date.
I keep coming back to my own work, because it is the only honest benchmark I have. In 2020 I spent three months simulating slippage vectors on Ethereum mainnet for Curve's stable pools. That produced a fifteen-thousand-word paper with reproducible scripts. The scripts are the report. If you disagreed with my conclusion, you ran the code. In 2021 I traced metadata handling across the ERC-721 standard and found the centralized backend under the decentralized label. That was a measurement, not a category. In 2022 I reverse-engineered the LUNA/UST loop until I found the exact block where the peg-breaking feedback became mathematically irreversible. That was a deterministic failure map. You could point to the state transition. The template above points at nothing.
What would it take to fill this document? Less than you think. Each "N/A" has a definable input. The technical section wants the verified bytecode hash and the admin-key graph. The tokenomics section wants the vesting contract addresses. The regulatory section wants the upgrade-authority address and the last twelve months of treasury spends. The market section wants a single afternoon of API calls. The ecosystem section wants the oracle and bridge addresses from the protocol config. None of this requires privileged access. All of it requires that someone actually look. The template was cheaper than looking. That is the whole story.
By 2026 I moved into the intersection of AI agents and on-chain execution, testing verifiable-compute protocols where a model proves its inference in zero knowledge. I found a gas-optimization bug in the proof-verification logic that cut transaction cost by forty percent. That finding existed because I read the circuit. An agent-driven report could have generated the same document above in four seconds — and it did, in effect, because the document is a form, and a form is what a model produces when it has learned format but not evidence. This is the failure mode of the entire generation: fluency without measurement. The report is plausible. Plausibility is not verifiability.
Here is the contrarian part, and I mean it fully. The empty report is more honest than the filled one. A report that writes "N/A" in the regulatory row is telling you the truth: no one checked. A report that writes "compliant" in the regulatory row after a fifteen-minute skim is lying with more confidence. I have read hundreds of filled templates. They are worse. They replace silence with false signal, and false signal is what gets capital destroyed. The empty document is a confession. The filled document is a forgery. Given the choice between a confession and a forgery, the confession is the better input, because at minimum it does not mislead.
But do not mistake honesty for usefulness. The empty template is honest and useless at once. It is the analyst's fear, compressed into a schema and shipped as a deliverable. It says: I did not have time, or access, or willingness to take a position that could be wrong. Refusing to be wrong is easy when you refuse to say anything. That is not rigor. That is paralysis wearing rigor's clothes. The format lets the author hide, and the reader, starving for signal in a bear market, mistakes absence for caution.
The fix is not a better template. The fix is fewer, deeper files. One protocol, one measurement, one falsifiable claim. Replace the nine-section chassis with a single question: what breaks this, and at what number? Then answer it with data someone else can reproduce. If a report cannot be wrong, it cannot be right, and the empty schema is the purest proof of that principle.
So the next time a research file lands in front of you with fourteen tables and sixty fields, open the technical section first and read the bytecode citation. If it cites a hash, keep reading. If it cites "N/A," close the file and go to the block explorer yourself. The answer to "is my asset safe?" is never in the template. It is on-chain, in the diff, in the admin keys, in the vesting contract that does not match the docs. Reversing the stack to find the original intent is the only analysis that survives a bear market. Everything else is a form with a footer, waiting for a reader who never checks.
