At 09:14 CET on a Tuesday in February 2026, a 4,700-word research document crossed my desk. It arrived with the full apparatus of institutional rigor: nine analytical sections, forty-three structured tables, a six-category risk matrix, a completed Howey test, a token distribution breakdown, and a transmission map tracing impact from upstream infrastructure to downstream applications. I read it twice. Then I counted the substantive claims. The document contained eleven instances of the string "N/A — insufficient information," four annotations reading "confidence: not applicable," and zero verifiable on-chain data points. Every field that should have held a number held a placeholder. The architecture was immaculate. The building was empty.
I have spent sixteen years reading crypto research, most of that time with the specific job of deciding which claims were true. I have never seen a report so precisely structured and so completely hollow. It was not a failure of analysis. It was a failure of inputs — a nine-dimension machine running on zero fuel and still spinning out pages because the schema told it to.
The industrialization of crypto research began around 2019 and accelerated hard through the 2021 cycle. Before that, a protocol deep-dive was an essay. A founder wrote a blog post, a trader wrote a thread, someone with a Bloomberg terminal wrote a PDF. The format varied because the thinking varied. That changed the moment funds, exchanges, and data platforms recognized that "research" was itself a product that could be scaled, and that the scalable unit of research is the template.
You define the dimensions — technology, token economics, market structure, ecosystem position, regulatory posture, team and governance, risk, narrative, and supply-chain transmission. You build a table for each. You hire junior analysts, or, later, you point a language model at the schema, and you fill the cells. When the template is good, it is genuinely useful. A disciplined checklist catches the things a narrative misses entirely: the unlock cliff three quarters out, the single custodian standing behind a wrapped asset, the governance vote that quietly transferred upgrade authority to a three-of-five multisig nobody disclosed. My own career started inside exactly this machinery. In 2017, as a junior analyst at a Paris venture firm, I built a rigid checklist-based due diligence framework and used it to flag three token sales before they launched. Two of them failed. The checklist worked because every single box had an answer.
The failure mode appears when the template outlives its inputs. The nine dimensions become a fixed cost. The schema demands a table for "team and governance" whether or not a team is identifiable. It demands a token distribution chart whether or not a token exists. It demands a Howey analysis whether or not there is a security to analyze. And so the analyst — human or machine — does the only thing the schema allows. It fills the box with the shape of an answer: "N/A," "insufficient information," "not applicable." The report grows. The signal does not.
Here is the mechanic, stated plainly: a fixed analytical schema converts missing data into structural noise. The nine-dimension framework I received was not wrong. Each dimension is a legitimate lens. The problem is that the framework cannot distinguish between "I analyzed this thoroughly and found nothing" and "there was nothing here to analyze." Both render as the same empty cell. To a reader skimming the table, an empty box looks like a checked box. Coverage is silently mistaken for diligence, and the reader who trusts the table has been handed a format where substance should have been.
I have watched this exact substitution before, at the code level, where the consequences are enforceable rather than reputational. In 2020, during the first DeFi Summer, I spent weeks reading early Uniswap and Compound contracts line by line while everyone else chased yield. I found a logic error in a lending protocol's interest-rate calculation — a minor miscalculation in isolation, but one that compounded under specific utilization conditions. I reported it privately to the core team before it was exposed. The lesson was never that the code was buggy; code is always buggy. The lesson was that the audit report covering that function carried a green checkmark for "interest rate logic" because the auditor had confirmed a function existed, not that the function was correct. The box was filled. The substance was absent. Code is law only if the audit trail is unbroken — and an unbroken trail requires an actual entry, not a placeholder that reads like one.
Let me make that concrete across the sectors where empty templates do the most damage, because the abstraction hides the cost.
Layer 2. The transmission map in the report I received traced ecosystem dependencies up and down a chain of infrastructure, protocol, and application layers. Every node read "N/A." This is the analytical equivalent of what the Layer 2 market actually did to itself. By 2025 there were dozens of rollups and validiums, each with its own bridge, its own sequencer, its own incentive program, all competing for the same finite population of real users. The aggregate TVL number rose, because the number of chains rose. The number of recurring, non-incentivized, paying users barely moved. A template that lists fourteen chains as "ecosystem partners" produces a table that looks like growth and actually measures fragmentation — the same scarce liquidity sliced into fourteen thinner pieces, none of them deep enough to absorb a real order without slippage. The map is not the territory. When the map has fourteen pins and the territory has one road, the template has lied without printing a single false statement.
DeFi. The token economics section of the empty report included a supply table with rows for "team," "early investors," "community/liquidity," and "treasury/ecosystem fund." Every cell: "N/A." But the shape of that table is itself a claim. It implies a token with a standard four-way split, a vesting schedule, and a liquidity-mining program — because that is the only kind of token the template knows how to describe. That default shape has been the industry's default failure for five straight years. Liquidity mining APY is the project subsidizing its own TVL figure; the moment emissions taper, the mercenary capital leaves and the "community/liquidity" row collapses back to its organic floor, which is usually close to zero. A template that always draws four rows will always find four rows of risk, and will never tell you that the one line that actually determines whether any of it is real is the ratio of emissions to retention — and that line is not in the schema at all.
NFTs and digital assets. The report's narrative section assessed a story cycle and concluded "N/A." Consider what actually happened to the creator economy the template's asset class was supposed to support. The royalty model depended entirely on marketplaces choosing to enforce it. When the dominant venue reduced enforcement to win back volume, the revenue floor for creators dropped to whatever secondary sales would voluntarily pay, which is functionally near zero. There is no sustainable on-chain business model for a creator whose only leveraged asset is a royalty that the venue can switch off with a settings change. The template has a "royalty" line item and fills it "N/A." It cannot express the structural fact that the line item was never enforceable to begin with. Once again: a filled box disguised as an empty one, and an empty insight dressed up as coverage.
Regulatory compliance. The report carried a completed Howey test — four factors, four cells, four "N/A"s. This is where the empty template turns actively misleading, because regulatory boxes are the ones institutional readers trust most. In 2024, when the first wave of spot Bitcoin ETFs was approved, I read the SEC filing documents line by line, hunting for the constraints that would actually bind liquidity: the custody arrangement, the surveillance-sharing agreements, the creation-and-redemption mechanics. Those documents did real work because every clause was specific and every claim was citable. A Howey table with four empty cells does the opposite. It signals "we thought about securities law" while saying nothing about whether the asset in question has any of the four elements. It borrows the authority of a compliance framework and spends it on a blank page.
The deeper pattern is that templates optimize for completeness, and completeness is the enemy of the specific. A good piece of analysis says one true, falsifiable thing. A template cannot say "the only signal that matters here is X," because the schema has already pre-decided that nine things matter equally. When the analyst is forced to weight each of the nine the same, the one real signal is diluted to one-ninth — and if the inputs are missing, even that ninth evaporates. You are left holding the shape of rigor and the weight of nothing. The document is long enough to feel expensive and empty enough to be worthless, and length itself becomes the disguise.
I want to be precise about where the blame sits, because "AI wrote a bad report" is the lazy diagnosis and it lets everyone off the hook. The empty template predates any useful language model. I watched human junior analysts produce the same artifact in 2018, filling due-diligence documents with "team not yet identified" and "tokenomics TBD" and shipping them anyway, because the document had to ship by Friday and it had to hit the page count the partner expected. The model did not invent the hollow report. It industrialized a workflow that humans had already normalized and never fixed. The schema was the original sin. The model was merely faster at committing it.
There is a second-order harm that the industry systematically underrates, and it is the one I care most about. An empty report is not neutral. It occupies the slot where a real report should have gone. It gets retweeted as "comprehensive." It gets cited in the next analyst's work. It builds a citation trail — an audit trail, in my language — that points to nothing, and then the following analyst cites the citation rather than the underlying data. In 2021, at the peak of the NFT market, I built an automated script to track whale wallet movements and minting patterns across multiple blocks, because the headline volume on a flagship collection looked impossible. It was. Roughly 60 percent of initial volume traced to wash trading between wallets I could follow end to end, hash to hash. The published "organic growth" figure rested on a chain of reports that had all cited each other and not one of which had reconciled the transaction hashes. That is what an empty-but-cited template produces at scale: a consensual hallucination with footnotes.
The counterintuitive reading is that the N/A report is not the enemy of good research. It is its most honest confession.
Consider the alternative. Take the same empty inputs and remove the integrity. Now the analyst, or the model, feels the pressure to fill the cells with confidence instead of placeholders. The token economics table gets plausible-sounding percentages. The team section gets a paragraph about "experienced operators" with no names attached. The risk matrix gets "medium" in every row, because "medium" is the answer that invites the fewest follow-up questions. This report reads far better than the N/A report. It ranks higher, travels further, and is far more dangerous, because it launders missing data into apparent knowledge. The N/A report at least drops the body where someone can find it. The confident report hides it behind a smile and a chart.
This is why my contrarian claim is uncomfortable: the "N/A — insufficient information" string is the single most honest token in that entire 4,700-word document. It is the only place where the report tells the truth about what it knows. Every cell reading "N/A" is a cell that refused to hallucinate — and refusing to hallucinate is a discipline, not a failure. The scandal is not that the report contains those strings. The scandal is that the schema made them invisible, that a reader skimming forty-three tables perceives structured effort and never perceives structured ignorance. The document optimized for the reader's trust instead of the reader's decisions.
The real fix is not to kill the template. Templates are how you catch the unlock cliff and the hidden custodian, and I will keep using them. The fix is to make emptiness loud. A schema should render missing inputs in red, at the very top, with a count and a denominator. It should refuse to publish when coverage crosses below a threshold. It should force the analyst to state, in one sentence, the single falsifiable claim the report supports — and if that sentence comes back empty, the report should not exist. The industry's problem was never too little structure. It was structure with no floor, and a floor is the only part of a building that you cannot fake.
The next time a research document arrives in your inbox, do not read the tables first. Read the empty cells. Count them. If the document has forty-three tables and eleven "N/A" strings, the useful number is not forty-three — it is eleven, and every one of those eleven is telling you exactly what nobody verified. Then ask the only question that survives contact with an empty template: which single cell, if filled with a real number, would actually change your position? If the report cannot answer that question, it has handed you a format to admire and a signal to ignore. Watch the emissions-to-retention ratio on the next incentive program you evaluate, and watch the retention curve thirty days after the emissions stop. That number is rarely in the schema. It is always the one that matters.
