I opened a "deep analysis report" last week that ran four thousand words. Every cell in its risk matrix read N/A. Every revenue projection: insufficient data. It carried a disclaimer, a glossary, and a star-rating system that awarded zero stars across all four dimensions โ technical value, investment value, timeliness, reference. Structurally, it was flawless. It was also a confession: a document built to look like diligence, admitting on every line that the analyst had nothing to work with.
In 2017, at twenty-three, I audited forty whitepapers for a Baltic ICO platform. Eighty percent had no economic viability โ not weak economics, absent economics. I built a values-first review framework because a token that doesn't know what it's for doesn't have a tokenomics problem. It has an honesty problem. Eight years later, the format changed. The honesty didn't. We automated the hollowness and gave it a template.
The crypto research industry became a machine somewhere between 2023 and 2025. The count of published protocol reports โ exchange desks, boutique funds, solo analysts with a Substack โ grew faster than the count of protocols. Almost none of that growth came from hiring. It came from generation. A large language model doesn't sleep, doesn't demand carry, and will produce twelve thousand words about a whitepaper it cannot read.
Bull markets don't reward skepticism. They reward throughput. When a token moves thirty percent in a week, a fund needs a document that looks like diligence before the IC meeting, and the document is a deliverable, not a discovery. I know this because I've been on the institutional side of it: in 2025 I drafted a paper arguing that institutional capital can accelerate decentralization if it's governed by DAOs rather than corporations, and three banks cited it. I watched the bankers read it. They read the abstract, the risk section, and the conclusion. The middle โ the part where the work lives โ got skimmed. That is where the hollowness begins.
So the incentive is set. Output volume is legible. Rigor is not. A report with N/A in every cell costs nothing to produce and looks identical to its peers in a PDF viewer โ same fonts, same tables, same footer. The market cannot tell them apart, which means the market will keep buying them.
The regulatory backdrop sharpens the incentive. When OFAC sanctioned Tornado Cash in 2022, it didn't just freeze a mixer; it signaled that publishing immutable code could be treated as a sanctionable act. The Fifth Circuit pushed back in 2025, ruling that immutable smart contracts are not "property" under the relevant statute. But the chill had already landed. Developers who write open-source tools now price legal risk into every commit. In that climate, an analyst who publishes N/A across a page isn't lazy. They're insured. Emptiness is a liability shield.
The anatomy of a hollow report matters because it tells you exactly where the pipeline broke. Start upstream. A generative system handed a headline and no body text cannot produce analysis; it can only produce the shape of analysis. Headers, tables, a risk matrix with color coding, a confident disclaimer. The skeleton survives. The substance evaporates. What you're reading isn't a failure of the model. It's a faithful rendering of its inputs, which were empty.
This is the inverse of what I learned in 2020, dissecting Compound's governance at a Warsaw audit firm. Compound's mechanics were legible. You could read the timelock, the quorum threshold, the delegation graph, the COMP distribution schedule. The information existed; the failure was translation. I wrote "Governance is Politics, Not Code" because the politics was sitting in the open, unread. When it hit ten thousand reads, I understood something: the scarcity was never data. It was the willingness to read it.
The empty report is a different failure entirely. Its problem is upstream โ nobody supplied the evidence. Someone asked a question and forgot to attach the thing being questioned. The system did the only reasonable thing: it reflected the absence back at them.
A report with substance would tell you that Uniswap V4's hooks turn the DEX into programmable Lego โ the same composability that lets a developer write a dynamic fee curve lets them write a fund-draining edge case, and the complexity spike is quietly scaring off most builders before they ship. It would tell you that cross-chain bridges have absorbed more than $2.5 billion in cumulative exploits โ Ronin, Poly Network, Wormhole, Nomad โ while the industry keeps routing user funds through them anyway. Those sentences require reading. A system given no text cannot write them. A template given no text writes around the hole, filling it with the aesthetics of care: bolded headers, confidence intervals, a glossary.
Here's where I refuse the easy conclusion. A report that says N/A is more honest than one that says everything confidently. In 2021 I ran an NFT marketplace campaign for fifty women creators; it did 500 ETH in volume and drew a wall of sexist noise in the community channels. The reports written about that project were full of certainty and wrong about all of it. Blank is a posture I can respect. Fabrication is not.
The real danger isn't the hollow report. It's the fluent one โ the automated diligence memo that fills every cell with plausible numbers: a 12% APR, a 40% unlock next quarter, a team "ex-Google," none of it verified, all of it formatted to survive a skim. Humans fill silence. Models fill templates. Neutrality is a story we tell ourselves about the code we haven't read โ and the blank template is the only genuinely neutral document in the room. It told me nothing, and it told me honestly.
The question worth sitting with isn't whether we can generate more research. It's whether we'll pay for the part that can't be generated โ the reading, the disagreement, the audit that returns an inconvenient answer. True ownership begins where the server ends. Debate is the compiler for better consensus. A report that ends in N/A isn't the end of analysis. It's the only honest place to start one. The blank page is not a verdict. It's an invitation.