Last Tuesday morning, a 42-page research report arrived in my inbox. Gradient cover page. A compliance disclaimer tailored to four separate jurisdictions. A table of contents with eleven numbered entries. And on page 19, under the heading "Token Economics," a table with eleven rows.
Every row said N/A.
I read the whole thing twice โ not out of confusion, but out of a strange professional admiration. Somewhere across those 42 pages, a team of analysts had engineered a document containing zero falsifiable claims about a protocol holding roughly $740 million in user deposits. Not one metric could be proven wrong. Not one figure could be independently checked. Not one sentence would ever have to be retracted, because none of them had ever really been said.
That report is not an anomaly. Over the past seven months I have collected 61 research documents on mid-cap DeFi protocols and scored them against a single question: if this thesis is wrong, what specific, observable event would tell me so? Forty-three of them had no answer. They were structurally unfalsifiable. They were scaffolding pretending to be signal.
What unsettles me is that I don't think the analysts were lazy. I think they were honest โ more honest, possibly, than the ones who filled the blanks with confident numbers.
When I started building in this industry, I was a junior developer in Los Angeles with an unshakable faith in the compiler. Code either ran or it didn't. Late 2017 disabused me of that quickly. I had personally introduced fifteen friends to an ecosystem that included a project called MyToken, a name I now use as shorthand for an entire category of failure. I watched people I had known since high school liquidate savings accounts on the strength of a whitepaper that was, in retrospect, beautifully typeset nothing.
Their money didn't disappear because the code had a bug. The code did what the code said. Their money disappeared because the documents surrounding the code were designed to produce a feeling rather than a fact. Trust is the only protocol that matters, and MyToken had built a user interface for trust while running a database for extraction.
So I started keeping a folder. Fifty failed projects, meticulously documented. I wasn't hunting for exploits anymore โ I was hunting for rhetorical patterns. And the pattern was remarkably consistent: the more pages a project produced, the fewer claims it actually made. Length was a substitute for specificity. Volume was the camouflage.
By mid-2020 I had co-founded Ethos Circle, a Discord community for non-technical professionals trying to understand yield farming. We onboarded 2,500 members. Then October 2020 happened, exploits tore through the ecosystem, and I spent 72 straight hours moderating channels and translating postmortems into plain-language safety checklists. We kept 85% of our members through that panic โ not because we had better information than anyone else, but because we were willing to say we don't know yet out loud, and then go find out.
That lesson is the reason I'm writing this. In a sideways market, the pressure to fill space with confident-sounding analysis is enormous. The cost of succumbing to it gets paid later, by people who made decisions.
Here is what template debt looks like from the inside.
There are roughly six archetypes that recur in crypto research, and each has a signature failure mode. The "ecosystem overview" inventories every partnership announcement without ever asking whether any of them produced a transaction. The "team deep dive" lists credentials without asking whether those people still work on the thing. The "tokenomics breakdown" shows a vesting cliff chart without asking who controls the multisig governing emissions. The "competitive landscape" compares TVL figures across protocols using different accounting standards. The "roadmap" is a list of verbs. The "risk section" is one paragraph about smart contract risk, followed by four pages about the risk that the token might go up.
None of this is dishonest in the narrow sense. It is a genre that has optimized for the appearance of completeness, and completeness is precisely the wrong target. Completeness is what you produce when you cannot afford to admit what you don't know.
So I built a personal rubric. It has four questions, and I want to walk through them, because the answers tell you almost everything about whether a piece of research earned the time it took to load.
The first: does the document contain a number that could have been different? This sounds trivial. It isn't. A number that could not have been different is decorative. "The protocol supports EVM-compatible chains" is decorative. "In the last 30 days, 1,840 unique addresses executed more than one swap, down 34% from the prior period" is a number that could have been different. If the research contains none of those, it has not measured anything. It has described a category.
The second: is there a named mechanism by which value accrues to the token holder, or is the assumption load-bearing? The phrase "the token will capture value as the ecosystem grows" is used so often it has stopped registering as a claim. It is a claim. In the large majority of cases I've examined, the mechanism is simply assumed. I have watched protocols with $300 million in TVL and a governance token that has never once received a protocol-level revenue distribution. The research about them ran 40 pages and never asked.
The third: what is the failure condition? Not "what are the risks" โ that produces boilerplate. What specific sequence of observable events would prove the thesis wrong? If someone cannot state that, they do not have a thesis. They have a mood.
The fourth, and the one I care about most: who is the analysis for? This is the ethical-auditor question. If the answer is a reader who must decide within 72 hours, the document should be short, falsifiable, and honest about its gaps. If the answer is a reader who wants to feel like they understand something, the document will be long, confident, and inert. Most crypto research published during consolidation is the second kind โ and it is not a coincidence that consolidation is when volume gets published.
Now let me be concrete, because I have a low tolerance for abstraction presented as insight.
Take the current DeFi development landscape. Uniswap V4 is the most architecturally interesting thing to happen to automated market makers in years. Hooks turn the DEX into programmable Lego, letting developers attach custom logic to liquidity pools at defined lifecycle points. I wrote approvingly about the design when it landed, and I still think it is elegant. But elegance at the primitive layer is not adoption at the builder layer.
Here is the number that matters, and it is absent from most V4 research I've read. The jump from V3 to V4 is not a version increment; it is a change in kind. V3 lured developers with concentrated liquidity โ novel, but expressible in a diagram. V4 asks them to reason about the hook lifecycle, the ordering of callbacks relative to the swap, the security boundary between a hook and the pool it modifies, and the fact that a badly written hook creates an entire class of extraction that no existing audit checklist covers.
I have spent time this year with builders who tried to ship on V4. The pattern is not "this is too hard to learn." It is "this is hard to be confident about." Those are different problems, and the second is much harder to solve, because confidence is not a documentation deliverable. My working estimate is that somewhere around 90% of developers who express intent to ship a hook-based protocol will not ship one โ not because tooling fails them, but because the complexity spike forces a re-evaluation of whether the hook is load-bearing at all. Most of the time it isn't. The thing they wanted to build could have been a vault, or a router, or an off-chain service.
That is not a criticism of Uniswap. It is a criticism of an analysis culture that treats a launch as a completed fact. The interesting question about V4 in a flat market is not what hooks can do. It is how many hooks are in production, how much liquidity they hold, and how many of their authors would ship the same hook again knowing what they now know. That third question is the one nobody asks, and it is the one that predicts the next two years.
Bitcoin deserves the same treatment, and this is where my discomfort runs highest.
I want to be precise, because this position has cost me a few arguments. I don't believe the ETF approval was a fraud. I believe it was a genuine, hard-won, legally legitimate milestone. I also believe it completed a transition the original design did not anticipate, and pretending otherwise is nostalgia dressed as analysis.
The observable shift is this: the marginal Bitcoin buyer is now a model-portfolio allocator responding to a risk-parity framework, not a person who wants to be their own bank. That is nobody's moral failing. It is simply a different game. Yet a great deal of the research I respect has refused to update. It still opens with "peer-to-peer electronic cash" and then spends 30 pages discussing flows into a wrapper that requires a custodian, a prime broker, and a Thursday settlement window.
Code is law, but people are the context. The context changed. The code didn't. The scarcity guarantee is intact and the ownership distribution is not, and both facts are true simultaneously. In consolidation, that distinction matters more than it does in a bull market, because the correlation between spot flows and price is tighter and the noise that used to mask it is gone.
Then there is the omnichain narrative, which I have grown genuinely tired of.
I have read dozens of documents this year about applications deployed across seven, nine, twelve chains. The vocabulary never varies: unified liquidity, seamless UX, chain abstraction, users don't know which chain they're on. That last claim is the tell. If users don't know which chain they're on, then chain count is not a user-facing feature. It is a fundraise-facing feature. The number of chains your contracts are deployed to is a metric about your cap table, not about your users.
What I want from any piece of cross-chain research is a cohort curve. Not total transactions. Not cumulative addresses. A cohort curve. Take the wallets that first interacted in month one and ask how many still interact in month four, in month nine. I have pulled a handful of these for multi-chain deployments, and the shape is consistent: a launch spike driven by incentives, a decay curve that resembles every other incentive-driven decay curve, and a long tail composed largely of wallets active on exactly one chain โ the one paying.
Across roughly a dozen multi-chain deployments I've examined, the share of monthly active wallets transacting on more than two chains in the same month has consistently come in under 5%. That is the number. Everything else is marketing copy formatted as a chart.
I want to name the contrarian thing clearly, because it is what I have been circling.
The instinct when you encounter forty pages of N/A is to treat it as a failure โ a gap, a hole where knowledge should be. I have come to think that instinct is backwards. The N/A report may be the most truthful document in the folder.
Here is what I mean. An analyst operating in a flat market faces a specific pressure. Volume expectations do not fall because the data got thin. Editors want a piece. Clients want a view. The social cost of writing "there is not enough observable information to form a thesis about this protocol this quarter" is enormous, and the reward for producing a confident framework is immediate. So most analysts fill the space. They substitute structural completeness for epistemic honesty, and the reader receives a document that feels like knowledge and functions like noise.
The N/A report is a refusal. It is a public act of not-knowing. In an industry whose dominant failure mode has been confident people saying confident things about things they could not observe, that refusal is not a deficiency. It is disclosure.
There is a harder version of this, pointing the other way. Some formats we treat as diligence do not lack information โ they over-supply it. The 80-page technical audit that produces a severity table and never states whether the team's economic incentives align with users. The security review that examines the contract and not the multisig. The transparency dashboard publishing 400 metrics, none of which is the concentration of governance power. Volume in these cases is not incomplete. It is misdirected. And misdirected volume is worse than a blank page, because it invites the reader to conclude the important questions have been covered.
This is where Community over coin, always stops being a slogan and becomes an audit criterion. When I evaluate a protocol, I ask a question no template contains: who has the standing to say no? Not who holds the votes โ who has standing. Can long-term users block an emissions change that dilutes them? Can a contributor raise a concern without risking their compensation? During the winter of 2022 I ran weekly town halls through Project Phoenix and mentored 50 junior developers through career pivots. We lost 40% of our community to despair before stabilizing, then grew 20% as people came looking for something that did not move. What I learned is that the governance layer that matters is not the one in the contract. It is the one in the room.
Anonymity is a shield, not a lifestyle. I mean that operationally. The pseudonymous contributor who ships code and disappears is a legitimate and often necessary participant. The anonymous delegate holding 4% of supply with no identity at stake is a structural risk โ and no research document I have read this year even has a field for it.
So where does this leave us, in a market that is flat and waiting?
I have a different proposal than the one I usually hear. The standard advice during consolidation is to do more research. The honest advice is to do less research and make it falsifiable. Write down three numbers about the protocol you care about. Write down the event that would prove you wrong. Write down who benefits if you're right. Then close the laptop and wait โ because the sideways market is not asking you to hold more opinions. It is asking you to hold fewer, better-anchored ones, long enough for the data to arrive.
In 2025 I helped convene the Values-Based Crypto Alliance, 30 community leaders and institutional representatives drafting ethical engagement guidelines we called the LA Principles. The hardest argument we had was not about privacy or consent. It was about disclosure standards. The institutions wanted to know what a good-faith report looks like. The community members wanted to know what a bad-faith one looks like. We could not agree, and the reason we could not agree is that this industry has never defined the difference between a document that informs and a document that performs.
That is the project the next cycle actually needs. Not another framework. Not another 42 pages.
A definition.
When a research report leaves eleven rows blank, is that an admission of ignorance or a confession of something else โ and which of the two have we been paying for?