We didn't build a news filter to be clever. We built it because we were tired of being lied to by our own certainty.
A few weeks ago, our small research group in Manila fed a piece of breaking crypto news into the analysis pipeline we had spent most of a year assembling. The headline was loud. The claims were confident. The pipeline returned nothing โ not a crash, not an exception, but a refusal. Nine analytical dimensions, the entire scaffolding we had built, all arrived at the same verdict: the input was empty. There was no list of information points to reason over. There was no who-did-what, no number attached to anyone's action. There was only the shape of a story, walking around in the clothes of a fact.
That single non-output taught us more about this market than any report we published all year. This is a piece about that lesson โ about what happens when you build an honest filter and the honest answer is "I cannot analyze this," and about why, in a sideways market where direction is scarce and conviction is expensive, that refusal may be the most valuable skill a crypto reader can learn to perform.
Context: the market we stopped trusting
For ten years I have watched the crypto news cycle do the same thing in slow motion. A token launches, a narrative attaches to it, a number arrives from nowhere, and thousands of people โ colleagues, students, the small-business owners I now teach basic wallet hygiene โ inherit the risk of a sentence someone else invented. In 2021 I watched the mania from the inside. I was a final-year computer science student, and my entire dormitory went into simultaneous financial collapse over a wave of NFT launches. We didn't retreat into cynicism. We organized a weekend workshop for forty peers, taught them hardware wallets, taught them how to verify a smart contract's source, and then I manually audited the top five trending NFT projects by hand. One of them was a rug pull. I flagged it two days before launch, and that intervention saved an estimated fifteen thousand dollars in combined student savings. The experience taught me something no trading course will: technical literacy is a form of social protection.
Since then, the stakes have only risen. Spot Bitcoin ETFs pulled crypto into the institutional bloodstream, and the divide between people who read the prospectus and people who read the headline has become the defining fault line of this cycle. Bitcoin changed too. Once described as peer-to-peer electronic cash, it now trades like a risk asset wearing a Wall Street ticker โ held in custodial wrappers, its velocity determined by macro calendars rather than by the merchants who were supposed to accept it. That is not a failure of the technology. It is a migration of purpose, and it means the average reader sits further from the truth than ever, because institutions have every incentive to be loud and no incentive to be legible.
In 2025 that gap became personal. I founded an academy aimed at translating complex regulatory frameworks into accessible guides for small businesses, and we partnered with three local banks to build a curriculum for five hundred SME owners โ compliance basics, wallet security, the difference between custody and self-custody. We secured a small regional grant to keep it running. What struck me was not how little they knew. It was how much confident, wrong information they had already absorbed. They had been reading headlines for years and mistaking the volume of coverage for the weight of evidence.
So we built the filter. Nine dimensions, each of which must be fed by real information points before it is allowed to speak. We borrowed the discipline from audit work: if the fact is not there, the conclusion is not available. What follows is what each dimension asks โ and, more importantly, what each dimension refuses to accept.
The nine dimensions, and what each one will not let you fake
Dimension one, the technical layer. Before anything else, we ask where in the stack a project actually lives. Is it a base-layer consensus change, a rollup, an application, an oracle, a bridge? This matters because the same headline โ "we now support zero-knowledge proofs" โ means completely different things depending on the answer. A base layer claiming a proof system is making a claim about its validity model; an application claiming it is making a claim about a library it imported. We also hold the line on audit status. An audit is not a guarantee; it is a snapshot of assumptions. The real question is never "are they audited" but "what was assumed when they audited." Who holds the upgrade key? Is the contract upgradeable through a proxy, and if so, who controls the proxy admin? Is there a trusted setup, and who ran the ceremony? Bridges and cross-chain messaging layers carry the heaviest technical risk of all, because their security model is an accumulation of every assumption they borrow from the chains they touch. These are the questions that separate engineering from marketing, and our filter will not answer any of them from a press release alone.
Dimension two, tokenomics โ the soundness test. Every token is one of a few types: gas, governance, utility, or a claim on liquidity. The classification sounds academic until you apply the structural test I learned the hard way during the DeFi winter of 2022. In that bear market I led a resilience collective of about two hundred members; we collectively audited lending protocols and contributed fifteen findings to major open-source projects through public contests. What we learned is that only one question really matters: where does the yield come from? If the yield is paid from real protocol revenue โ fees, interest, liquidations โ you are looking at a business. If it is paid from new deposits, you are looking at a structure that can only survive by growing forever, and nothing grows forever. Vesting is the second half of the test. Unlock cliffs are the calendar of future supply. A token with a large cliff twelve months away is a token whose price is, in part, a countdown.
Dimension three, the market face. This is where most readers stop, and where we insist on continuing. A piece of news is one of two things: a promise or a delivery. Promises get priced on announcement. Deliveries get priced on realization, and the gap between the two is where people lose money. We also track positioning โ the funding rate and open interest โ because positioning tells you who is trapped. When funding is heavily positive while price stays flat, the market is long and wrong, and a surprise in either direction will liquidate one side violently. Here is the honest limitation, and we say it out loud rather than guessing: funding rates, total value locked market share, and governance participation cannot be read from a text. They require live data from on-chain and third-party sources. A filter that invents these numbers is worse than no filter at all.
Dimension four, the ecological niche. Every protocol sits inside a supply chain, and its fate is largely determined by the health of its neighbors. We map upstream and downstream dependencies: who does this protocol rely on for liquidity, for data, for security? Then we look for the two signals that cannot be faked โ developer and user retention. Commit activity over time, not a single impressive week. Wallet retention curves, not total addresses created. A protocol can buy attention, but it cannot buy retention at a price that scales. In 2024 I ran a pilot project integrating a decentralized compute network with autonomous agents for content verification here in the Philippines, working alongside a team of developers and two sociologists. We processed ten thousand data points and cut misinformation by roughly forty percent. The lesson was not that the technology worked. It was that retention in that system depended entirely on whether users trusted the verification layer โ and trust, unlike compute, does not scale by paying for it.
Dimension five, regulatory posture. The four prongs of the classic securities test โ investment of money, common enterprise, expectation of profit, reliance on the efforts of others โ brush against almost every crypto token in existence. Our filter forces a jurisdiction-by-jurisdiction read: is this asset classified, tolerated, or contested in the markets where it actually trades? Is there KYC and AML at the on-ramp? Europe's framework, the shifting enforcement posture in the United States, and the licensing regimes across Southeast Asia each produce a different answer, and that answer resets the risk floor of everything stacked above it.
Dimension six, team and governance. Is the team doxxed? What is the vote concentration โ the number of addresses required to control a majority? Who invested, and at what tier, because a lead backer with a board seat is a different animal from a community grant. And roadmap delivery rate, measured backward, not forward. A team that has shipped eleven of twelve commitments is a very different bet from a team that has shipped two of twelve and is now promising nine more.
Dimension seven, the risk matrix. Six categories, always the same six: technical, market, operational, regulatory, competitive, and narrative. Technical risk lives in the code. Market risk lives in positioning. Operational risk lives in the multisig and the people behind it. Regulatory risk lives in the courts. Competitive risk lives in the next launch. Narrative risk lives in the crowd. Most failures are compound โ a technical flaw that becomes a narrative collapse, because a community was promised a certainty the code never offered.
Dimension eight, narrative expectations. Narratives have a life cycle: emergence, consensus, saturation, exhaustion. The filter asks a single uncomfortable question โ how much of this story is already in the price? When fully diluted valuation runs far ahead of revenue, you are holding a belief, not a claim on cash flows. The expectation gap is the distance between what the story promises and what the numbers deliver, and it is usually where value is either created or destroyed.
Dimension nine, transmission across the industry chain. Finally, we ask who else feels this, and when. A change at the base layer reaches exchanges within days, infrastructure providers within weeks, applications within months, and users within a year. Six domains matter: infrastructure, exchanges, applications, users, regulators, and capital. A piece of news that touches none of them is not news. It is content.
The contrarian turn: the refusal is the product
Here is the part that unsettles people.
The most honest output of any analysis system is "I do not know," and in crypto that output is systematically under-produced because the market pays for confidence and punishes hesitation. We did not design our filter to be right more often than everyone else. We designed it to be unable to lie. When the information point list is empty, the filter halts. It does not fill the gap with narrative, because narrative is the gap.
This is why the framework I just described can be dangerous in the wrong hands. Give it to someone who needs a signal every hour and it becomes a machine for manufacturing certainty โ nine tidy boxes, each one stuffed with a plausible guess, producing a report that looks analytical and is actually astrology with footnotes. The structure becomes the new lie. The boxes make the invention feel earned.
We watched exactly this happen with the cross-chain narrative. For two years the market was told the future belonged to "omnichain applications," contracts deployed across eight or twelve chains at once, and the frameworks dutifully filled in their boxes: technical layer, check; ecosystem, check; roadmap, check. What almost nobody asked was the only question a user actually has โ do you care how many chains the contract lives on? You do not. You care whether it works, what it costs, and who is safe. The complexity was overwhelmingly a feature of the fundraising story, not of the user's experience. A framework that does not begin with the user's question will happily rationalize the investor's answer.
That is the contrarian core of this piece. The value of the nine dimensions is not positive. It is negative. It is a filter precisely because it kills things. It kills the story with no information point. It kills the yield with no revenue. It kills the roadmap with no shipment. And when it kills enough stories, what remains standing has earned a different kind of trust โ the kind that never needs to be asserted, because it was never granted cheaply. So when our pipeline returned nothing for that loud headline, it was not broken. It was working. And in a sideways market, where the direction everyone wants is nowhere to be found, the willingness to say "there is nothing here" is itself a positioning decision. It keeps your capital idle when it should be idle.
Takeaway: a thought for the agent age
We are walking into a year when autonomous agents transact with other autonomous agents, when wallets make decisions without a human present, and when the volume of machine-generated claims will make today's news cycle look quiet. I have spent the past year producing a series of conversations asking what it means for machines to hold value at all, and the answer keeps returning to the same place: trust cannot be automated if it was never defined in the first place.
The protocols that survive the agent economy will not be the ones with the most confident voices. They will be the ones whose data is verifiable and whose silence is honest. So the question I leave with you is not which token to buy in this chop, but something prior: when a claim reaches you next week, dressed in the clothes of a fact, will you have a filter โ or will you have a feeling? We didn't build ours to predict the market. We built it so that when the input is empty, we would have the courage to say so.