Nine Dimensions of Nothing: Why Crypto's Most Dangerous Asset Is a Blank Data Field

I opened the file on a Tuesday. Nine tabs, one for each dimension my team audits before a protocol is allowed anywhere near our curriculum. Technical architecture. Token distribution. Market structure. Ecosystem position. Regulatory posture. Team and governance. Risk matrix. Narrative. Supply-chain transmission. Every tab carried the same content — a single word repeated in every cell, in the same neutral font, with no apology attached: Not applicable. Not pending. Not under NDA. Not the audit lands after mainnet. Not applicable. Someone had built a nine-dimension template with obvious care and then filled it with a uniform refusal to be described. I have spent twenty-two years reading crypto documents. I have read whitepapers written across three languages, token tables so convoluted they needed their own glossary, and pitch decks that described the replacement of global settlement infrastructure in eleven slides. I had never read anything this honest. It was 11 p.m. when the file opened. I stayed until three in the morning — not because the file was empty, but because of what the emptiness disclosed. Truth decays slowly. Usually you have to wait years to watch it happen. A blank due-diligence file is not an edge case in a bear market. It is the modal case. Walk any listing aggregator right now and count the assets whose documentation link resolves to a Notion page with three bullets and a Telegram invite. Count the ones whose GitHub last accepted a commit in 2022 and whose last release tag is still marked pre-alpha. Count the audits that are PDFs with no date, no commit hash, and no scope statement — documents that certify nothing because they were never asked to certify anything in particular. I have done this count twice in the past fourteen months, and the ratio has moved in one direction only. The industry has built an entire analytical apparatus — dashboards, scoring models, sentiment indices, on-chain intelligence platforms — on the assumption that the truth is present and merely buried. Almost none of that apparatus is designed for the case where the truth is simply not there. We are extraordinarily good at excavating. We are nearly helpless in front of an empty room. There are three states a protocol can occupy with respect to disclosure, and conflating them is the most expensive mistake a retail analyst makes in a bad market. The first is opacity. Information exists and is withheld. Opacity is a strategy. It has an author, an intent, and a timeline. It can be negotiated with, waited out, or leaked. The second is immaturity. Information does not yet exist because the thing it describes does not yet exist. Immaturity is a phase. It ends — either in a shipped product or in a dead repository — and a disciplined analyst can usually date the terminal condition. The third is absence. Information does not exist and there is no mechanism by which it will come to exist. Absence is not a phase. It is a structure. It does not resolve, because nothing is generating the pressure that would resolve it. The file on my desk was the third state. And the third state is precisely the one our industry has the least vocabulary for, because our entire analytical culture is premised on hidden truth waiting to be excavated. Every scoring model I have ever used assumes the existence of an object. None of them contains a cell for the object not being there. In a bull market, absence is invisible. Price fills the space. Narrative is a solvent — pour enough of it in and any void starts to look like a container. A token with no documentation and a rising chart generates more conviction than a token with a complete specification and a flat chart, because the chart is read as evidence that someone else has done the work. In a bear market, that solvent evaporates, and what remains is the container with nothing in it. I want to be careful about my own history here, because I have contributed to this problem. In late 2017, I spent three months translating the Tezos whitepaper and its technical FAQ into accessible Chinese content, and it reached more than fifty thousand readers. That documentation was dense, strange, and self-amending — a governance model that let the protocol rewrite itself through stakeholder vote. I did not agree with all of it. I did not need to. The point is that a forty-page governance specification existed, and my job was translation, not invention. I was standing on a floor. By 2020 I was writing ethical lending guides with the MakerDAO community, working through collateral mechanics with around two thousand individual users. When the collateral crisis hit that spring and liquidations cascaded through an auction system that was not designed for a fast market, I spent two weeks manually verifying on-chain data to explain to frightened people what had actually happened to their positions. That work was possible because every claim could be checked. Collateral ratios, oracle prices, auction bids — all of it was sitting in public state, waiting to be read. The transparency was not philosophical. It was operational. Then 2022 taught me something worse. Terra and FTX did not fail because documentation was missing. They failed because documentation was abundant and fabricated. FTX published more, faster, and prettier than almost anyone in the industry. The information was not absent. It was unverifiable, which is the more dangerous condition, because unverifiable information satisfies the appetite for diligence without performing any of its functions. I spent six months after that collapse away from commentary, auditing identity protocol code to understand how sovereignty gets implemented at the primitive level. I published a fifteen-thousand-word piece on dignity in decentralization that a very large number of people read, most of them, I think, looking for company rather than answers. That experience reframed my model of the enemy. The threat is not the empty file. The empty file is easy. The empty file tells you to leave, and you leave, and you keep your capital. The threat is the full file that has been constructed to feel like an answer. Which is why the file on my desk unsettled me in a way that a fabricated whitepaper never could. It was honest about being empty. And my framework — nine dimensions, each designed to produce a verdict — had no verdict to produce, because there was no object. So I did the only thing left. I analyzed the emptiness itself, dimension by dimension, and documented what each void would have contained. What follows is that audit. The first thing to establish is what an audit actually is, because the word has been hollowed out to the point of meaninglessness in this industry. A real audit is a bounded artifact. It names the auditor. It names the commit hash of the code reviewed — a specific cryptographic point-in-time, not a repository. It states scope, including what was excluded. It lists findings by severity, with the ones marked critical or high accompanied by evidence, remediation status, and a re-review. Without the commit hash, the document is a marketing asset. Without findings of any severity, the audit is either extraordinarily narrow or extraordinarily dishonest, because every nontrivial system has findings. The file contained none of this. Not because the audit was bad, but because the question had never been asked and the object it would have been asked about did not exist. Based on my own audit experience, the economics here explain a great deal. A serious third-party review of a mid-complexity protocol runs somewhere in the range of thirty to one hundred fifty thousand dollars over four to twelve weeks of calendar time, with the good firms booked out two to four months. In a bull market, that cost is noise against a treasury denominated in a token that is up eight hundred percent. In a bear market, it is one of the first line items cut, because it produces no visible artifact that a retail buyer will notice on a chart. The cut is rational at the level of the individual team and catastrophic at the level of the system. Code over hype. That phrase has been my working rule since I left my economics desk in 2018, and it has one implication that people tend to skip: if there is no code, the rule produces no output. Not a negative output. Not a neutral output. No output. You cannot evaluate what has not been instantiated. What I could evaluate was the structural signature of the void. And the signature is remarkably consistent. Protocols with the thinnest documentation carry the heaviest administrative surface: upgradeable proxy patterns with no declared upgrade authority, admin keys held by an address with no on-chain identity, multi-signature thresholds set low enough that a single compromised signer threatens the system. Three-of-five with named, distributed signers and a forty-eight-hour timelock is a governance posture. Two-of-three with anonymous signers and no timelock is a liability with a friendly logo. That correlation is not a coincidence. It is a selection effect. Teams building toward a defensible protocol write specifications because specificity is an asset in a competitive market — it attracts developers, integrators, and auditors, and it costs nothing to produce if the thinking has been done. Teams building toward an exit avoid specificity because specificity is a record. An explicit claim can be falsified. A vague claim can only be disappointed, and disappointment has no legal remedy and no on-chain proof. When you see an information void, you are not looking at laziness. You are looking at a risk-management decision, made deliberately, by someone who understood what a commitment would cost them. In 2026, my consortium built a verification layer that required human ethical sign-off for high-value autonomous transactions, and we piloted it with five hundred users. The design lesson that stayed with me was not about AI. It was about signature. The sign-off only carries moral weight if there is a defined object being signed off on — a transaction with parameters, a counterparty with an identity, a consequence with a shape. A human signature attached to an undefined action is theater with a biometric. The same is true of due diligence. A framework applied to a void produces signatures without objects. It manufactures the feeling of oversight and delivers none of it. There is no such thing as an unknown unlock schedule. There is only an unlock schedule you have not read. This is the single most important sentence in this article, and it is the one most people argue with, so let me be exact. Supply is a mechanical fact. A token contract defines a total supply. Vesting contracts, whether they are purpose-built, Sablier streams, Streamflow schedules, or plain token-lock contracts, define release curves as functions of block height and time. Distributions to team, investors, advisors, and treasury are recorded as transfers, and those transfers are permanent. If the schedule is not in the documentation, it is still in the ledger — and the ledger is strictly more reliable than the documentation, because the documentation can be edited and the ledger cannot be un-transferred. What this means practically is that the absence of tokenomics disclosure is not a knowledge problem. It is a labor problem. Somebody can reconstruct the distribution in an afternoon with a block explorer and a spreadsheet. The team is betting that nobody will. In my experience, they are usually right, and that bet has a very high expected value for them. So run the arithmetic, because the arithmetic is what the void is hiding. Take a token with a hundred million fixed supply. Assume twelve percent circulating at the observation date and a forty percent allocation to team, investors, and advisors. Assume a twelve-month cliff followed by thirty-six months of linear release. That is forty million tokens unlocking across thirty-six months, roughly 1.11 million tokens per month. Now assume that insiders liquidate only twenty percent of what they receive — a conservative assumption, since rational actors with a zero cost basis in a declining market tend to liquidate considerably more. That is roughly 222,000 tokens of genuine sell pressure per month against whatever net new demand exists. In a bull market, that number is absorbed without comment. New inflow exceeds it, the chart rises, and the unlock becomes a non-event that reinforces the belief that unlocks do not matter. In a bear market it is the entire market. The float is not deep enough to absorb it, the marginal buyer is not present, and the price has exactly one direction available. The unlock was always there. Only the absorbing capacity changed. The fully diluted value ratio is the diagnostic that survives the void, because it can be computed from two numbers. A token trading at eight hundred million dollars circulating against an eight billion dollar fully diluted valuation is telling you that ninety percent of the supply is waiting. That is not a valuation. It is a schedule. And a schedule with a date on it is not a risk you have to speculate about — it is a risk you can read off a calendar. Hold the line here, because this is where discipline gets tested. The absence of tokenomics documentation is not a reason to be confused. It is a reason to assume the unfavorable case, because the team that declines to publish a schedule almost never declines because the schedule is generous. Asymmetric disclosure is itself a signal, and it points in one direction. If information truly does not exist, then price is not discovering anything. It is reflecting the beliefs of whoever is currently transacting, and those beliefs are anchored to nothing except each other. That is the definition of a reflexive asset, and reflexive assets do not have price levels. They have price momentum. Consider the arithmetic of a launch where the entry price is set by the venue's own marketing rather than by any discovery process. I have been tracking the peak multiples of exchange-launched offerings across cycles, and the compression is stark. What cleared multiples in the tens to low hundreds in the 2019 through 2021 window now clears in the single digits to low teens. The mechanism is not mysterious. When the venue allocates the supply, sets the narrative, defines the entry through a subscription cap, and then lists the asset on its own order book, the economics of the trade are being set by the counterparty to the trade. Early buyers are not discovering a price. They are accepting a price. The margin that once accrued to subscribers has migrated upstream, to the venue and to the project, because both parties now understand that the demand is manufactured from their own user base. A closed loop can be profitable for the operator and structurally unprofitable for everyone standing inside it. What this does to market structure is measurable. When a single venue carries the majority of an asset's thirty-day volume, the order book is not a market. It is a display. Depth thins outside a narrow band, and the price you observe is a quote rather than a level — a number that exists because nobody has tried to cross it in size. In the token with no information, this configuration is the norm rather than the exception, because a distributed venue strategy requires a sophisticated market maker and a reason to deploy one. The void and the concentration arrive together. I want to resist the reflex to read manipulation into every empty structure. Most of it is simpler and more mundane. A protocol with no documentation has no observable demand curve, so no rational market maker will commit inventory to it, so liquidity concentrates on whichever venue is willing to list it for marketing reasons, so the effective price becomes a function of that venue's incentives rather than of any buyer's willingness to pay. The emptiness produces the concentration, and the concentration disguises the emptiness. You cannot tell which came first from the outside, and that ambiguity is the point. Network effects are countable. That is what makes the ecosystem dimension the least forgiving of the nine, because it is the one where the void cannot be dressed up. You can write a compelling narrative about a protocol with no users. You cannot produce integrations that do not exist. A protocol's position in a value chain is defined by three measurable things. What it depends on upstream — which chains, oracles, bridges, sequencers, and data providers it cannot function without. What depends on it downstream — which applications, vaults, index products, or automated strategies call it. And whether developers are still building against it, which is visible in commit activity over time rather than in commit totals. A protocol with no upstream dependencies and no downstream integrators is not a protocol. It is a repository. The distinction matters because repositories do not accumulate network effects — they accumulate stars. And the file on my desk contained no evidence of either direction. No dependency graph. No integration registry. No developer activity. Not because the ecosystem was small. Because there was no axis along which to place it. The developer signal deserves particular attention, because it is the metric most easily faked and most easily read correctly. Commit count is a vanity measure. Commit quality is not. A repository with two thousand commits in a single quarter and no test suite, no release tags, no issue triage, and a changing set of author identities is not a development effort — it is a stream of noise calibrated to a contribution graph. The reliable signals are boring: how long the median contributor has been contributing, whether the commit history shows sustained work between funding rounds, whether the code reviewed in an audit is the code that shipped. Boring signals are hard to fabricate because fabricating them requires actually doing the work. If you want a single proxy for ecosystem health that survives an information void, use retention. Transactions are cheap. Wallets are free. But a user who returns in week eight after depositing capital has made a decision that costs them something. Protocols that cannot retain a hundred users cannot retain ten thousand, and there is no marketing budget large enough to close that gap permanently. Retention curves are visible for any protocol with on-chain state, regardless of what its documentation claims. Which means that in the void, the answer is not missing. It is simply located somewhere nobody bothered to point you toward. The regulatory dimension of an information void is genuinely interesting, because the standard securities analysis requires an identifiable set of actors. The Howey framework asks whether there was an investment of money in a common enterprise with an expectation of profit derived from the efforts of others. Every element of that test presumes an other. Strip the other out and the analysis has no subject. Practically, though, regulators do not resolve the ambiguity in the direction the void would prefer. Anonymity does not produce exemption. It produces a different enforcement theory. When no team is identifiable, the target becomes the venue that listed the asset, the infrastructure that served it, and the promotional network that distributed it — all of which have identifiable operators, banking relationships, and reasons to avoid attention. Anonymity at the protocol layer redistributes legal risk downward and outward to the people who touch the money, and those people generally know it, which is why the venues with the most aggressive listings in a bear market are also the ones hiring the most compliance staff. The absence of a team is not the absence of a promoter. It is the perfect promotion. A named founder can be sued, subpoenaed, and discredited. A decentralized claim cannot be served with process. The structural incentive to remain unspecified is therefore strong, and it grows stronger the more legal pressure the industry absorbs, which means the information void I am describing is not a temporary artifact of a young market. It is an equilibrium that regulatory pressure actively produces. None of this is legal advice and I would be a fool to present it as such. What it is, is a reason to treat regulatory silence as a risk input rather than a neutral default. In a jurisdiction-hunting market, the projects that most need clarity about their own status are the ones least likely to publish it, because publishing it would require restricting who can hold the asset — and restricting who can hold the asset would eliminate most of the demand. Anonymous teams are not automatically bad. I have watched two-person pseudonymous teams ship more real infrastructure than hundred-person foundations, and I have personally written checks into that arrangement. The distinction that matters is not identity. It is accountability surface. A named team risks reputation, legal exposure, and career. An anonymous team with a long, consistent, publicly verifiable history risks the same things in a different currency — it risks losing the track record it spent years building. The accountability exists. It is simply denominated in credibility instead of a birth certificate. What does not have accountability surface is a team with no history, no prior work, no consistent identity across venues, and no verifiable claims. In that configuration there is nothing to lose and everything to gain, and the file becomes a description of an opportunity rather than a project. When I consult for a team, the first thing I ask them to do is write down what they would lose if they were publicly wrong. Teams that can answer immediately are almost always building something. Teams that pause are answering the question with the pause. Governance amplifies the same signal. Participation rates in mid-sized decentralized organizations are brutal to look at: single-digit turnout on most votes, with spikes to a third or more only when the proposal alters fee distribution. That pattern is not apathy — it is rational attention allocation, because most proposals do not move value. The consequence, though, is that the population deciding governance is the population with the most concentrated stake, and the population with the most concentrated stake is the population most able to exit before the consequences land. Quorum achieved through a small number of large delegations is not legitimacy. It is arithmetic with a costume on. In the void, governance cannot be evaluated at all, because there is no proposal history to read, no delegation graph to trace, and no participation record to compare against peers. All that can be said is that the absence of a governance record in a project that intends to govern itself is not a gap in the information. It is the governance model. Risks in an information void do not add. They multiply. That is the single most important structural point in this entire analysis, and it is the one that intuition gets wrong. When you evaluate a protocol with partial information, you are pricing a set of known risks against a set of known mitigations. Audit findings you can trade off against a timelock. Concentrated token distribution you can trade off against a long vesting schedule. Low liquidity you can trade off against a hard supply cap. The risks are individual, and each carries an independent probability, and the joint probability of failure is the product of individual survival probabilities — but with enough mitigations in place, that product can still be comfortably small. In a void, there are no mitigations, because a mitigation is a statement, and statements require disclosure. So every failure mode remains live at its unmitigated probability, and the joint survival probability collapses to the product of many small numbers. If you have six unmitigated failure modes each with a five percent chance of materializing in a given year, survival across all six is roughly seventy-four percent. Extend that to twelve modes and it falls to fifty-four percent. Add the modes you cannot enumerate because you cannot see them, and the number approaches the probability that the thing works by accident. This is why the correct output of a due-diligence process applied to a void is not a low score. It is a refusal. I have had this argument with colleagues who insist that refusing to analyze is a failure of analytical nerve. I think the opposite. A framework that cannot return refusal is not a framework. It is a machine for converting absence into confidence, which is exactly what this industry has been selling for a decade. So my recommendation, in the specific case of the file on my desk, was not reduce exposure. It was decline. Not because I had found something bad. Because I had found nothing at all, and the absence of evidence in a domain where evidence is trivially producible is itself the strongest evidence available. There are thousands of protocols with public audits, published schedules, live repositories, and verifiable retention curves. Allocating any marginal capital to one that publishes none of these is not a bet on the protocol. It is a bet that the void will be filled by someone else's money before your own exit. Risks do not add. They multiply. And when you cannot see the factors, you cannot compute the product — you can only observe that it is not on your side. The tale is the last thing to go, and the most dangerous thing to trust. In a market with information, narrative has to compete against disclosure — a story that contradicts the data can be dismissed by anyone willing to read the data. In a market without information, narrative has no competitor. Any story fits, because nothing contradicts it, and a story that nothing can contradict is not an explanation. It is a placeholder. This gives voids a peculiar narrative profile that I have grown to recognize at a glance. They attract maximalist framing, because extreme claims are the only claims that cannot be checked against a spec. They attract institutional language with no institutional behavior, because a treasury strategy cannot be contradicted if it is never executed. They attract community energy that is loudest in the first three months and structurally unable to sustain itself into the fourth, because community in the absence of a product is a marketing expenditure, and marketing expenditures scale with revenue, and there is no revenue. I have watched this pattern long enough to date it. Narrative half-life in a bear market is shorter than most people assume — months, not years, for stories without a delivery mechanism. The stories that survive are the ones attached to something that can be measured, because measurability is what allows a story to be updated rather than merely repeated. A rollup that publishes its blob consumption can have a credible story about scaling. A protocol that publishes nothing can only have a story about potential, and potential has a terminal velocity. Here is the contrarian observation. The most dangerous narrative is not the aggressive one. It is the one that fits anything. Vague stories are not weak stories. They are maximally compatible stories, and compatibility is what allows a community to keep believing after the first three delivery dates pass, because a claim with no edges cannot be contradicted by reality. A sharp claim dies on contact with the world. A shapeless claim survives indefinitely on nothing. Emptiness does not stay local. That is the last thing the void hides, and in some ways the most consequential, because the transmission path runs through infrastructure that people consider safe. Consider what an empty protocol borrows. It borrows the credibility of the chain it is deployed on, the legitimacy of the exchange that lists it, the perceived safety of the wallet that holds it, and the professionalism of the analytics platform that covers it. None of those borrowed assets were ever pledged. They were absorbed, silently, from the surrounding stack. And because the absorption is silent, the reputational cost does not appear anywhere until the moment of default, at which point it distributes across the whole chain of custody. This is not a speculative mechanism. It is visible in the fee markets that underwrite the base layers, and it is visible in the cost structure of the rollup ecosystem that sits on top of them. The blob space introduced with Dencun was priced as if it were abundant, and abundance was always temporary. Blob capacity per block is bounded by protocol parameters, and the demand for that capacity has been compounding since the upgrade shipped — data availability requirements from rollups, from bridges, from proving systems, all drawing on the same fixed pool. When demand is below the target, the blob fee stays at its minimum. When demand pushes past the target, the fee market engages, and it engages in a way that is far more violent than the gas market people are used to, because the adjustment is exponential rather than linear. The period where rollups could advertise sub-cent fees was not a design achievement. It was an under-priced resource being consumed by early movers. That window is closing, and when it closes, the cost of posting data will be repriced over a very short interval, and the rollups will pass that cost to the users who were told that low fees were permanent. I have written this before and I will keep writing it: the thing that makes a rollup cheap is not the rollup. It is the base layer subsidy embedded in the blob fee mechanism, and subsidies end. When they do, the users who migrated on the promise of cost will discover that they migrated on the promise of timing. Bitcoin's data-inscription experiments rhyme with this in an even sharper way. Block space on Bitcoin is roughly one to four megabytes of weight, contested by monetary transactions that people genuinely need to settle. Loading arbitrary data into that space through witness discounts does not create a new use case for the chain — it creates a bidding war in which the new use case wins by paying more, and the monetary users lose by paying more. The analogy I keep returning to is hauling freight in a vehicle designed for something else entirely. The cargo does not arrive well. The vehicle does not survive intact. And the person who owned the car did not agree to the arrangement. The transmission insight generalizes beyond any single chain. An empty protocol does not remain an isolated empty protocol. It borrows trust from infrastructure that has earned it, defaults on that trust, and returns the reputational damage to the infrastructure, which is precisely the infrastructure that the bear market needs to remain credible. The list of things that survive a cycle is short, and every empty file on the list makes it shorter. Now let me say the thing that unsettles me most, because I did not arrive at it easily and I do not fully enjoy holding it. The void is the most honest disclosure in this industry. Everything else we read has been optimized. Whitepapers are optimized for the reader they need to convince. Audits are optimized for the scope they need to cover. Token tables are optimized for the comparison they need to survive. Roadmaps are optimized for the funding round they need to close. I have spent twenty-two years reading documents that were written to be read a certain way, and I have developed a permanent discount rate on all of them. The discount rate is not cynicism. It is calibration. I have watched enough highly optimized documents turn out to be beautiful architecture around nothing. This has a consequence that nobody in my position likes to state plainly. Our analytical frameworks are, in significant part, theater. We built machines that demand that nine dimensions of cells be filled so that we can feel that diligence has happened. FTX filled every cell. Terra filled every cell. The scoring models that rated them highly were not malfunctioning. They were doing exactly what they were designed to do — converting the presence of documentation into the appearance of safety, because presence is what the models could measure. The file on my desk returned no score, and in doing so it refused the entire performance. In a strange way I trust it more than the forty-page documents, because it makes no claim that requires me to suspend disbelief. It simply declines. Which brings me to the harder admission. The problem is not that some assets are empty. The problem is that our most respected assets are only marginally less so, in the sense that matters, because depth is not a proxy for truth. A hundred-page governance document that no one has ever falsified is not more informative than a blank page. It is just more expensive to produce, and expense is what we have been mistaking for evidence. So the correction I want to propose is not a return to completeness. It is a move toward falsifiability. The question that should govern allocation in the next cycle is not how much a team has published. It is what would prove them wrong. A published unlock schedule can be proven wrong against a block explorer. A published fee distribution can be proven wrong against a treasury address. A published audit against a commit hash can be proven wrong by finding the finding that was omitted. Every one of those claims carries the possibility of its own refutation, and that possibility is the only thing that has ever made a document worth reading. Code over hype has never been a slogan about code being good. It is a claim about which of the two can be wrong. Code can be wrong. It runs, it reverts, it gets exploited, and all three outcomes are on the record within minutes. Hype cannot be wrong, which is exactly why it has been so widely preferred. If you take one practical thing from this audit, take the refusal. Build a due-diligence process whose primary output is a decision to decline, and give that output the same dignity as a buy signal. The process will look unproductive. It will feel like an admission that you could not figure something out. That feeling is correct, and the feeling is also the entire value of the exercise, because in a bear market your only durable edge is the set of things you never touched. The emptiness in that file is not an anomaly. It is the growing edge of the market — the frontier where the cost of pretending has finally exceeded the cost of not existing. What I want to know, and what I will be watching through this entire cycle, is whether that frontier becomes a place where careful people simply stop looking, or whether it becomes the standard against which every surviving project is eventually measured. Build anyway, but build so that the falsification is possible. Because the only document worth trusting is one that has told you how to catch it lying. And if a project cannot tell you how to catch it lying, the question is no longer whether it is telling the truth. The question is why you are still reading.