The Empty Input Problem: Why Crypto Research Dies Without Information Points

We didn’t just hunt alpha; we rewired the game. But every rewire starts with a wire. The most dangerous document in crypto is not a hacked bridge. It is a research report with an empty data field. Earlier this week, I found an internal memo that made me stop scrolling. It was titled “Second Stage Deep Analysis: Unable to Execute — Input Data Key Missing.” The opening lines were a confession: the first-stage output had failed to provide a title, a source, an article type, a domain tag, a core viewpoint, or — most catastrophic of all — an information point list. The memo asked a single question: how do we analyze what we cannot see? In a bull market, that question sounds like an excuse. We are all supposed to be speed-reading whitepapers, calculating APYs, and tweeting about alpha. But the more I thought about the memo, the more I realized it was the most honest piece of research I had read in months. It refused to hallucinate. It refused to write a conclusion that was not anchored to a fact. That is rare in crypto, and it is becoming rarer every time the market pumps. What exactly is an information point? It is the smallest unit of analysable reality in this industry. “The protocol’s multisig requires 3-of-5 signatures.” “The TVL dropped from $2.1 billion to $780 million in one week.” “The GitHub repository has 14 open issues and two commits since March.” That is it. No chart. No narrative. No hype. Just a fact. Every serious due-diligence exercise I know starts with a list of these atomic facts. The memo’s nine-dimensional framework was built around that list. Without the list, the framework was a skeleton with no organs. From core dev trenches to community heartbeat, I have watched this problem destroy projects. In 2017, I paused an academic career to audit early Solidity contracts for an ambitious project called EtherHouse. I found four re-entrancy vulnerabilities before the infamous DAO hack by reading raw code. The information points were the color of a function call, the order of state updates, the presence of a lock. I did not need a thesis. I needed input. The code was the input. When I later wrote about “code as law,” I was really writing about information points as the only trustworthy legal evidence in a decentralized system. I have been thinking about that memo all week because it contains a table that is both frustrating and beautiful. The first column lists every field a serious analysis should receive: article title, information source, article type, domain tags, core viewpoint, information point list, involved protocol names, time sensitivity, author position. The second column shows the current status — every single one is marked as missing. The third column describes the impact: no analysis anchor, no credibility weight, no genre recognition, no domain hypothesis, no leverage point, no material for any dimension, no entity anchor, no time premium, no narrative bias correction. It is a complete map of the research process turned into a tombstone. Most analysts would have papered over the gaps. They would have written something like “the project shows strong momentum” and shipped it. The memo did not. That is the beginning of integrity. Let me take the nine dimensions one by one and show what happens when the information point list is empty. This is not an academic exercise; it is the difference between a conviction trade and a coin flip. The technical dimension asks: Is this a layer 1, a rollup, an application, or a bridge? What are the security assumptions? How does it compare with competitors? You cannot answer any of those questions if you do not know the protocol’s name. But even the name is not enough; you need the actual implementation. I keep a running example in my head: Uniswap V4’s hooks. The marketing layer called the DEX “programmable Lego.” The information point is that hooks are callbacks executed before and after swaps. The nuance is that every hook interaction expands attack surface and cognitive load. My honest belief is that the complexity spike will scare off 90% of developers. But I can only defend that belief because I have looked at examples, measured how hard they are to compose, and watched the post-launch adoption curve. If you only have the phrase “programmable Lego,” you cannot see the cliff. This is also why I have been publicly bored by the data availability arms race. Dedicated DA layers are an answer to a real problem — data must be available so that fraud proofs can run — but 99% of rollups do not generate enough data to justify a separate chain. That claim is only defensible if you have the bytes: blob usage per rollup, compression ratios, average transaction size. Leave the information points empty and the modular thesis becomes modular content. The tokenomics dimension asks about supply structure, unlock pressure, Ponzi risk, and value capture. Empty input means you cannot know whether an unlock event is a cliff or a continuous drip. You cannot know whether value flows to the token holder or to an off-chain entity. You cannot know whether emissions are paying for growth or paying for the last buyer. In 2022, I spent three months dissecting the Terra collapse. The information point that mattered most was not the price of LUNA or the market cap of UST. It was the absence of an external anchor. UST was printed by burning LUNA. There was no collateral audit, no circuit breaker, no feedback loop that could survive a bank run. A framework that forbids speculation would have flagged this as “unanalyzable” before the collapse. That is precisely the point: the empty field was the conclusion. The market dimension is next. Without a price anchor or a TVL number, you cannot judge whether a token is in the discovery phase or the distribution phase. In a bull market, this is where FOMO replaces judgment. I have seen a 200% pump generate a beautiful research deck that simply ignored the information point that 40% of the liquid supply had been transferred to a single wallet. The message “send the chart first” is not cynicism; it is methodology. The ecosystem dimension is the favorite hiding place for vibes. The memo wants to know developer health, user growth authenticity, and supply-chain dependencies. During DeFi Summer, I forked three automated market makers in a Jakarta co-working space and launched UniBarter, a localized AMM for Indonesian crypto traders. We attracted 500 users in two weeks. The happy information point was “500 users.” The uncomfortable information point was “the same 12 wallets provided 80% of the liquidity, and most of the volume came from yield hunters chasing a Telegram signal.” The first number made for a great Twitter thread; the second number predicted the exodus. Empty input produces the phrase “a thriving ecosystem” with no evidence. I have learned to mistrust that phrase. The regulatory dimension is a minefield. The Howey test has four elements: investment of money, common enterprise, expectation of profit, and efforts of others. Each element is an information point. If you don’t know whether the token was sold through a public presale, you can’t assess the first element. If you don’t know who controls the treasury, you can’t assess the second. If you don’t know whether the whitepaper promised returns, you can’t assess the third. If you don’t know whether the team operates a centralized sequencer, you can’t assess the fourth. The cleanest conclusion from empty inputs is: this token is too risky to touch. The team and governance dimension looks at backgrounds, investor quality, and decentralization. I have learned to ignore the LinkedIn photos and ask for the multisig threshold. A project can have a Nobel laureate on the advisory board and a 2-of-2 Gnosis Safe controlled by two founders. That is an information point. If you don’t have it, you are evaluating a press release, not a protocol. The risk matrix dimension is the most abused. Most risk matrices are generic: “smart contract risk: medium,” “market risk: high.” That is not analysis; it is desktop publishing. The memo’s framework demanded a risk matrix with rows tied to specific information points. Without those rows, every project looks the same, and the matrix becomes a placebo. The narrative and expectations dimension is personal to me. In 2021, I attended a virtual NFT summit in Bali and met artists turning images into community governance tokens. I co-founded NFTforChange, linking digital collectibles to Indonesian reforestation projects. We minted 1,000 NFTs and raised $50,000 in Ether. The narrative was noble. The information point that changed everything was the floor price curve: from a peak of 0.15 ETH to 0.02 ETH in six weeks. If you had only read the mint announcement, you would have missed the whole story. Narrative analysis requires longitudinal data: what was promised, what was delivered, and what is being said now. The industry-chain transmission dimension asks about upstream and downstream effects. If the protocol is anonymous, you cannot trace the chain. If the protocol is a small DeFi app, you can still ask what happens to its bridge, its oracle, and its aggregator if its TVL halves. The memo had no protocol name at all. It could not even begin. This is why the memo’s priority list matters. It defined P0 fields: information point list, protocol name, article type, and source. It defined P1 fields: article title, publication date, and author position. I agree with every choice. Source determines how much trust you assign to each fact. Article type determines whether you are reading a research report or a sponsored announcement. Author position determines narrative bias. I would add one more P0: the link to the raw material. In crypto, verifiability is oxygen. The memo may have been written out of frustration, but it encoded the same lesson I teach at BlockJakarta: no input, no output. The deeper problem is not the memo. It is the ecosystem that produced it. We reward confidence over evidence. We retweet threads that say “we are so early” and skip the ones that say “our query returned zero rows.” We have built an entire content economy on top of a foundation of missing information points. The bull market does not punish this behavior. It rewards it. Rising prices forgive every analytical sin, and people mistake luck for skill. A memo that refuses to feed the machine is a small rebellion. Take the phrase “institutional adoption.” It is the most common empty input in our industry. Someone will tweet “BlackRock is buying ETH” and generate 10,000 words of analysis. But the information point is “BlackRock filed an S-1 for an Ethereum ETF” — and even that is a filing, not a purchase. The gap between the filing and the analysis is where narratives get built. In a bull market, you can fill that gap with hope. In a bear market, you fill it with panic. Neither is analysis. Time sensitivity is the most forgotten field. A yield farm APY from 2020 is a historical artifact. A TVL number from Tuesday can be useless by Thursday. The memo’s P1 field “publication date” should actually be P0. I would argue time is the first thing to check before any chart. If you do not know when a fact was true, you do not know if it is true now. The memo also listed “author position” as P1, but that too should be higher. A report written by a token team is not a report; it is a pitch. A report written by a short seller is a different kind of pitch. You need that context before you can weigh a single information point. There is a simple test I use for every research report I read. Replace the protocol name with another protocol name. If the article still makes sense, it contains no information points. A real analysis is specific enough that changing the subject breaks the logic. The memo failed that test before it even started, because it had no subject. That is why it was honest enough to stop. Let me give you a more personal example. During my audit of EtherHouse, I found a function that looked like a donation. The information point was the sequence of operations: the function sent Ether first, then updated the balance. That is a reentrancy flaw. A polished audit report could have said “no issues found” by checking only the compiler version and deleting the word “call.” The difference is the information point list. The list catches the flaw because it forces you to write down every state-changing operation in order. The same discipline applies to token analysis. When I see a freshly funded project with a $100 million war chest and all the marketing pages, the first thing I ask for is its information points. If the press release says “institutional-grade technology” but the code repo is private, that is a red flag bigger than any chart. If the tokenomics page says “fair launch” but the top 10 addresses hold 80% of the supply, the phrase means nothing. The information point is the distribution table. Without it, you are not doing research; you are doing wishful thinking. A useful analysis framework is a chain of falsifiable claims. Each claim should map to an information point. If the mapping is broken, the chain is broken. This is why I have, for years, said that the Lightning Network has been half-dead. That statement looks like an opinion. Behind it is a chain: routing failure rates are absurd; channel management is too complex; the user experience is a niche hobby. Each link has a data point. If you do not believe me, run your own payments over public Lightning routes and count the failures. That is the information point. Now the contrarian turn. The memo’s refusal to execute is a feature, not a bug. In an industry where analysts confidently vomit conclusions from empty inputs, intellectual honesty is a competitive edge. The most valuable sentence in the memo is: “This violates my professional standard, which prohibits speculation.” That sentence should be printed on a t-shirt. Too many people think the job is to have an opinion. It isn’t. The job is to know which information points would change your opinion. The memo did not know any, and it had the discipline to say so. But the memo also has a blind spot. It treats empty input as an obstacle when sometimes absence of input is the input. The Lightning Network is my favorite example. For seven years, I have argued that Lightning is half-dead. Routing failure rates are absurd. Channel management is too complex. The whole UX is a niche hobby. However, I only reached that conclusion because I was willing to dig through sparse data: node statistics, failed payment percentages, channel churn. If I had refused to analyze because the data was incomplete, I would have missed the reality. The empty field is not always a block; sometimes it is a door. The same is true for DA layers. They are overhyped because the key data — bytes published per rollup — is not public enough. The memo would say “unable to execute.” A better analyst would say “that silence itself is a finding.” When the market sleeps, the architects wake up. And the first act of architecture is to list what you do not know. Here is the uncomfortable synthesis. The industry does not need more data-hoarding analysts, and it does not need more confident speculators. It needs analysts who can hold two ideas at once: I don’t have the information point to prove this, and the absence of information points is itself information. The memo was right to refuse a hollow analysis. It was also wrong to stop there. The empty input table was the beginning of the research, not the end. The next step should have been to say: I cannot analyze this specific project, but I can analyze the shape of the silence — why is the data missing, who benefits from the absence, and what would change my view. That is the edge. So here is my advice to anyone trying to survive this bull market without becoming a bagholder. Stop asking for the conclusion. Start asking for the information point list. If someone cannot show you the raw atomic facts behind their recommendation — the code diff, the vesting schedule, the wallet distribution, the blob usage — they do not have a thesis. They have a vibe. Education is the new mining rig for the mind, and every mining rig needs ore. The ore is not a headline. It is a single verifiable fact about a protocol’s code, token, or users. The next time someone hands you a 50-page deep research report, open it to page two and look for the empty fields. They are the only part of the document that cannot lie. In a market that runs on hallucination, truth is the edge.