{"course":{"slug":"prompting","title":"Getting Real Work out of a Model","url":"https://addaly.com/learn/prompting","level":"beginner","track":"start"},"lesson":{"slug":"checking-what-is-missing","title":"The hardest check: what was left out","url":"https://addaly.com/learn/prompting/checking-what-is-missing","position":62,"of":72,"minutes":9,"answer":"An omission leaves no mark in the answer, so force it into view: enumerate and count before analysing, require a line for every checklist item including \"not present\".","keyIdea":"An omission leaves no mark in the answer, so force it into view: enumerate and count before analysing, require a line for every checklist item including \"not present\", and ask what was found but left out."},"body":{"format":"markdown","text":"## An absence has no signature\n\nEvery other failure in this module leaves a mark. A fabricated citation can be searched for. A wrong number is out of proportion. A misread clause has a real sentence you can go and read.\n\nAn omission leaves nothing. A summary of four payment terms that covers three reads exactly like a summary that covers four. You cannot see it by looking at the answer, because looking at the answer is not where it is.\n\nThis is the failure most likely to reach a decision, and it gets the least attention.\n\n## Where omissions come from\n\n**Compression.** Summarising is selection, and the model's sense of what matters is not yours. The clause that matters to you may be unremarkable in general.\n\n**Position.** Material in the middle of a long context is used less reliably. A document's middle section is where things quietly go missing.\n\n**A stopping habit.** Asked to list items, models tend to produce a comfortable number (five, seven, ten) and stop. If there are fourteen, you often get ten and a fluent closing sentence.\n\n**Retrieval.** With a long attached file, only some passages may have reached the model at all.\n\n**Your own framing.** Ask \"what are the risks?\" and you get risks. A cost that is not a risk, a dependency, an opportunity: none of them were asked for and none will appear.\n\n## Count before you analyse\n\nThe single most effective technique, and it takes one extra step.\n\n> First: list every clause that mentions payment, by clause number and first five words. Do not summarise or analyse. Count them and give me the total.\n\nNow you have a number. If it says four and you expected more, you have caught the omission before any analysis was built on it. Then run the analysis over the enumerated list.\n\n> Figure: Payment clauses found in a sixty-page contract. Bars (clauses): Actually in the document 14; In a one-pass summary 9; After enumerate, count, then analyse 14. The summary that covered nine read exactly like a summary that covered fourteen. The count is what made the gap visible, before any analysis was built on it.\n\nLocating and reasoning are different jobs. Doing them separately is the theme of half this course, and this is the sharpest instance.\n\n## Ask for the boundary of the answer\n\nThree lines that turn silence into information:\n\n> After the answer, list: anything relevant you found but did not include, and why. Anything I asked about that the document does not address. Anything you were unsure whether to include.\n\nThe third one is the surprising one. It regularly surfaces the item that was on the edge of the selection, which, being an edge case, is disproportionately likely to be the thing you cared about.\n\n## Adversarial checks\n\nAsk a question whose answer you know, or can quickly establish, and see whether the answer holds up:\n\n- \"Does this document say anything about X?\" where you know it does. A false negative tells you retrieval or attention is failing.\n- \"How many sections does this document have?\" against the actual count.\n- \"Quote the last sentence of the document.\" Catches truncation instantly.\n\nFor anything long and important, the last of these takes five seconds and is worth doing every time.\n\n## The structural fix\n\nFor recurring work, a checklist beats vigilance.\n\nIf you review contracts, you have a list of the twelve things you always check. Put the list in the prompt and require a line for each, including \"not present\". Now an omission is a visible row rather than a silence.\n\nThat is the same move as the fail-visibly lesson, applied to coverage instead of values: **make absence occupy space**. An empty row is something you can see. Nothing is not.\n\n## The limit worth stating\n\nNone of this catches an omission of something you did not know to ask about, and no method can. A checklist covers what is on the checklist. Enumeration covers what you thought to enumerate.\n\nWhich is the honest reason that, for anything consequential, somebody who knows the domain reads the document. The tools make that person faster. They do not replace the fact that only a person who knows what should be there can notice that it is not.\n\n## Why this failure is under-reported\n\nNobody writes a complaint about an omission, because nobody knows it happened. The wrong figure gets corrected loudly; the missing clause is discovered eighteen months later, by a lawyer, in a dispute.\n\nSo the anecdotes people share are all about the visible failures, and the impression forms that models mostly invent things. Invention is the noisy failure. Omission is the common one, and if you adopt one habit from this module it should be enumerate-then-analyse, because it is the only one that addresses the failure you will never otherwise hear about."},"check":{"question":"A manager asks for a summary of a 40-page supplier agreement and receives a clear two-page summary. Which single request would most reliably reveal whether anything important was omitted?","options":[{"text":"\"First list every clause heading in order with its number, and give the total count.\"","correct":true},{"text":"\"Are you confident this summary is complete?\"","correct":false,"tempting":"A leading question about completeness, answered by a system that cannot see what it did not include."},{"text":"\"Make the summary longer and more detailed.\"","correct":false,"tempting":"More words about the same selected material; the omitted sections stay omitted."},{"text":"\"Summarise it again and compare the two summaries.\"","correct":false,"tempting":"Two passes with the same selection pressure tend to omit the same material, and agreement would then look reassuring."}]},"prev":{"slug":"a-second-opinion","title":"Asking a second model","url":"https://addaly.com/learn/prompting/a-second-opinion"},"next":{"slug":"how-much-checking","title":"How much checking is enough","url":"https://addaly.com/learn/prompting/how-much-checking"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'The hardest check: what was left out', in Getting Real Work out of a Model. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/prompting/checking-what-is-missing (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"The hardest check: what was left out","courseTitle":"Getting Real Work out of a Model","module":{"index":7,"slug":"checking-the-answer","title":"Checking the answer"},"answer":null,"keyIdea":"An omission leaves no mark in the answer, so force it into view: enumerate and count before analysing, require a line for every checklist item including “not present”, and ask what was found but left out.","hero":{"kind":"figure","figure":{"t":"figure","label":"Figure","figure":{"type":"bars","unit":"clauses","title":"Payment clauses found in a sixty-page contract","series":[{"label":"Actually in the document","value":14},{"label":"In a one-pass summary","value":9},{"label":"After enumerate, count, then analyse","value":14,"highlight":true}],"caption":"The summary that covered nine read exactly like a summary that covered fourteen. The count is what made the gap visible, before any analysis was built on it."},"alt":"Figure: Payment clauses found in a sixty-page contract. Bars (clauses): Actually in the document 14; In a one-pass summary 9; After enumerate, count, then analyse 14. The summary that covered nine read exactly like a summary that covered fourteen. The count is what made the gap visible, before any analysis was built on it."}},"blocks":[{"t":"h2","id":"an-absence-has-no-signature","runs":[{"text":"An absence has no signature"}]},{"t":"p","runs":[{"text":"Every other failure in this module leaves a mark. A fabricated citation can be searched for. A wrong number is out of proportion. A misread clause has a real sentence you can go and read."}]},{"t":"p","runs":[{"text":"An omission leaves nothing. A summary of four payment terms that covers three reads exactly like a summary that covers four. You cannot see it by looking at the answer, because looking at the answer is not where it is."}]},{"t":"p","runs":[{"text":"This is the failure most likely to reach a decision, and it gets the least attention."}]},{"t":"h2","id":"where-omissions-come-from","runs":[{"text":"Where omissions come from"}]},{"t":"p","runs":[{"text":"Compression.","b":true},{"text":" Summarising is selection, and the model’s sense of what matters is not yours. The clause that matters to you may be unremarkable in general."}]},{"t":"p","runs":[{"text":"Position.","b":true},{"text":" Material in the middle of a long context is used less reliably. A document’s middle section is where things quietly go missing."}]},{"t":"p","runs":[{"text":"A stopping habit.","b":true},{"text":" Asked to list items, models tend to produce a comfortable number (five, seven, ten) and stop. If there are fourteen, you often get ten and a fluent closing sentence."}]},{"t":"p","runs":[{"text":"Retrieval.","b":true},{"text":" With a long attached file, only some passages may have reached the model at all."}]},{"t":"p","runs":[{"text":"Your own framing.","b":true},{"text":" Ask “what are the risks?” and you get risks. A cost that is not a risk, a dependency, an opportunity: none of them were asked for and none will appear."}]},{"t":"h2","id":"count-before-you-analyse","runs":[{"text":"Count before you analyse"}]},{"t":"p","runs":[{"text":"The single most effective technique, and it takes one extra step."}]},{"t":"quote","runs":[{"text":"First: list every clause that mentions payment, by clause number and first five words. Do not summarise or analyse. Count them and give me the total."}]},{"t":"p","runs":[{"text":"Now you have a number. If it says four and you expected more, you have caught the omission before any analysis was built on it. Then run the analysis over the enumerated list."}]},{"t":"p","runs":[{"text":"Locating and reasoning are different jobs. Doing them separately is the theme of half this course, and this is the sharpest instance."}]},{"t":"h2","id":"ask-for-the-boundary-of-the-answer","runs":[{"text":"Ask for the boundary of the answer"}]},{"t":"p","runs":[{"text":"Three lines that turn silence into information:"}]},{"t":"quote","runs":[{"text":"After the answer, list: anything relevant you found but did not include, and why. Anything I asked about that the document does not address. Anything you were unsure whether to include."}]},{"t":"p","runs":[{"text":"The third one is the surprising one. It regularly surfaces the item that was on the edge of the selection, which, being an edge case, is disproportionately likely to be the thing you cared about."}]},{"t":"h2","id":"adversarial-checks","runs":[{"text":"Adversarial checks"}]},{"t":"p","runs":[{"text":"Ask a question whose answer you know, or can quickly establish, and see whether the answer holds up:"}]},{"t":"ul","items":[[{"text":"“Does this document say anything about X?” where you know it does. A false negative tells you retrieval or attention is failing."}],[{"text":"“How many sections does this document have?” against the actual count."}],[{"text":"“Quote the last sentence of the document.” Catches truncation instantly."}]]},{"t":"p","runs":[{"text":"For anything long and important, the last of these takes five seconds and is worth doing every time."}]},{"t":"h2","id":"the-structural-fix","runs":[{"text":"The structural fix"}]},{"t":"p","runs":[{"text":"For recurring work, a checklist beats vigilance."}]},{"t":"p","runs":[{"text":"If you review contracts, you have a list of the twelve things you always check. Put the list in the prompt and require a line for each, including “not present”. Now an omission is a visible row rather than a silence."}]},{"t":"p","runs":[{"text":"That is the same move as the fail-visibly lesson, applied to coverage instead of values: "},{"text":"make absence occupy space","b":true},{"text":". An empty row is something you can see. Nothing is not."}]},{"t":"h2","id":"the-limit-worth-stating","runs":[{"text":"The limit worth stating"}],"role":"limit"},{"t":"p","runs":[{"text":"None of this catches an omission of something you did not know to ask about, and no method can. A checklist covers what is on the checklist. Enumeration covers what you thought to enumerate."}]},{"t":"p","runs":[{"text":"Which is the honest reason that, for anything consequential, somebody who knows the domain reads the document. The tools make that person faster. They do not replace the fact that only a person who knows what should be there can notice that it is not."}]},{"t":"h2","id":"why-this-failure-is-under-reported","runs":[{"text":"Why this failure is under-reported"}]},{"t":"p","runs":[{"text":"Nobody writes a complaint about an omission, because nobody knows it happened. The wrong figure gets corrected loudly; the missing clause is discovered eighteen months later, by a lawyer, in a dispute."}]},{"t":"p","runs":[{"text":"So the anecdotes people share are all about the visible failures, and the impression forms that models mostly invent things. Invention is the noisy failure. Omission is the common one, and if you adopt one habit from this module it should be enumerate-then-analyse, because it is the only one that addresses the failure you will never otherwise hear about."}]}],"check":{"question":[{"text":"A manager asks for a summary of a 40-page supplier agreement and receives a clear two-page summary. Which single request would most reliably reveal whether anything important was omitted?"}],"options":[{"text":[{"text":"“First list every clause heading in order with its number, and give the total count.”"}],"correct":true,"tempting":null},{"text":[{"text":"“Are you confident this summary is complete?”"}],"correct":false,"tempting":[{"text":"A leading question about completeness, answered by a system that cannot see what it did not include."}]},{"text":[{"text":"“Make the summary longer and more detailed.”"}],"correct":false,"tempting":[{"text":"More words about the same selected material; the omitted sections stay omitted."}]},{"text":[{"text":"“Summarise it again and compare the two summaries.”"}],"correct":false,"tempting":[{"text":"Two passes with the same selection pressure tend to omit the same material, and agreement would then look reassuring."}]}],"order":[3,0,1,2],"why":null},"prev":{"slug":"a-second-opinion","title":"Asking a second model","minutes":8},"next":{"slug":"how-much-checking","title":"How much checking is enough","minutes":8}}}