Every hospital has a second floor plan, and it is never posted by the entrance. The first plan shows the wards, the lifts, the cafeteria and the way to radiology. The second shows where things are known: which system holds the laboratory results, where records from another hospital go when they arrive, which binder at the nurses’ station holds the thing nobody has typed in, and which person on which shift remembers what was said to the family on Tuesday. Visitors need only the first plan. A new doctor needs both, and the second is the harder to learn, because nobody has drawn it.
The first skill of training is medicine itself: what a finding means, what follows from it, what ought to be done. The second is knowing where the finding lives, in what condition, and who can vouch for it. The two are learned together, in the same building, and they are easily mistaken for each other in a way that can be quietly demoralising. A new intern who cannot find a result in his first week looks, to himself and sometimes to others, like a doctor who does not know enough. Often he knows plenty. What he lacks is the address.
My own training crossed an awkward moment in the history of that second plan. In Amritsar, where I studied medicine until 2002, the hospital record was overwhelmingly paper, and its geography was physical: a folder with a thickness, sections that could be thumbed through, entries in recognisable hands. At the University of Arkansas for Medical Sciences, where I was an intern in 2007–2008, electronic records were already an ordinary part of clinical work, but they were not one system. A laboratory application knew one part of the patient and a radiology system another; some orders and some documentation were electronic; paper survived, and so did the fax machine. I have described that patchwork in a longer essay and will not reconstruct its screens here, partly because I could not do so honestly at this distance and partly because the question it taught is larger than any of them. Where does this fact live? is a question about location. Asked well, it becomes a question about what a fact is.
Begin with the simplest-sounding report a new doctor makes: I can’t find it. It sounds like one observation. It may be any of four. The thing may be absent: never made, never written, never done. It may be inaccessible: it exists, but behind a permission, a password, another institution’s wall. It may be misfiled: present, but under a heading or a date or a spelling where no reasonable person would look. Or it may simply be unfamiliar to the seeker: present, correctly filed, in a place he has not yet learned. These states call for different remedies, the first for a decision, the second for a request, the third for a correction, the fourth for a lesson, and a good deal of early frustration comes from treating them as one.
Two Kinds of Knowledge #
In 1775 Samuel Johnson called with James Boswell and Sir Joshua Reynolds on a Mr. Cambridge, and the moment they had made their bow to him in his library, Boswell records, Johnson “ran eagerly to one side of the room, intent on poring over the backs of the books.” Reynolds observed, aside, that Johnson ran to the books as he himself ran to pictures, with the advantage on the painter’s side, since he could see much more of the pictures than Johnson could of the books. When their host politely confessed to the same habit, and wondered why anyone should want to look at the backs of books, Johnson wheeled round and answered at once. “Sir, the reason is very plain. Knowledge is of two kinds. We know a subject ourselves, or we know where we can find information upon it.” The first thing in any inquiry, he went on, is to know what books have treated of the subject, and that is why one looks at catalogues, and at the backs of books in libraries.
It is the best short statement I know of this essay’s subject, and it came from a man who had earned the right to make it. Johnson’s Dictionary, published twenty years earlier, in 1755, defined some forty-two thousand words and illustrated their senses with about 114,000 quotations from some five hundred authors, passages he had marked in books for clerks to copy out. It was a work of both kinds of knowledge at once: of knowing words himself, and of knowing precisely where their uses could be found and saying so, author by author. Every definition arrived with its evidence attached, and the evidence arrived with its address.
Medicine needs both of Johnson’s kinds, and it needs a third that his sentence leaves out. A doctor must know a great deal himself, because there are moments when nothing can be looked up. He must know where to find a great deal more. And he must know, once he has found it, what condition it is in: whether this is the current edition, whether the page has since been corrected, whether the catalogue’s description matches what is inside the covers. Reynolds’s joke points at the gap. From across the room Johnson could read the spines, which told him what was there and where it stood; he could not read the pages. A spine is an address. It is not the text.
The complete electronic record makes that distinction impossible to ignore. It keeps everything, at the same volume and with the same confidence: every vital sign at every hour, every templated sentence, every repeated list, every result back to the first, all of it retained and present and very little of it ranked. Nobody can carry such a record in his head, and nobody should try. Johnson’s second kind of knowledge is what releases a doctor from the impossible task: he need not hold everything if he knows where everything lives. But the release is conditional. It depends on the addresses being reliable, and on the summaries a busy doctor actually reads keeping their spines, each compressed line pointing to the volume it came from, so that the next reader can take the volume down if the line turns out to matter. A summary without addresses does not abbreviate the record. It quietly replaces it.
Shelved by Arrival #
A century after Johnson’s visit, the shelves themselves were reformed. When Melvil Dewey’s system appeared, most American libraries gave each book a fixed position on the shelf, assigned by its height and the date it was acquired. A reader who wanted a volume had to know its place or find someone who did. Dewey devised the alternative while employed in the library of Amherst College, and in 1876 he published it as a forty-four-page pamphlet with the patient title A Classification and Subject Index for Cataloguing and Arranging the Books and Pamphlets of a Library. It rested on what came to be called relative location. A book’s address would no longer be a spot in the building but a number expressing its subject, so that books stood beside their nearest neighbours in meaning wherever that subject happened to be shelved, and a whole collection could be moved without one address becoming false. Dewey turned the address of a book into a statement about what was in it.
Seventy years later Vannevar Bush complained that the reform had not gone far enough. In “As We May Think”, published in The Atlantic Monthly in July 1945, he objected to every system that filed information alphabetically or numerically and found it, when it was found at all, by tracing it down from subclass to subclass. An item filed that way, he observed, “can be in only one place, unless duplicates are used.” The mind, Bush argued, works by association, leaping from one item to the next along trails, and he imagined a desk-sized machine, the memex, that would let a reader build and keep such trails through a library on microfilm, annotate them, and hand a copy to a colleague. Among his examples was a physician, puzzled by a patient’s reactions, following a trail laid down while studying an earlier, similar case, running through analogous histories with side references to the anatomy.
Hospital information, for all its digitisation, is still largely shelved by arrival. A fact tends to live where it was produced. The laboratory’s result lives in the laboratory’s system because the laboratory made it; the specialist’s opinion lives in the specialist’s note; the summary from another hospital lives wherever that hospital chose to send it. Each location is sensible from the point of view of the department that created the fact. From the point of view of the patient’s problem, the facts are scattered across the building like books shelved by date of purchase.
Bush’s qualifying clause has turned out to be the important one. In a modern record a fact can be in many places, because duplicates are used constantly: copied forward, summarised, carried into a list, quoted in a letter. The question of where a fact lives has quietly become a question of which of its copies is the original and whether the copies still agree. That second question belongs to another essay. Here it is enough to notice that the address of a fact is no longer single, and that a new doctor learning the building is learning a map on which the same fact may appear in several places at once, not always in the same condition.
Coordinates #
A fact in a medical record has more than one address, and learning to read them is most of the second skill. Imagine a laboratory value that appears on a screen in the middle of a busy day. It is a single number in a single place. To know what it can bear, a doctor needs at least five further coordinates.
The first is its source: which laboratory, which instrument, which kind of specimen, or, if it was typed in by hand, whose hand and copied from what. The second is its place of storage, which tells him what else is likely to be stored beside it and what is likely to be missing. The third is time, and time comes in more than one kind: when the sample was taken, when it was resulted, when it was entered, when it was last looked at. A value can be fresh in the record and old in the body. The fourth is status. The fifth is ownership: who ordered it, who is expected to act on it, and who will be asked about it if nobody does.
Status is the coordinate most easily ignored, because a number looks finished whatever its condition. The technical standard that many health systems now use to exchange records, HL7’s FHIR, gives a diagnostic report ten possible statuses in its fourth release: registered, partial, preliminary, final, amended, corrected, appended, cancelled, entered in error, and unknown. Read slowly, the list is a small treatise on the life of a fact. Registered means the existence of the report is known but nothing is yet available. Preliminary means some verified early results exist and others do not. Final means complete and verified by an authorised person. Amended, corrected and appended name three different things that can happen to a fact after it has been declared final. And the definition of entered in error contains the most candid sentence I know in health-information engineering: the report has been withdrawn, and “this electronic record should never have existed”, though, the definition adds, real-world decisions may have been based on it.
I quote a data standard in an essay because it says plainly what clinical culture often leaves implicit. A fact in a record is not a stone. It is an event in progress, and its address includes where in that progress it stands. A standard can express those states; whether a particular system displays them, and whether a tired reader notices them, is another matter, and no standard can guarantee either. Finding the page is the first step. Knowing what claim the page can support is the work.
Ownership is the coordinate that breaks most quietly in a teaching hospital, because trainees move. A first-year doctor commonly changes services every few weeks, and a test he ordered on his last day on one ward may return when he is already on another, routed to a name that is no longer attached to that patient’s care. The result exists, in the right system, under the right patient, with the right status, and its fifth coordinate points at someone who has gone. Nothing is missing except the person whose job it was to look. That kind of gap cannot be found by searching harder, because nothing is lost; it can only be prevented by making ownership explicit at the moment it changes hands.
Browsing, Querying, Asking #
There are three ordinary ways to look for a fact, and each has its use and its characteristic failure.
The first is browsing. A paper chart was mostly browsed; the eye moved through it and, in moving, passed things it was not looking for: a consultant’s note two pages before the one wanted, a remark in a margin, a medication chart with a crossing-out that told a small story of its own. Browsing shows the neighbours of what you came for, which is the virtue Dewey built into his shelves. Its failures are fatigue and luck. It is slow, it misses whatever lies beyond the point where one stopped, and it was never as thorough as nostalgia for paper suggests. Paper charts lost pages, held only what had been written in that building, and were often in someone else’s hands at the moment they were needed.
The second is querying. An electronic system is largely asked: one chooses a screen, a category, a date range, a search term, and it returns what was requested quickly and completely. Its virtues are speed, reach and durability; it can be consulted from another floor in the middle of the night. Its failure mirrors browsing’s. It rewards those who already know what to ask and where to ask it. The fact one did not think to request does not drift past the eye. It waits, perfectly preserved, in a place one had no reason to visit. Modern systems soften this with searches across the whole record and displays of related items, and these help; but they still return answers to questions, and a beginner’s difficulty is often that he does not yet know the question.
The third is asking a person, and it is the oldest. It is fast when the right person is near, it brings interpretation along with location, and it can answer the question behind the question: not only where the result is, but why it was ordered and whether anyone has already acted on it. Its failure is that people are absent, busy, tired and occasionally wrong, and that what they say leaves no trace unless someone writes it down. Bush’s memex was, in one light, an attempt to give the third way the durability of the second: a trail is a record of how one item led to another, a piece of someone’s reasoning made portable. A good handoff is the nearest thing a hospital has to a memex trail. It does not merely list facts; it says which facts led to which, and where to look if the trail goes cold.
A careful new doctor learns all three, and learns which to use first: querying for the thing he knows he needs, asking for the thing he does not know he needs, and browsing, when time allows, for the thing nobody thought to mention.
Kinds of Absence #
The hardest search is the one that ends in nothing. Consider a hypothetical colleague, early on a busy day, asking whether an investigation planned some weeks ago came back normal. The searcher looks and finds no result. I can’t find it is a true report and a nearly useless one, because it can conceal at least six different situations, each pointing to a different next question.
Perhaps the test was discussed but never ordered; then the question is whether it is still wanted. Perhaps it was ordered and never performed; then the question is why, and whether the patient knows. Perhaps it was performed somewhere else, at another hospital or an outside laboratory, and the result lives in that institution’s records; then the question is how to obtain it. Perhaps it was performed here and is still pending, or was resulted under a heading, a date or a variant of the patient’s name that a reasonable search would not reach. Perhaps it exists but is restricted from the searcher’s view. Or perhaps it was cancelled, or entered in error and withdrawn. Until these possibilities are separated, the next step is guesswork about the archive and not reasoning about the patient.
The honest report is often smaller than the question. I can see that it was requested; I cannot yet establish that it happened. The sentence is useful because it preserves a boundary that a hurried summary easily erases, the boundary between an intention and an event. A plan to test is not a test, an order is not a result, and the absence of a result is not a normal result. Each of those confusions has a familiar shape in clinical life, and each begins with a search that ended in silence and was reported as an answer.
The same standard that lists a report’s statuses keeps a vocabulary for missing values, and it rewards reading for the same reason. A value may be absent because it is temporarily unknown, because it was never asked, because it was asked and the source did not know, because it was asked and the source declined to say, because it is masked for privacy, or because the test was not performed. These are not technicalities. They mark the difference between a gap that will close on its own and one that needs someone to act, between a silence that carries information and a silence that is merely silent. A record that could say which kind of absence it holds would be more useful than one that fills the space with something confident. A well-marked gap is itself a kind of fact.
Punjabi, my first language, has a word for this condition that I find hard to improve on. ਪਤਾ (patā) means an address, a whereabouts, a trace, and also knowledge: mainūṅ patā hai is the everyday way of saying I know, and comes out literally as “there is knowledge to me”. I have written elsewhere about what that double meaning does for the sentence I don’t know yet. Here the useful word is its negative. Something missing is ਲਾਪਤਾ (lāpatā), without an address. The word does not say that the thing does not exist. It says that nobody can presently say where it is, which is precisely the condition of the result one cannot find, and a better description of it than either normal or not done. I am aware that in Punjab’s recent history the same word has been applied to people as well as papers, and I do not borrow that weight for a hospital essay. But it keeps me from treating any absence in a record as merely clerical.
The Human Index #
The deepest layer of a hospital’s second floor plan is not stored in any system. It is held in people. In 1985 the psychologist Daniel Wegner proposed the idea of transactive memory: that groups, first studied in couples and families and later in working teams, store knowledge across their members, each of whom carries a share of it together with a sense of who holds the rest. A team behaves less like one large memory than like several smaller ones sharing a directory.
A ward is such a team, and its directory is mostly unwritten. The night nurse knows how the patient actually slept. The pharmacist knows why a dose was changed. The unit clerk knows where faxed records end up. The senior resident knows which consultant prefers a telephone call to a page. A new doctor can be well read and still be, in this one respect, a stranger without the directory. His ignorance is real, but it is often ignorance of who and where, and more reading will not cure it. Asking will, together with remembering the answers and noticing whose answers turn out to be right.
Rural Punjab has long had a human index of this kind for the most consequential records in village life. The patwari, the village-level revenue official, keeps the land records, and the practical question of who holds which field, under what entry, has often been answered as much by the person who knows his way around the registers as by the registers themselves. The arrangement has familiar virtues and familiar hazards. Knowledge of where a record lives is a form of power, and a village’s dependence on one person’s familiarity with a register can be exploited as easily as it can be served. A hospital’s human index rarely has that kind of leverage, but it shares the position: the indispensable route to a fact is often a person whose labour nobody has formally counted.
That labour has costs that are easy to miss. The nurse or clerk who becomes a ward’s unofficial search engine is interrupted constantly, credited rarely, and missed only when absent. Every turnover of staff takes part of the directory with it, and a unit that has lost its experienced people may keep every one of its records and still not know where anything is. An institution that wants its knowledge to outlast the people who hold it has to write the directory down: not every fact, which the record already holds, but the map of where facts go and who can explain them.
There is an opposite risk, and it grows with seniority. In 2011 a much-discussed paper in Science by Betsy Sparrow, Jenny Liu and Daniel Wegner reported that people who expected to be able to look something up later remembered where it was kept better than they remembered the thing itself. Attempts to repeat parts of that work have produced mixed results, which is worth knowing before building a philosophy on it. Still, the worry it named is familiar in medicine. A physician who knows exactly where every fact is stored and carries none of them has become a very efficient librarian of his own ignorance. Clinical judgement needs the address and the meaning together, in one head, at the moment of decision.
The Newcomer’s Map #
For a doctor new to a hospital, and especially for one new to a country’s hospitals, failures of wayfinding are easily read as failures of knowledge. The misreading is understandable. From across a room, a doctor who cannot find the result and a doctor who would not know what to do with it look much the same. But they need different help, and the wrong help can make matters worse. Explaining the medicine to someone who needed the address is mildly insulting; supplying the address to someone who needed the medicine is dangerous.
The remedy is to teach the second floor plan explicitly, as part of orientation and not as folklore: where results from each kind of test appear, where outside records go, how a question reaches the person who can answer it at three in the morning, which information lives only in people, and who those people are. Hospitals that do this well tend to do it modestly, through a few written pages and a senior resident’s willingness to walk a newcomer through his first week. It costs little, and it changes the meaning of where does this live? from a confession into a professional request.
A question about where something lives can itself be asked well or badly, and the difference is worth teaching. Where are the results? invites a shrug. I’ve looked through the last month of laboratory results and found nothing from the outside clinic; where do their reports usually land? invites an answer, because it shows the map the asker already has and marks exactly where it runs out. It protects the listener’s time and the asker’s standing at once: it is plainly the question of someone who has done the work and reached the edge of local knowledge, not of someone who has not begun. Its form is the form of the honest report of an absence. It says what was searched, how, and what was not found, and asks only for the missing coordinate. A beginner who learns to ask that way learns in the same motion how to report a gap to a senior colleague, which is among the more consequential sentences in clinical life.
It would be too easy, though, to make orientation the explanation for every difficulty. Some information is hard to find because the system genuinely hides it: because two systems do not exchange what they hold, because a result is filed under a heading nobody would try, because a document arrived as an image that no search can read. A newcomer who cannot find such a thing is not poorly oriented. He has found a fault in the building. Part of an institution’s humility is a readiness to hear a newcomer’s question as possible evidence about itself.
So I would like the question to lose its embarrassment. Where does this live? sounds elementary even when the thing sought is complicated, and pride often prefers to pretend it knows the route. Pretending does not shorten the route. Asked plainly, the question makes the missing link visible, and the people who answer it best are often not the most senior people in the building.
From Retrieval to Provenance #
The clinical record today is far more integrated than it was twenty years ago, and software can search it faster than any intern could. Increasingly, software can also summarise it: a paragraph that gathers a history from a hundred notes, a sentence that answers a question in plain prose. That speed changes the second skill without retiring it. When retrieval was slow, finding a fact was the achievement, and the fact usually arrived with its address attached, because one had gone to fetch it. When retrieval is instant, the fact can arrive without having been anywhere, and the address becomes the thing one has to ask for.
Take a single sentence in a hypothetical summary: No prior history of kidney disease. It is fluent, plausible and quite possibly true. To know what it can bear, the reader has to follow it back. Is it drawn from a note in which someone wrote those words after asking the patient? From the absence of any such diagnosis in a problem list, which is a different and weaker thing? From a normal laboratory value three years old? Or from silence, a record in which nothing to the contrary happened to be found? Each is a different claim, and only the first is what the sentence appears to make. A summary that can show its trail, sentence by sentence, lets the reader draw that distinction in seconds. A summary that cannot is asking to be believed on its fluency.
Johnson’s Dictionary offers a better pattern for such summaries than much of what has been built so far. Its definitions were brief, and could afford to be, because each sense came with a quotation and each quotation with a name: a reader who doubted a definition could see the evidence and, if he cared to, go to the book. A clinical summary built on that pattern would be longer by a few words a line and shorter by many minutes of doubt. Every compressed claim would carry the note, the date and the author it came from, and a reader in a hurry could rely on it precisely because a reader with a question could check it.
The standards for exchanging records have a place for this as well. FHIR’s provenance resource exists to record the activity that created or changed a piece of information and the people and systems involved, so that a reader can judge its reliability or know where to go to investigate its origins. A field for provenance is not provenance, and whether a given system fills it faithfully is a question about that system. But the field’s existence is a quiet admission that the address of a fact includes its history. Over my working life the question in the title has changed its shape. More often now it arrives as where did this come from?, which is the same question asked by someone who has learned to be wary of a fact with no address.
Knowing Where to Stop #
Every clinic day carries a silent reply to all of this. Nobody can reconstruct the archive at every visit. A clinic day does not allow an hour of searching per patient, and a doctor who treated every gap as a mystery to be solved would see very few people and help fewer. Exhaustive searching can even become a performance, a diligence that reassures the searcher more than it helps the patient.
The objection is right, and the answer to it is not more searching but better sorting. Some gaps bear on the decision in front of you: the result that would change the plan, the history that would alter the risk, the record without which a prescription is a guess. Others are background incompleteness, the ordinary condition of every record, and chasing them is a poor use of the hour. The skill lies in knowing which kind of gap one is standing in, and spending effort in proportion to the difference the missing thing could make. A missing fact that would change the plan deserves a telephone call to another hospital. An interesting detail that would change nothing deserves a note and no more.
Johnson knew this problem from the inside. In the preface to his Dictionary he confessed that he had begun by resolving to leave nothing unexamined, and had soon found that to deliberate whenever he doubted and enquire whenever he was ignorant would have prolonged the work without end, and perhaps without much improvement. One enquiry, he wrote, only gave occasion to another, book referred to book, and “to search was not always to find, and to find was not always to be informed.” Pursuing perfection on those terms was like chasing the sun to the hill where it seemed to rest, and finding it no nearer. So he set limits to his work, which would in time be “ended, though not completed.” No better description exists of a responsible search in a busy clinic. It ends because it must, it is not complete because it cannot be, and its honesty lies in saying which parts were finished.
What remains, when the search has gone as far as it reasonably can, is a statement. I would like every clinician, and every system that assists one, to be able to make it plainly: what was found, and where; what was looked for and not found, and what kind of absence that is; and how the incompleteness bears on the decision being made. Such a statement is less impressive than a confident summary. It is more useful, because it tells the next reader where the edges are.
The second floor plan is never finished. Hospitals change their systems, people leave, new kinds of record arrive and old ones are migrated, not always intact. But the question that makes the plan legible stays the same, and it deserves to be asked aloud, by beginners and by everyone else: where does this fact live, how did it get there, what state is it in, and who can tell me what it means?
Evidence and Editorial Note #
Documented findings. Public biography: MBBS in Amritsar, 2002; PGY-1 at the University of Arkansas for Medical Sciences, 2007–2008. The author’s published account, in the flagship essay on this site, that paper records predominated in his Amritsar training and that electronic records at UAMS in 2007–2008 were in ordinary use but spread across separate laboratory, radiology, order and documentation systems, with paper and fax surviving. Fixed shelf positions by height and date of acquisition in most American libraries when Dewey’s system appeared; Dewey’s development of the scheme at Amherst College from 1873; the 1876 forty-four-page pamphlet, its title, and the concepts of relative location and relative index: Wikipedia, “Dewey Decimal Classification”. Vannevar Bush, “As We May Think”, The Atlantic Monthly, July 1945: the objection to alphabetical and numerical filing, the sentence quoted, association and trails, the memex, and the physician example, as quoted and summarised in Wikipedia’s articles “As We May Think” and “Memex”. HL7 FHIR Release 4 (version 4.0.1): the ten DiagnosticReport status codes and their definitions, including the quoted definition of entered-in-error; the DataAbsentReason codes (unknown, asked-unknown, temp-unknown, not-asked, asked-declined, masked, not-performed and others); the scope of the Provenance resource: hl7.org value-set, code-system and resource pages. Johnson’s visit to Mr. Cambridge’s library in 1775 with Boswell and Reynolds, Reynolds’s aside, and the “two kinds” of knowledge, quoted: Boswell’s Life of Johnson (1791; Project Gutenberg text). Johnson’s Dictionary (15 April 1755), its 42,773-word list, about 114,000 quotations from about 500 authors, and clerical copying of the passages he marked: Wikipedia, “A Dictionary of the English Language”. The preface passage on enquiry, “to search was not always to find”, the chase of the sun, and the work “ended, though not completed”: Johnson’s Preface to the Dictionary (Project Gutenberg text). Daniel Wegner’s proposal of transactive memory in 1985, its study first in couples and families and later in teams, and its definition as individual knowledge plus knowledge of who knows what: Wikipedia, “Transactive memory”. The patwari as the village-level revenue official responsible for land records in Punjab: Wikipedia, “Village accountant”. Sparrow, Liu and Wegner, Science (2011), on remembering where information is stored, and the 2018, 2020 and 2021 replication attempts: McGill University Office for Science and Society (linked). ਪਤਾ as address, trace and knowledge, and the construction mainūṅ patā hai: Wiktionary, with standard Punjabi usage; ਲਾਪਤਾ as “missing”: standard Punjabi and Hindi usage. Vendor representations. None. FHIR is an open standard named for its published definitions; no product is named or assessed.
Recollection and tradition. The author’s training sites and dates are his published biography. No specific record layout, screen, search, colleague or patient is recalled. The laboratory value, the colleague’s question about an earlier investigation and the summary sentence are hypothetical illustrations and are introduced as such; the ward roles in “The Human Index”, the routing of results to trainees who have changed services, and the example of a well-asked question about location are general descriptions of how hospital teams work, not recollections of particular people or occasions. The remark on the wider uses of ਲਾਪਤਾ in Punjab is general and makes no claim about any person or case.
Analysis and forecast. The four states behind “I can’t find it”, the five coordinates of a fact, the reading of the FHIR status list, the reading of Johnson’s two kinds of knowledge and of Reynolds’s joke, the third kind (knowing the condition of what is found), the comparison of the handoff to a memex trail and of a good summary to Johnson’s illustrated definitions, the three modes of search, the six kinds of absence, and the view that a well-marked gap is more useful than a confident summary are the author’s interpretation as of November 6, 2026. Independence. This essay is not sponsored by any institution, company, health system or technology developer. It is personal reflection, not medical advice; readers seeking care should contact their own physician’s office.
About the Author #
Kanwar Partap Singh Gill, MD, traces his family roots to Khadur Sahib in Punjab’s Majha region and grew up in Amritsar. He attended Spring Dale School and completed pre-medical studies at Khalsa College in 1994–1996. He earned his MBBS from Sri Guru Ram Das Institute of Medical Sciences and Research in Amritsar in 2002.
He completed PGY-1 at the University of Arkansas for Medical Sciences in Little Rock in 2007–2008, and PGY-2 and PGY-3 with UCSF Fresno at Community Regional Medical Center from 2008 through 2010. He has practiced medicine in California since 2010.
His career has crossed four distinct information eras: predominantly paper-based medicine, early and fragmented electronic records, the integrated-EHR era, and the emerging age of generative and agentic artificial intelligence.