Recently, while thinking about AI-assisted writing, I agreed with what seemed like a reasonable condition of authorship, that a person should be answerable to the text carrying their name. If someone publishes an essay, I believed, they should understand what it claims, recognize the thinking inside it, and accept responsibility for having placed it before other people. Such expectation seemed reasonable at the time, even fair. After all, claiming authorship should not allow one to hide behind language whose implications they cannot follow or defend. I had skipped past the word answerable because I agreed with it. Later, I began wondering what kind of answer it demanded.
In practice, being answerable to a text often means being able to defend it verbally and under scrutiny. The writer must retrieve the argument quickly, organize it aloud, regulate whatever anxiety the encounter produces, and respond in a manner that reassures the listener that the writing could only have come from them. What began as a reasonable demand for intellectual responsibility suddenly acquires several additional requirements, none of which is identical to understanding what one wrote. This occurred to me because writing has become the form through which I express myself most fully, as my ability to speak has degraded over the years. Once a charismatic speaker, I must now do with writing as the primary medium through which my thoughts become available to other people. Based on what I write, someone might reasonably conclude that I can sustain a complicated argument. Place me in a live exchange, however, and I may be unable to provide the performance they expect from the person behind the prose. Does that make me less answerable to my writing?
The question extends beyond authorship. We repeatedly attach unrelated capabilities to one another and then behave as though the relationship between them were natural. A great writer should be a great speaker; someone who understands a subject should be able to explain it without preparation; a rational person should remain coherent while being judged and required to justify themselves to someone with power over their future. While those abilities can coexist, they are not the same ability. Authorship may be one of the categories onto which we place more weight than it can carry. While "Author" names a relationship between a person and a work, we often treat it as a comprehensive certification of the person behind it.
Employment interviews perform a similar kind of compression. An interview may be designed to estimate professional competence, but it also measures familiarity with interview conventions. The latter might matter for the work (and this is my being charitable) but it is mainly incidental to it. Yet, the final result is usually described as though the interview had isolated the precise thing it intended to measure. I have experienced the distortion in both directions. Years ago, when I was still a good speaker, I interviewed with a company's founder. He was so impressed by one of my answers that the company accelerated the hiring process and created a position for me. My speech had made something visible to him (intelligence, potential, confidence, perhaps, even, an unusual way of thinking) and the organization reorganized itself slightly to make room for what it believed it had found. When the rubber met the road, I was unable to perform as expected. For a long time, I understood the experience as proof that I could talk but could not do, that I was a smooth speaker with nothing behind the performance. The company had believed me, reality had tested the belief, and I had been exposed.
With time, that interpretation became less convincing: my intelligence could not undo the years of organizational confusion and mismanagement into which I had been placed. Perhaps I was smart, but that did not make me the company's savior. When we say that someone "spoke well but could not perform," we may mean that fluent speech led us to infer intelligence, and intelligence led us to assume that the person would not require adequate conditions in order to thrive. Charisma didn't reveal nothing about me, but it encouraged too much inference.
My present difficulty with speech may do the same thing in reverse. Where eloquence once caused people to imagine capacities capable of overcoming any environment, hesitation may now cause them to overlook capacities that may well exist and would flourish under the right one. In both cases, the proxy becomes a verdict. This is why the language surrounding failed assessment is important. "The candidate could not demonstrate competence" sounds cautious, but it still places the failure primarily inside the candidate. Sometimes, the more accurate conclusion would be that the process did not allow us to determine whether the candidate was competent. The difference doesn't require an employer to pretend that competence was established, nor does it mean every uncertain candidate should be hired. But, it certainly asks the institution to remain honest about the boundaries of what it knows.
"Our process could not read you" is not the same conclusion as "there was not enough in you to read." The reassuring response is to propose a better assessment: ask the person how they communicate best; allow a written response, more time, a quieter environment, or another opportunity to demonstrate the relevant skill. Often, those changes would help. But, I have also experienced the inability to perform better under accommodation. Trauma does not necessarily disappear because the room is quieter; anxiety can migrate into the new format; distrust can make even an apparently considerate encounter feel dangerous. More time may simply become more time in which to struggle. This complicates the comforting story in which the ordinary assessment fails, an accommodation is offered, and hidden competence finally emerges. Sometimes, accommodation vindicates the person. Sometimes, it does not. If answerability depends on successful vindication, then accommodation merely becomes another test a person can fail.
This leaves us somewhere considerably less satisfying. If the original assessment was imperfect and the improved assessment still cannot reassure us, what exactly do we know? Perhaps, the person is capable and our process has failed to reveal it. Perhaps, they are not capable in the way we hoped. Perhaps, the difference itself is more complicated than the process allows us to see. "Our test could not establish your competence" does not mean "you are competent," any more than it means "you are incompetent." The honest conclusion might simply be that we don't know. I think this is difficult for us because assessments are designed to produce answers, and to do so within an expected timeframe. An employer cannot interview someone forever; a teacher eventually has to assign a grade. There is nothing inherently unjust about that. The trouble begins when the practical need for a decision suddenly becomes confidence that the process has revealed the person themselves.
This brings me back to the question that started all of this: what does it mean to be answerable to a text? I had initially treated the idea as straightforward. If I put my name on something, surely I should be able to answer for it. But, the interview problem has made me wonder whether I was collapsing two different things: being answerable to a text and being able to demonstrate that answerability through whatever assessment another person happens to use. Suppose I write something and, months later, someone asks me to explain a particular passage. I stumble, I cannot immediately reconstruct how I arrived at it. I remember the intuition but not the reasoning; I understand what I meant but explain it badly; the question reveals an implication I had never considered; I simply cannot retrieve the thought under scrutiny. Something has certainly been learned from this encounter, but what? That I cannot explain the passage under these conditions? That I do not understand it? That I never understood it? That the text is somehow less mine? These conclusions are not interchangeable, even if we often move between them as though they were.
Perhaps, this is because a text and an account of how the text came to exist are different objects. Writing can preserve a thought that speech can no longer easily retrieve. It can contain a connection reached slowly over several days that cannot be reconstructed in thirty seconds. Still, it can contain something the writer accepted too quickly, understood only partially, or failed to interrogate at all. The difficulty is that these possibilities may look remarkably similar from the outside. A person who understands but cannot explain and a person who never understood may both hesitate when questioned. Eloquence can conceal confusion just as the lack thereof can conceal understanding.
My own history makes that difficult for me to ignore (which should make one wonder why I initially accepted the answerability criteria so quickly in the first place). There was a time when I could speak with enough confidence that people inferred capacities I did not necessarily possess. I could perform well enough in an interview to make someone imagine an employee who was not quite the person they would later encounter at work. Now, I have to reckon with the opposite possibility: that someone may fail to produce the expected performance and cause us to infer an absence that the performance cannot establish. Neither version makes assessment useless, but it certainly makes it inconclusive.
If a writer cannot explain an argument they have published, I think a reader is perfectly entitled to become less confident that the writer understands it. If repeated questioning exposes contradiction after contradiction, that evidence should count. We do not owe every claim of understanding unquestioning belief simply because human beings are difficult to assess. But, evidence that gives us reason to doubt is not always evidence that gives us reason to know. There is a difference between "I am no longer confident that you understand this" and "I now know that you do not." This creates an uncomfortable limit on what another person can verify about a writer. I can read the text, ask questions, watch the writer revise. I can even ask what they meant, why they made a particular choice, what they think follows from their argument, or whether they still endorse it. All of these things give me evidence about their relationship to the work. None gives me direct access to that relationship. Eventually, I am making an inference from what the writer can make legible to me.
The writer faces a corresponding limit. I cannot simply announce that I understand something and require everyone else to believe me. My private certainty does not become another person's evidence merely because I possess it. If I repeatedly fail to demonstrate any meaningful relationship to something bearing my name, other people are allowed to find that suspicious. They may even have good reason to withhold the judgment I want from them. The fact that their test is imperfect does not obligate them to conclude in my favor. So, perhaps answerability and demonstrability need to be separated without being severed. If they were identical, then the person best able to perform understanding would necessarily be the person who understood best, which I know from experience is not true. But, if they had nothing to do with each other, then answerability would become an entirely private claim immune to scrutiny. I could put my name on anything, insist that some inaccessible relationship existed between me and the work, and treat every request for evidence as an illegitimate demand. That cannot be right either.
What, then, should answerability require? I am drawn to a less theatrical definition: perhaps, I should recognize the thinking in a text as something I understand or endorse; I should understand its central claims well enough to know what I am placing before other people; I should have exercised judgment over what appears under my name. And, perhaps most importantly, I should remain responsible for it when someone discovers a problem. "I wrote it" should mean that criticism eventually reaches me rather than disappearing into whatever process helped produce the words.
Even this does not give us perfect answerability, since a writer can misunderstand an implication of their own argument, or they can discover that something they endorsed yesterday no longer survives scrutiny today, among other things. If answerability requires complete cognitive sovereignty over a text, then, I am not sure writers have ever been answerable in the way we sometimes imagine.
Perhaps, we ask for this sovereignty because it would make authorship easier to police. If the writer can always reproduce the thought on command, we can reassure ourselves that the thought belongs to them. If they cannot, doubt enters. But, doubt may be unavoidable whenever one mind tries to establish what is happening inside another. The best we can sometimes do is gather evidence and remain proportionate in what we infer from it. This is what the interview taught me that I had not initially understood about writing. Every assessment has a boundary between what it observes and what it concludes. Sometimes, that boundary is narrow enough that we barely notice it. Sometimes, a person's opacity makes it impossible to ignore.
We still have to hire people, evaluate work, assign responsibility, and decide what we believe. The challenge is to resist turning those necessary judgments into claims of knowledge larger than the evidence can support. "I cannot verify this" is a real conclusion. "I doubt this" is a real conclusion. "I do not have enough evidence to trust this" is a real conclusion. None is identical to "I know what is absent inside you." And, so, I return to the question I had skipped past because I thought I already knew the answer. Am I answerable to my text? I think I am, but not because I can guarantee that I will reproduce its reasoning whenever someone asks. I am answerable insofar as I recognize what I have put before other people and remain available to the consequences of having said it. There may be moments when I cannot make that relationship legible to someone else, and they may be justified in remaining uncertain about me.