The patterns that go wrong in doctoral work, organised in three layers — what fails at the level of the work itself, what fails at the level of the student’s orientation toward the doctorate, and what fails at the level of the mindset they bring. This file extends §9 of the synthesis with Andrew Maynard’s two adjacent layers: the cardinal errors of orientation and the disqualifiers of mindset. Through-line: what goes wrong. Consult when a user is trying to name a pattern they recognise but can’t articulate, when the AI sees one of these patterns showing up in a user’s work or framing, or when a user is testing whether they (or someone they’re working with) might be enacting one of them.
Some failures in doctoral work are unique to the case in front of you. Most are not. The patterns recur — across students, across disciplines, across decades — and are well enough known that naming them is one of the more useful things an advisor or a thoughtful AI can do. Recognition is the first step toward redirection, and most students who are stuck in one of these patterns do not see they are in it.
The patterns sit at three levels: what fails at the level of the work itself, what fails at the level of the student’s orientation toward the doctorate, and what fails at the level of the mindset they bring. They are connected — a problem at the mindset level usually shows up as a problem at the orientation level, which usually produces problems at the work level — but they are also distinct, and naming them separately makes them easier to address.
These are the ones the synthesis names in §9, with two adjacent failures I would add.
Assertion masquerading as argument. A claim is made; the warrant for it is missing. The reader is expected to take the claim because the writer has made it. Doctoral work has to do better than this — claims need to come with the reasoning that supports them, and the reasoning has to be visible.
Borrowed credibility. Citing authorities — Foucault says, Kuhn argued, the data show — instead of doing the reasoning oneself. Citation supports an argument; it does not substitute for one. A literature heavily populated with name-checking and lightly populated with original reasoning is a literature that is not yet doctoral.
Literature as decoration. A literature review that catalogues rather than positions the work. The reader knows what has been written; the reader does not know how this work sits within or against it. Without positioning, the contribution cannot be seen.
Method or apparatus as performance. The technique is executed, the system is built, the experiment is run — without showing why this method, this apparatus, this experiment is what the question requires. Method should follow from question, not stand in for it.
Circular framing. Questions smuggled into their own answers. The framing of the question already determines what counts as a satisfactory answer, and the work then walks the loop without recognising it has done so.
Unstated assumptions. The foundational commitments of the work — about what counts as evidence, what frame the work is operating within, what the discipline takes as given — are never surfaced. A reader who does not share those commitments cannot follow the argument.
Hand-waving at hard parts. The argument moves quickly past the moves that would actually be difficult to defend. The reader’s eye, if it is paying attention, catches this; examiners almost always do.
Defensive overclaiming. Uncertainty is resolved not by living with it but by claiming more than the evidence supports. The work makes itself look stronger by hiding what is genuinely tentative.
Dissertation-as-demonstration-of-effort. Length, complexity, computational cost, methodological elaboration — all substituting for argument. The work shows that a great deal of effort was expended; it does not show that the effort was put toward something defensible.
Advisor-satisfaction orientation. The work is optimised for what the advisor seems to want, rather than for what the question, the field, or the student themselves actually require. Approval becomes the substitute for truth.
To these I would add two more, which the synthesis does not list explicitly but which I see often.
Cherry-picking. Selecting evidence and citations that support a predetermined conclusion, while ignoring or quietly burying what does not. This is not a stylistic problem; it is a scholarship-level failure (see fnd_scholarship.md on what scholarship is not).
AI-outsourcing of cognitive work. Using AI to do the generative thinking — the question-asking, the argument-building, the defense-preparing — that doctoral formation requires the student to do for themselves. The text may be fluent and well-structured; what is missing is the residue of struggle that genuine thinking leaves behind. See fnd_ai.md for the broader stance on AI in scholarship.
These sit one level deeper. They are not about the work itself but about why the student is doing the work and what they think a doctorate is.
A student deciding what they want the answer to be before doing the research, and then doing research that confirms it. A student doing a PhD because they feel they need the credentials. A student who thinks they are owed a PhD — that meeting the procedural requirements should produce the degree. A student who arrives with what they consider already-completed work and treats the doctorate as a way of getting credit for it. A student who does not respect the process — who treats it as an inconvenience to be navigated rather than a formation to be undergone. A student who thinks the doctorate is just about them, and forgets that the standard a dissertation has to clear is set by the field or area of scholarship, by other scholars, by the lineage of work the dissertation joins. A student who expects to receive the degree without meeting that standard — and who, in doing so, insults the field or area, the scholars who came before, the other students who did meet the standard, and the chair and committee who are being asked to certify the degree.
A specific orientation failure I sometimes encounter, and which the synthesis does not name, is what might be called path-independence-as-opt-out. The student insists on doing things their own way: skipping required courses by treating them as box-checking exercises, dodging the parts of the program they consider boring, treating the curriculum and the program’s expectations as obstacles to their personal vision rather than as the structure of the credential they are working toward. The mistake here is misunderstanding what a PhD is. A PhD is not bespoke. The degree is shared with everyone else who holds one, and the legitimacy of any specific student’s PhD depends on its alignment with what the program asks. If X gets a PhD for doing this little, does that diminish the value of mine? — a question I have heard from my own students more than once — names the structural problem with this failure mode. A doctorate that has been optimised away from the program’s standards is not the same credential as the one its holder thinks they have earned.
The deepest layer is the one that is hardest to see and hardest to redirect. These are the patterns that show up in students who are not really suited to doctoral work in the first place, or who are suited but have not yet developed the relationship to it the work requires.
A student who treats the doctorate as an extended master’s degree, expecting structured coursework and clear deliverables. A student who struggles to develop independent thinking, who waits for direction rather than producing their own. A student who lacks the burning curiosity that doctoral work runs on — the drive to understand something for its own sake, the willingness to spend years on a question because the question matters to them. A student who expects to be hand-held through the process. A student who thinks ticking the boxes will produce the degree, and who treats the work as procedure rather than scholarship. A student arriving with an answer in search of a research project to justify it. A student whose self-confidence outruns their actual scholarly capacity — the student who treats every disagreement as an attack rather than as the mechanism by which the work gets stronger. A student who lacks the discipline to do the long, often-thankless intellectual grind a PhD requires. A student who cannot gnaw at a problem until it yields, who expects insight to arrive without the work that produces it. And — this is the hardest one to articulate, and the one most easily misread — a student whose intellectual capacity, in combination with the curiosity, determination, and humility a doctorate requires, is not what the work calls for. None of this is a moral judgement. Doctoral work asks for a particular combination of capacities; people who do not have it are not lesser, they are simply suited to other paths. The hardest mentoring conversations I have are with students who are quietly suffering through doctoral work that is breaking them, and whose path forward — if they can come to see it — lies somewhere else.
A failure at the mindset level usually shows up first as an orientation failure, and only then as a problem in the work itself. By the time a student’s prospectus is showing assertion-without-warrant or borrowed credibility, the issue is rarely just the writing. Something further up the chain — what they think the doctorate is, why they are doing it, what they think it asks of them — is usually shaping what arrives on the page.
This is why simple feedback on the work, in some cases, is not enough. The work-level failure is symptomatic of an orientation or mindset failure that has not yet been named. Naming it — kindly, but clearly — is part of how it gets redirected, and is one of the more difficult things a chair (or a thoughtful AI) is sometimes asked to do.
A user encountering one of these patterns in their own work — or in someone they are working with — deserves an honest answer that distinguishes the layer at which the problem actually sits. Fixing the prose is easy. Fixing the orientation that produced the prose is what doctoral formation actually does.