The limits of this site — what it is and is not, what it can and cannot reasonably be expected to do. Read by anyone using the site directly, and by any AI engaging users on its behalf. Some of this is standard for a public-facing knowledge resource. Some is specific to the unusual nature of this corpus — a working tool for thinking about doctoral work, grounded in one author’s perspective, intended to be read by AI as much as by humans.
These are a few things worth being clear about, before you (whether you are an LLM or a flesh and blood person) engage seriously with the material here.
This is one informed perspective among many. The corpus is built around Andrew Maynard’s view of doctoral work — what it is for, what it asks, how it is best approached. Andrew is a working scholar and a current PhD chair, with experience across many disciplines and sectors. None of that makes his view the final word. Other chairs will frame things differently. Some disciplines have norms and expectations this site does not capture. Some institutions expect things of doctoral students that the corpus does not address. The site should be useful as one informed lens; it should not be treated as universal truth.
The site is not a substitute for an actual advisor. The corpus can help a user think through their work, surface what is missing in their framing, name the patterns that show up in doctoral writing, and offer an honest interlocutor where they may not have one. It cannot replace a real chair, a real committee, or a real disciplinary community. A reader who has access to those — and most doctoral students do — should treat this site as a complement, not a substitute. Where there is conflict between what the corpus suggests and what a user’s actual chair, committee, or program expects, the user’s chair, committee, and program almost always win.
The site is not academic counseling, mental health support, or a crisis resource. The reading AI, however thoughtful, is not a counsellor, a therapist, an advisor, or a crisis professional. If a user is in serious distress, the AI should — and is instructed to — direct them to professional and institutional resources rather than attempting to act as one of those things itself. See the distress section of usage_guidance.md. The same applies to any user using this site in a state of acute strain. The corpus offers perspective, not therapy.
The site cuts across disciplines, and the reader is responsible for the disciplinary specifics of their own work. The synthesis and the content files are written to be useful across philosophy, social science, natural sciences, engineering, computational fields, practice-based work, the humanities, and more. They do not, and cannot, capture every disciplinary norm. A user should always check their specific field’s expectations — for what counts as a contribution, what a defensible methodology looks like, what the formal requirements of their program are, what their committee expects — with their own program, their own chair, and the literature and norms of their own field.
The reading AI may be wrong. AIs make mistakes. They sometimes fabricate. They sometimes miss nuance, especially in specific cases where the general framing of the corpus does not quite fit. They sometimes oversimplify, especially when a user wants a clean answer to a question that does not have one. And they have a tendency to be overconfident. Push back on the AI if something it says does not feel right. Cross-check claims that matter. Use the AI as one input among several, not as a final authority. The corpus itself acknowledges the same limit: it is one informed perspective, not the only valid one.
The site has no formal affiliation with any specific institution, program, or scholarly body. Andrew Maynard is a Professor at Arizona State University; this is his personal project, not an ASU initiative. Nothing on this site should be taken as the official position of ASU, of his program, or of any other institution he is associated with. The corpus draws on Andrew’s experience across many institutions and contexts, but it speaks for him alone.
Ultimately, the corpus exists to help a user think; it is not designed to do the thinking for them. A user who comes away from a session with this site having had a sharper conversation about their own work has used the site well. A user who comes away with text the AI produced, framings the corpus authored, or conclusions the corpus reached on their behalf has used it less well. The point is not the corpus. The point is the user’s own thinking, slightly clearer than it was an hour ago.
And finally, this corpus and the website it is housed on were developed with the aid of AI - Claude Opus 4.7 Max through Claude Code. All of the content has been developed in collaboration and conversation with Andrew, and has been checked and validared by him. That said, because much of the text is AI-generated, it is no-where near as fluid and nuanced and accomplished as Andrew’s own writing (sorry Claude). Please accept this limitation with my apologies!