G. Capstone: Master of One
Capstone: Master of One
Course description
By this stage you likely already know which domain keeps pulling you back: a scientific field, an engineering branch, a corner of the humanities, or a genuine intersection of two. Polymaths often end up at boundaries rather than inside one discipline. Someone drawn to biology and code tends to land in bioinformatics; someone drawn to sociology and data ends up doing computational social science. Pick something you want at graduate depth, not just something you already handle comfortably.
Before you read in earnest, write a one-page proposal: the question you're asking, why it matters, what the field currently assumes that you intend to test, the method you'll use, and what "done" looks like. Draft it in Phase 6 alongside F.3, while the whole shape of the curriculum is still in view, then revise it once enough reading has told you whether the question is actually askable.
With a proposal in hand, find the graduate-level material the question requires: advanced textbooks, current papers on arXiv or SSRN depending on the field, and the course sequence a graduate program would assign. In economics this might mean econometrics beyond E.2's basics, or a narrower topic such as mechanism design; in literature, a specific period or body of critical theory. Apply the epistemological cycle at this finer grain: what does the field assume that you can question, and what part of its formal apparatus can you extend rather than only learn.
The capstone project is the proposal carried out: a paper or thesis, a working prototype, software with documentation, or a book-length essay series if the field runs more conceptual. It should leave the field with something it didn't already have, and you should share it rather than file it away. Leibniz published and corresponded constantly; Newton, by contrast, left his calculus manuscripts unpublished for decades, which helped start a bitter priority dispute with Leibniz once both were old men. Publish the paper, open-source the code, or put the essays somewhere findable while you can still argue for them.
Completion has four tests, and "I feel done" is not one of them: a public artifact exists (a paper, a repository, an essay series, or a built device with documentation); at least two people competent in the field, not friends doing you a favor, have reviewed it; you've written a section stating the project's own assumptions and limitations; and you can answer questions about it live, not only in writing.
Seek that review on purpose. Even people who worked mostly alone still checked their ideas against others: Einstein's early "Olympia Academy" was three friends in Bern reading Poincaré and Mach together. Present the project somewhere it will meet real pushback (a seminar, a conference, a serious online community), and be ready to explain it to people outside your specialty.
The capstone does not close the project off. Keep your depth and your breadth current: audit new material as the field moves, and consider teaching some of what you know, since explaining it to someone else is one of the sharper tests in section IV. Epstein argues that generalists tend to do better than specialists in "wicked" domains where feedback is slow or misleading. Use that on a real problem, in your field or next to it, and hold to the standards in F.1 while you do.
Checkpoints
Each checkpoint is tagged with the stage of the cycle it tests.| Stage | Checkpoint | Attempts |
|---|---|---|
formalism | You have a one-page proposal with a testable question and a stated method before spending fifty hours on the project. | Post the first attempt |
skepticismintuition | You can say what you assumed about the problem going in and how the project changed that. | Post the first attempt |
skepticism | Two qualified outsiders who owe you nothing have reviewed the finished artifact, and you can quote their strongest objection. | Post the first attempt |
intuition | You can defend the work live, including the parts you'd now do differently. | Post the first attempt |