Timing: Assigned in Class 12, due before Class 13.
Prerequisites: Your ballpark item exists either as a pull request against econ-ark/ballpark (per Matsya, ballpark, and Dolo Plus draft) or as a merged directory there. You have read the newly-articulated ballpark quality criteria in CONTRIBUTING.md. If you received a tier classification in your Class 12 assessment, re-read it — it tells you exactly where your item stands on the Draft / Primer / Formalized scale and what’s missing to reach the next tier.
Read this first: a fully worked, step-by-step example of a student successfully carrying out this assignment on the Benhabib, Bisin, and Luo (2019) ballpark item is published at
llorracc.github.io/workspace-course-topics/worked-examples/benhabib-formalize-finalize/. The worked example is structured as ten rounds (Round 0 Setup through Round 10 YAML generation) and ends with a consolidated “Lessons learned” section. Skim it before starting your own work, and refer back to specific rounds when you hit the analogous step on your own paper.
Use a strong AI model (Claude Opus 4.7 on the “high” or “xhigh” or “max” setting recommended) as a critical reviewer of your ballpark item. Have it read your committed artifacts against the quality criteria in the new CONTRIBUTING.md, and produce a prioritized list of substantive changes that would push the item toward — and polish it within — the Formalized tier.
Then pick the top 3–5 changes and actually make them, as a follow-up PR against your ballpark entry. The goal of this assignment is not generic polish; it is to formalize what is not yet formal, and to finalize (close open ends, resolve workaround flags, fill in gaps) what is already formal.
This is the kind of critical-review task where a strong model materially outperforms a fast one. The review produces a prioritized list grounded in multiple documents (your notebooks + CONTRIBUTING.md + the paper PDF); quality matters more than speed. Use Claude Opus 4.7 in Cursor or Claude Code — confirm the model indicator in the UI before sending the prompt. If Opus 4.7 is unavailable, Sonnet 4.6 is an acceptable substitute; avoid Haiku for this task.
We will craft this prompt together in class. The template below is the starting point; adapt it for your ballpark by replacing the Benhabib paths with yours and tailoring the item-specific questions in Section 1.
Please review the ballpark entry for Benhabib, Bisin, and Luo (2019) at
https://github.com/llorracc/ballpark/tree/master/models/We-Would-Like-In-Econ-ARK/Benhabib_et_al_2019against the quality criteria articulated in
https://github.com/llorracc/ballpark/blob/master/CONTRIBUTING.mdThe entry is currently at Primer tier. I want a prioritized list of substantive changes that would help the item reach — and be finalized within — the Formalized tier. Focus on the economics, the model statement, the literature coverage, the dolo-plus / DDSL formalization (if any), and the
AGENTS.md. Not cosmetic changes (formatting, typos, citation style).Specifically, please address:
“The Model” section in
Benhabib_et_al_2019_summary.ipynb. Is the recursive formulation explicit enough that matsya can build the Formalized-tier layer on top of it? Does it admit an unambiguous perch decomposition, or does it carry ambiguities that will force a matsya or a later formalizer to reinterpret?Literature coverage. Are the 3–6 papers in
_prior-literature.ipynbthe right ones? Missing any foundational work? Is the subsequent-literature notebook capturing the most influential follow-ups through today?What are the paper’s main distinguishing features? Is this surfaced clearly in
_summary.ipynb?
AGENTS.md. Are the “Known model features requiring attention” complete? Are the “Common next tasks” well-scoped (grounded in actual gaps, not aspirational)?Anything else a thoughtful graduate student would want clarified before attempting the Formalized tier for this paper.
Output format: a numbered list of 4-6 proposed changes. For each give (a) the problem in one sentence, (b) the proposed substantive change, (c) where in the item the change goes (file, section). Omit cosmetic items.
Tell your Cursor / Claude Code thread, at the start of the work:
Create a working branch in your ballpark fork named <citekey>-formalize-finalize (e.g. benhabib-formalize-finalize for the Benhabib et al. (2019) worked example). Branch from origin/master so your commits land on a clean base and will rebase cleanly when you open a PR. All commits in this assignment land on this branch — do not touch CONTRIBUTING.md or other repository infrastructure here; that is scoped to separate PRs.
Reuse your existing matsya --session name for this ballpark. If you worked on the same paper in Matsya, ballpark, and Dolo Plus draft (#130), your session was topics2026-<citekey>. Do not open a new session — every matsya call in this assignment should continue the accumulated server-side context so (a) instructor review can inspect the full thread, and (b) matsya has the context it built up. Record the session name in matsya-session.txt in your item directory if not already committed, and tell your Cursor / Claude Code session at the start of the thread: “For all matsya interactions in this thread, use --session topics2026-<citekey> — do not open a new session.”
AGENTS.md), include those explicitly in the prompt so the model addresses them.verification.md discipline from Matsya, ballpark, and Dolo Plus draft.<citekey>-formalize-finalize branch. Push the branch and open a pull request against your ballpark’s upstream (econ-ark/ballpark:master) titled Formalize and finalize <citekey>: revisions from Claude Opus 4.7 review.Submit the PR URL and include in the PR description:
CONTRIBUTING.md tier checklist.CONTRIBUTING.md — the quality criteria and tier system.models/We-Would-Like-In-Econ-ARK/Benhabib_et_al_2019/AGENTS.md — reference AGENTS.md.