Decisions: dispositions for the 24 open lecture-proposal issues

[mmcky]

An audit of every open issue in this repo that proposes a piece of lecture content found 24 of them, some dating to 2019. Three describe work that shipped years ago and was simply never closed; three more were built and then retired or reversed; the rest were never started. This issue is where we decide what happens to each one.

The full audit — with the concrete evidence behind every call, file by file — is in reports/lecture-proposal-issues-audit-2026-07-22.md.

How to respond: tick a box to accept the proposed disposition. To disagree, leave it unticked and say so in a comment — an unticked box means "not yet decided", not "rejected". Nothing here has been actioned; no issue has been touched.

Read these three first

These need care in how they close rather than whether, because the obvious closing comment would record something that isn't quite true.

#28 — four of Tom's seven bullets shipped, not all seven. Two genuinely remain unbuilt: Jordan form, whose only trace in the org is a commented-out remark in the Networks book source, and the Courant–Fischer min-max characterisation as a route to PCA (svd_intro.md derives PCA from the SVD instead). Close against what shipped and carry the two residuals forward as their own issues.

#29 — cite the public artifact, not the source. The published §2.2.2 of Economic Networks asserts the qualitative conclusion but ships the lognormal comparison as an unsolved exercise, and the worked solution lives in book-networks/archived/figures_source/granular.ipynb, which is in a private repo. The closing comment must point at networks.quantecon.org and the public PDF.

#89 — close as not planned, not as completed. The teaching content exists, so this reads at first like a delivered issue. It is not: the org reversed the issue's premise when lecture-python-programming#341 removed pandas_datareader in March 2025. Closing it as completed would put a misleading record in the repo.

Close as completed — the content exists

• [Lecture] Capital Structure #27 Capital Structure — the Bisin, Clementi & Gottardi paper became BCG_complete_mkts.md and BCG_incomplete_mkts.md in lecture-python-advanced.myst, merged September 2020
• [Lecture] eigenfunctions and eigenvalues #28 Eigenfunctions and eigenvalues — eigen_I.md, eigen_II.md, greek_square.md, plus chow_business_cycles.md and long_run_risk_operator.md (see note above; two residuals to spin out)
• [Lecture] Granular origins of aggregate fluctuations #29 Granular origins of aggregate fluctuations — Economic Networks §2.2.2 (see note above; cite the public PDF)

Close as not planned — built, then retired or reversed

• [New Series] Applications of Statistics #101 Applications of Statistics — executed exactly as specified in lecture-stats, but the site has not rebuilt since January 2024, is unlisted, and PROJECT: New finance lecture series — finance.quantecon.org (next after lecture-dp) #312 describes the topic-split experiment as retired
• [Lecture Series] Python for Econometrics and Machine Learning #50 Python for Econometrics and ML — the Stata half is real; statsmodels.md is 30 bytes and scikit-learn.md is 16, and all 15 commits fall in a three-day window in November 2021. Finer-grained tracking already lives in that repo's own issues
• Accessing Data using Python #89 Accessing data using Python — superseded and reversed (see note above)

Close as not planned — stale link-shares

No proposal was ever written down, and in several cases the source itself has decayed, so the issue no longer records what was wanted.

• [Lecture] Oberfield's theory of input-output architecture #22 Oberfield's theory of input-output architecture — zero of 14 lecture bibliographies cite it, and the Networks book explicitly declines to cover endogenous network formation
• [Lecture] LQ portfolio choice #24 LQ portfolio choice — the 2014 blog URL is now a parked domain; black_litterman.md predates the issue by four months
• [Machine Learning] Excellent use of Machine Learning as Companion Techniques for Economics #66 Machine learning for poverty maps — one "nice find" reply, no lecture named
• [math] Generating LaTeX that represent Python Functions #68 Generating LaTeX from Python functions — Google archived latexify_py in February 2025; the issue has no cross-reference in four years
• [reading] A New Way to Predict Probability Distributions #77 A new way to predict probability distributions — a vendor-feature tutorial since overtaken by conformal prediction
• [math history] Determinants from a history perspective #82 Determinants from a history perspective — the thread is now login-walled; every determinant mention across five linear-algebra lectures is mechanical

Convert to a discussion — live ideas, no unit of work

These still stand up intellectually. What none has is a champion, and as open issues they read as tracked work that isn't. Ten would collapse into two or three threads: #11, #22 and #29 are one production-networks conversation; #66, #70, #77, #81 and #203 are one reading list.

• [Lecture] New lecture on micro shocks by Baqaee and Farhi #11 Baqaee & Farhi micro shocks — an org-wide search for Hulten returns zero
• [Lecture] De Nardi richer income dynamics paper #12 De Nardi richer income dynamics — the volunteer has not touched a lecture repo since March 2022; the idea still fits the actively-extended IFP series
• [Lecture] Contagion and default? #25 Contagion and default — nothing across four separate arellano.md files, all Arellano 2008; one of the two links now 404s
• [Lecture] Economic geography? #26 Economic geography — Krugman's racetrack economy, still uncovered anywhere, including the Julia series and the sandpit repo
• [Book] Mathematical Logic through Python #67 Mathematical Logic through Python — outside the computational-economics remit, so no repo is a natural host
• [paper] Interesting application of Machine Learning #70 Text-as-data measurement of industrial policy — the paper's data and algorithms are not public, so it cannot be reproduced as a lecture; the reading-list value survives that
• [reading] Interesting comparison Statistics vs Deep Learning #81 Statistics versus deep learning for forecasting — a bare URL, untouched for 3.3 years
• [resource] Nowcasting Approach -- application to World Trade #104 Nowcasting world trade — a methodology reference with no lecture, series or owner named
• [Resource] A "resources" lecture series that enables explanation topics that can be referenced from lectures #141 A shared "resources" series — no site, no repo, no comment in 27 months (spin out the one real deliverable first, below)
• [teaching example] Working with AI needs detailed reviews? #203 Working with AI needs detailed reviews — the transcript is pasted inline rather than linked, so unlike the bare-URL issues its content has not decayed

Move to a lecture repo — actionable, mis-filed here

The only two entries specified tightly enough for someone to pick up and build.

• [Lecture] Gale Shapely matching #23 Gale–Shapley matching → lecture-python-intro, beside schelling.md, short_path.md and lp_intro.md. The tempting near-match in Economic Networks is linear assignment, not deferred acceptance
• [lecture] Building a micro GPT (Blog Post) #287 Building a micro GPT → lecture-python.myst, which already owns back_prop.md and finite_markov.md

Follow-up issues to create

• Pareto goodness-of-fit test in lecture-python-intro/lectures/heavy_tails.md — the one concrete deliverable buried inside [Resource] A "resources" lecture series that enables explanation topics that can be referenced from lectures #141, and a real gap regardless of what happens to that issue
• Jordan form in lecture-python-intro — residual from [Lecture] eigenfunctions and eigenvalues #28
• Courant–Fischer as a route to PCA in lecture-python.myst/svd_intro.md — residual from [Lecture] eigenfunctions and eigenvalues #28
• Add lecture-econometrics-machine-learning to the archive review in Review: Archive candidates for inactive repositories #267 — last pushed 2022-04-11, unlisted on quantecon.org

Scope of the audit

An issue was catalogued if it proposes a specific piece of lecture content — a new lecture, a new series, or a named topic to be written up. Cross-cutting maintenance sweeps, infrastructure and tooling are out, however much lecture content they eventually touch. Four issues sit close to that line and were ruled out: #258 (reorganisation, and largely executed now lecture-dp exists), #312 (presents as additive, but every lecture it names already exists), #301 (content production, but half licensing and infrastructure), and #173 (a maintenance sweep, and being reorganised out of meta under #343).

• Confirm the scope boundary above, in particular that Lecture Improvement Project #173 stays out

---

Every finding was checked against the current state of the lecture repos and then independently re-checked by a second pass tasked with refuting it. Candidate files were opened and read; a matching filename was never accepted as evidence of delivery.