Season 3, as a map — and the Kubernetes book, closed - #91
Merged
Conversation
Two things, and the second is why the first exists. THE KUBERNETES BOOK IS CLOSED (#88, #89, both closed unmerged). Its premise did not survive contact with its own sources. Three of the five Act I chapters read the same 24-page ACM Queue article, which is a magazine piece rather than a paper; the other two read Borg and Omega, which are not Kubernetes; and nine of fourteen chapters had no paper at all. Strip the answer-key framing and what is left is a book about Kubernetes internals, which is a fine book and not this site's. The ROADMAP carries it the way it carries calc-basic-view — built, rejected, and why — plus the rule that generalises: a book needs a primary source per chapter, or it is a different kind of book and has to say so before it starts. Grepping the sources for the words the chapter list promises costs a day instead of a season. SEASON 3 IS UP AS A CONTENTS PAGE. Thirteen chapters and a close, nothing written. Seasons 1 and 2 argue about data and both assume, without ever arguing for it, that a broken machine has stopped and that the machine itself is free. Season 3 reads the papers underneath that: FLP, Chandy-Lamport, partial synchrony, Paxos Made Live, the Byzantine papers, two choices, delay scheduling, Sparrow, Borg, and the 1999 result that puts a ceiling on what a cache is worth. It closes a hole the adversarial pass found and did not fix. #84 caught Chapter 23 claiming six times that Chapter 7 covered Chandy-Lamport, corrected the sentences, and left the absence — the algorithm is genuinely not in this book. FLP was missing on the same terms. Source: the CS 525 reading list, compared paper by paper. 9 of its 82 named papers are already chapters; 73 are not. The 2025 frontier is left alone, and the ML half stays book D's, which the roadmap assigns by name. THREE TESTS FAILED AND EVERY ONE WAS RIGHT TO. A season needs a route, a card and a page, and routes.test.ts said so. Two of them hard-coded the season list — notChapters = ['/papers/season/2'] and an equality against two literal paths — so adding a season made them report a change rather than a fault; both derive from SEASONS now. remainingLabel said "both seasons finished", which stops being true at three. And one guard was genuinely too strict. It banned an act having a figure without a summary OR a summary without a figure, on the grounds that both render as broken. Season 3 showed the asymmetry: an act whose chapters are not written has words and no drawing, and that is a state every season passes through. The index renders those in one column rather than framing an empty cell — the same hole #82 removed from the panels — and the rule now bans the orphan figure, plus a new one requiring a figure once an act has live chapters. 379 tests, 101 routes. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Deploying with
|
| Status | Name | Latest Commit | Updated (UTC) |
|---|---|---|---|
| ✅ Deployment successful! View logs |
ddia-live-comic | 4e7bd83 | Sep 05 2026, 01:14 AM |
…plans Sam asked whether season 3 covers the UIUC list and whether there is a season 5. No, and yes. Coverage, since it is the obvious temptation: of the 82 papers that overview names, 9 are already chapters and 13 are in the season 3 map, so after season 3 the book will have read 22 of 82 — and the course lists 664. Pursuing coverage turns the book into the syllabus, which already exists and is better at being one. Two candidates, from themes this book has nothing from and that are not about data, which is why they fit nowhere in seasons 1-3. Season 4, the measurement season: everything you have read is a claim, and here is what happened when somebody measured it. COST, where one laptop beats the published cluster results and which the site already cites. Windows NT file system usage, on the list for its method rather than its findings. RocksDB workloads at Facebook. Cloud performance reproducibility, where the answer is no by enough to invalidate conclusions. Strongest candidate, because it is chapter 17's line with evidence and because it is this book's own adversarial method turned on the field the book is about. Season 5, gray failure: it did not crash, it got slow and then quietly wrong. Upgrade failures, silent semantic violations, slow-fault tolerance, plus the verification thread. The natural sequel to season 3's act II — there a machine lies because somebody compromised it, here because it broke in a way nobody modelled. Its caveat is disqualifying until fixed: the papers skew 2021-2025, which is the maintenance class this file refuses elsewhere, so it needs an older floor before it is a season rather than a news roundup. Both recorded as NOT STARTED, under the rule that falls out of the Kubernetes book: no season starts until the previous one is written. Season 3 is 0 of 13. Docs only. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two things, and the second is why the first exists.
The Kubernetes book is closed
#88 and #89, both closed unmerged. The premise didn't survive contact with its own sources: three of the five Act I chapters read the same 24-page ACM Queue article (a magazine piece, not a paper), the other two read Borg and Omega (which are not Kubernetes), and nine of fourteen chapters had no paper at all.
The ROADMAP carries it the way it carries
calc-basic-view— built, rejected, and why — plus the rule that generalises: a book needs a primary source per chapter, or it's a different kind of book and has to say so before it starts. Grepping the sources for the words the chapter list promises costs a day instead of a season.Season 3 is up as a contents page
Thirteen chapters and a close, nothing written. Seasons 1 and 2 argue about data, and both assume without ever arguing for it that a broken machine has stopped and that the machine itself is free. Season 3 reads the papers underneath that:
It closes a hole the adversarial pass found and didn't fix. #84 caught Chapter 23 claiming six times that Chapter 7 covered Chandy–Lamport, corrected the sentences, and left the absence — the algorithm genuinely is not in this book. FLP was missing on the same terms: two chapters about reaching consensus with no impossibility result underneath them.
Source is the CS 525 reading list (UIUC, spring 2026), compared paper by paper: 9 of its 82 named papers are already chapters, 73 are not. The 2025 frontier is deliberately left alone — a paper from this year is a claim, a paper from 1985 is a fact — and the ML half stays book D's, which the roadmap assigns by name.
Three tests failed and every one was right to
A season needs a route, a card and a page, and
routes.test.tssaid so before I'd noticed.Two of them hard-coded the season list —
notChapters = ['/papers/season/2']and an equality against two literal paths — so adding a season made them report a change rather than a fault. Both derive fromSEASONSnow. AndremainingLabel()said "both seasons finished", which stops being true at three.One guard was genuinely too strict. It banned an act having a figure without a summary or a summary without a figure. Season 3 showed the asymmetry: an act whose chapters aren't written has words and no drawing, and that's a state every season passes through. The index renders those in one column rather than framing an empty cell — the same hole #82 removed from the panels — and the rule now bans the orphan figure, plus a new one requiring a figure once an act has live chapters.
379 tests · 101 routes · no overflow or console errors at 390px or 1440px.
🤖 Generated with Claude Code