Marvel's 42-hour Doomsday watch list is an onboarding doc
Marvel published an official watch list before Avengers: Doomsday — 15 titles, 42 hours. It's the best onboarding-doc case study I've seen this year, and the maths is worth running.
The official what to watch before Avengers: Doomsday list is out, and it is 15 titles long and roughly 42 hours of screen time. WIRED's Megan Tomos wrote it up on 5 September, and the number is what caught me.
Not because I'm precious about Marvel. Because I have spent this year writing onboarding docs, and Marvel just published the most honest one I've seen: here is the minimum you must consume before the new thing makes sense. Most engineering teams never admit that number out loud.
🎬 What Marvel actually picked, and what it left out
The "Countdown to Doomsday" playlist on Disney+ has 15 official selections, ahead of the film's cinema release on December 18. The MCU currently runs to 38 movies and 25 series, so this is a hard prune: 15 out of 63.
| # | Title | Year |
|---|---|---|
| 1 | X-Men | 2000 |
| 2 | X2: X-Men United | 2003 |
| 3 | Captain America: The First Avenger | 2011 |
| 4 | The Avengers | 2012 |
| 5 | Avengers: Infinity War | 2018 |
| 6 | Avengers: Endgame | 2019 |
| 7 | Loki (both seasons) | 2021–2023 |
| 8 | Shang-Chi and the Legend of the Ten Rings | 2021 |
| 9 | Spider-Man: No Way Home | 2021 |
| 10 | Black Panther: Wakanda Forever | 2022 |
| 11 | Doctor Strange in the Multiverse of Madness | 2022 |
| 12 | Deadpool & Wolverine | 2024 |
| 13 | Captain America: Brave New World | 2025 |
| 14 | Thunderbolts* | 2025 |
| 15 | The Fantastic Four: First Steps | 2025 |
Two things stand out. The first two entries predate the MCU entirely — they are 2000 and 2003 Fox films. WIRED notes fans read that as a signal, since Patrick Stewart, James Marsden, Ian McKellen and Alan Cumming are all rumoured to return. Rumoured is doing real work in that sentence; treat it as fan inference, not studio confirmation.
The second: The Marvels (2023) did not make the list. A title shipped inside the same universe, two years ago, deliberately excluded from the prerequisites.
Marvel also launched a Countdown to Avengers: Doomsday Official Podcast on Disney+, YouTube and audio platforms, running weekly recaps of the Infinity Saga.
🧮 Run the 42 hours through your own calendar
42 hours is abstract until you divide it. Here is the arithmetic, and I'm showing the working so you can check it rather than trust me.
| Metric | Value | How it's derived |
|---|---|---|
| Total runtime | 42 hours | Marvel's own figure, via WIRED |
| Days from today to 18 Dec | 104 | 5 Sep → 18 Dec 2026 |
| Weeks available | ~14.9 | 104 ÷ 7 |
| Per week | ~2.8 hours | 42 ÷ 14.9 |
| Per day | ~24 minutes | 42 × 60 ÷ 104 |
WIRED frames it as "roughly under three hours a week," which matches. Twenty-four minutes a day sounds survivable. It is also, for most working people I know, the entire discretionary slot in a weekday.
There is a second cost that a US-written article won't price for you: bandwidth. I don't know your bitrate, so don't take my number, take my method:
Do this yourself: check your player's stats overlay for actual MB/minute, multiply by 2,520 minutes. If you assume something in the region of 3 GB/hour for 1080p, 42 hours lands near 126 GB. On a metered home connection or a mobile package, that is a line item, not a rounding error. Verify the assumption before you budget on it.
Every title streams on Disney+, which WIRED describes as available in many markets globally. Check availability where you actually are rather than assuming; several titles are also rentable or buyable through Amazon Prime Video or Apple TV.
🛠️ This is a dependency graph, and Marvel pruned it in public
Strip the capes off and the structure is familiar to anyone who maintains a system with history.
- 63 modules exist. 15 are prerequisites. Marvel didn't tell people to read the whole repo. They published the subset that the new release actually links against.
- Two dependencies predate the repo. The 2000 and 2003 X-Men films are vendored-in external code from before the project started, still load-bearing.
- One recent module was dropped. The Marvels is two years old and excluded. In code terms, it exists, it still runs, and it is not on the critical path.
- The order is chronological, not by release importance. They optimised for a first-time reader's comprehension, not for the maintainers' sense of which parts were hardest to build.
Key takeaway: Every mature system has a prerequisite cost. The mature move is to measure it and publish the pruned list, not to pretend a newcomer can start anywhere.
Most onboarding docs I've read fail on exactly this. They list everything, weight nothing, and quietly push the pruning work onto the person with the least context to do it. If you can't name the 15 things a new engineer must read before your service makes sense, you don't have onboarding, you have a directory listing.
Concretely: take your own README and onboarding wiki, paste them into our reading time estimator, and see what number comes back. If your "quick start" is a 40-minute read, that's your 42 hours. Say it out loud in the doc.
💰 The honest part: this is subscriber retention
WIRED calls the list "a great bit of subscriber farming," and I think that read is correct and worth sitting with. A theatrical release in December is monetised once. A 42-hour prerequisite list monetises the three and a half months in between, on a subscription that has to justify itself every billing cycle.
| What it looks like | What it also does |
|---|---|
| Helpful catch-up guide | Fills ~15 weeks of subscription with owned content |
| Free podcast on YouTube | Top-of-funnel for the Disney+ playlist |
| Chronological ordering | Maximises titles watched vs. a "just watch Endgame" shortcut |
Both readings are true at once. It genuinely helps a confused viewer, and it is a retention mechanic. That's the pattern worth stealing for anyone building a product: the best retention loop is usually a real answer to a real question your users already have. Marvel didn't invent the confusion. They just measured it and shipped the fix on their own platform.
What this means for you
If you're catching up for December, the useful move is to book the 24 minutes a day now and check your data plan before you start, rather than attempting a 42-hour sprint in the last fortnight.
If you're building anything with accumulated history — a codebase, an API, a course, a product with four years of features — take the transferable lesson:
- Count your prerequisite hours. Not vibes. An actual number.
- Prune to the load-bearing subset. 15 of 63 is a defensible ratio. "Read everything" is not.
- Order it for the newcomer, not for you.
- Name your exclusions. Leaving The Marvels off the list is information, and it's kinder than silence.
Marvel's list will get argued about by fans for the next three months. That argument is the point. A published prerequisite list is falsifiable; an implicit one just makes people feel stupid.