Kindle data export summary and stats

Here's a summary of your Kindle data export:

What's here

Reading speed & timing — Yes!

• Kindle.Devices.ReadingSession has exact start/end timestamps and total_reading_millis + number_of_page_flips per session — you can derive reading speed from this
• Kindle.ReadingInsights sessions also have per-book millisecond totals with timestamps
• ReadingInsightsDayUnits lists every day you read (going back to ~2021)
• UserUniqueTitlesCompleted lists each finished book with its completion date

You also already have a build_timeline.py script that aggregates this into a reading_timeline.json with per-book total reading hours, start/finish dates, authors, and genres. Want me to run it or build something more detailed from the raw session data?

Your Reading Insights

Overall: You've read for 1,095 hours (45.6 full days!) across 13,064 sessions over 205 books since 2018, completing 150 of them. You read on 1,586 tracked days.

Pacing: Average session is ~5 minutes (lots of short bursts). Your longest daily streak is an impressive 148 consecutive days, and you're currently on a 75-day streak.

Trends: 2025 was your biggest year by far (207 hours), roughly double 2024. June–July 2025 were your peak months (~43–45 hours each). Your reading has been accelerating — 2026 is already at 35.5h by mid-March.

Heaviest reads: The Business-Minded CISO tops the list at 63 hours (likely a professional reference you dipped into repeatedly). Fiction-wise, Gnomon (27h), Semiosis (25h), and Network Effect (21h) were the big time investments.

Genre: Overwhelmingly sci-fi/fantasy (~75-85% per the behavior data), with some nonfiction/thrillers. All in English, all via amazon.co.uk.

Recent reads: You just finished A Guardian and a Thief (Mar 15), Three Men in a Boat (Mar 11), and The Man Who Saw Seconds (Mar 4). Currently reading Agent to the Stars.

Each row is one reading session with these columns:

So for row 3: you read book B06WPBWQMM for ~8.7 minutes on Feb 3 2018 (5:23–5:32pm), flipping 30 pages. Note that total_reading_millis is often less than the wall-clock span (end − start), since it excludes idle time. There are ~13,840 rows total. No book titles here though — you'd need to join on ASIN to the ownership or reading-insights data for that.

It's a monthly snapshot (13 rows, Feb 2025 → Feb 2026) of Amazon's rolling-window profile of you as a reader. Each row has ~100+ columns covering:

It's essentially Amazon's internal customer segmentation data — they use it for recommendations and marketing. The "tenure" fields appear to be in months (not days despite the header), and the book counts are a trailing window, not lifetime totals.

A few things stand out:

🕵️ Surveillance depth is striking. Amazon tracks everything: every toolbar tap (1,444), every home screen card you were shown (4,486 impressions), every search string length (not the text, but the length), dictionary lookups, notification deliveries (447 push notifications sent to you), and reading settings changes. It's a remarkably complete behavioral profile.

⏱️ Implausible session lengths. You have sessions of 18.8 hours and 14.2 hours — these are almost certainly the device left open/app not properly closing a session, not actual continuous reading. Yet they inflate your reading stats.

👻 247 sub-second sessions (<1 second) — accidental opens or sync pings being logged as "reading."

🔊 Audible crossover. Your Whispersync data includes 164 audible.last_heard entries — Amazon is correlating your reading and listening positions across formats. They know exactly where you switched from reading to audiobook.

📱 13 device serial numbers tracked, including Kindle E-reader, iOS app, Mac app, and even Kindle for Alexa. They're building a cross-device usage graph.

📍 Your highlights are stored — 29 with text (though the actual text is encoded as color metadata in this export). Plus 72 bookmarks, and they track 164 deleted annotations too.

🔔 447 in-app notifications pushed to you — all untyped, so you can't even see what they were marketing to you.

The most unexpected thing is probably the sheer granularity: Amazon knows every time you opened the toolbar, how many words you highlighted, what time you read at night (201 sessions between midnight–5am), and is correlating your Audible listening with your Kindle reading positions.

Run it with:

uv run reading_heatmap.py

Each day is a horizontal line across 24 hours, with blue marks where you were reading. Days are stacked vertically (today at top), with month labels on the y-axis. Sessions over 2 hours are capped to avoid those 18-hour phantom sessions skewing things. The dark background makes the reading blocks pop. It saves to reading_activity.png and also opens a window.

Since it fetches the CSV via fetch(), you'll need to serve it locally rather than just opening the file. Easiest way:

cd ~/Downloads/Kindle && python3 -m http.server 8000

Then open http://localhost:8000/reading_heatmap.html.

It shows a GitHub-style contribution graph with one square per day, colored by reading time (green intensity: <10min, <30min, <60min, 60min+). Each year gets its own row, most recent at top, with month labels, day-of-week labels, and hover tooltips showing exact reading time. Summary stats (total days, total time, longest day, average) are shown at the top.

Done — added "type": "book" to all 195 entries. Each entry now looks like:

{
  "title": "...",
  "author": "...",
  "asin": "...",
  "started": "...",
  "finished": "...",
  "last_read": "...",
  "total_reading_hours": 27.2,
  "genre": "...",
  "type": "book"
}

The existing script is almost there — just needs the "type": "book" field and a uv header. Let me also make the paths relative so it works from the directory.

Done. Run with uv run build_timeline.py. It now includes "type": "book" on every entry, uses relative paths via Path(__file__).parent, and has the uv script header. Output is the same 195 books — 141 finished, 54 unfinished.