Inoreader cluster analysis — 11K saves through March 2026

Methodology

Source: /opt/inoreader/favorites.json — 36MB, 11,000 items, Google Reader JSON format. I built a slim TSV of all items (id, title, url, feed, date, category count, annotation length, summary first 200 chars), then analyzed feed distribution first before sampling titles. The biggest structural finding: 10,991 of 11,000 items are from a single month (March 2026), with only 9 items from earlier dates. There are zero annotations across the entire dataset. Custom tags do not exist — categories contain only system states (reading-list, starred, fresh, read). This is not a curated personal library built over years; it's a snapshot of one month's active reading in Inoreader, where starred/saved = "I saw this while reading." I sampled all specialty and niche feeds in full, systematically sampled the major news feeds in batches of 20-30, and relied on feed names plus title patterns for the bulk classification. Where a feed title alone didn't disambiguate topic, I read 20-30 item titles from that feed to verify. The main limitation: there's no behavioral signal stronger than "he subscribed to this feed and saw this article." What I cannot know is which items he actually read vs. which appeared and were swept into favorites via Inoreader's behavior.

Cluster overview

|---------|-------------|--------------|----------------|-------------------|--------------|

Signal scoring: since there are no annotations, signal comes from (1) feed specificity — deliberately subscribing to a niche feed means something more than clicking into a news item, (2) concentration — one or two feeds dominating a cluster suggests intentional tracking rather than incidental exposure.

High-signal clusters (detailed)

Cluster 1: Spirituality, Astrology & Faith

What's in it: 65 items from Thought Catalog covering zodiac sign readings, divine timing, scripture-adjacent reflection, tarot, guardian angels, and faith in uncertainty — almost no overlap with the rest of the reading diet.
Count: ~65 items, almost entirely from Thought Catalog, March 2026
Source pattern: Single-feed concentration. Thought Catalog is the only spirituality source in the entire dataset. This wasn't a general news spillover — he subscribed to Thought Catalog specifically.
Annotation density: Zero. But the feed specificity is the signal.
Exemplars:

1. "The Missed Exit Theory: What's Meant For You Will Find You" — Thought Catalog

2. "What Your Guardian Angel Needs You To Know This Week (Tarot Reading)" — Thought Catalog

3. "If God Feels Quiet, Read This" — Thought Catalog

4. "I Found That If I Pray For God To Move A Mountain, I Must Be Prepared To Wake Up Next To A Shovel" — Thought Catalog

5. "What Ecclesiastes 3:1 Teaches Us About Divine Timing In Love, Career, + Other Blessings" — Thought Catalog

Proposed content series: "The Inner Operating System"

- Piece 1: What astrology gets right (and why calling yourself a skeptic is the tell) — the psychology of pattern-seeking in identity systems, whether zodiac or Myers-Briggs or creator archetypes

- Piece 2: Divine timing is just compounding — a faith framework for people who can't stop optimizing — the tension between effort culture and surrender, drawn from the "missed exit theory" pattern

- Piece 3: What the God-entrepreneur convergence actually means for your audience — why faith-adjacent messaging resonates in creator culture, what it signals about where the audience is

Why this fits the positioning: It's the only cluster in this dataset that's private rather than professional. That makes it real. Brandyn reads this content while consuming business, tech, and news — the coexistence is the story. A creator talking honestly about inner navigation amid external optimization cycles has a lane that almost nobody owns.

Cluster 2: Writing & Publishing Industry

What's in it: Jane Friedman's blog (industry news for writers and publishers) — new imprints, AI certification debates, ebook distribution, publishing conferences. Adjacent items from other feeds: AI copyright lawsuits (Encyclopaedia Britannica vs OpenAI), authors' rights.
Count: ~40 items; Jane Friedman (9 articles) anchors it; publishing-related items from WIRED, Scientific American, Fast Company add ~30 more
Source pattern: Jane Friedman is a deliberate, specific subscription — it's not a general media outlet, it's a trade blog for publishing professionals. Every save from that feed is intentional relative to saves from, say, BBC News.
Annotation density: Zero.
Exemplars:

1. "My Concerns About the Authors Guild Human Authored Certification—and Their Comprehensive Response" — Jane Friedman

2. "Free conference on AI and publishing" — Jane Friedman

3. "Ebook distributor Bookwire and Eleven Labs partner up" — Jane Friedman

4. "Encyclopaedia Britannica is the latest giant to sue OpenAI" — Fast Company

5. "This Bill Would Force AI Companies to Disclose Copyrighted Works" — PetaPixel (cross-cluster with photography)

Proposed content series: "What AI is Actually Doing to Publishing"

- Piece 1: The 'human-verified' label and why publishers are scrambling to matter again — the Authors Guild certification, what readers actually think, whether provenance labels will become a market signal

- Piece 2: From RSS to AI summaries: how discovery for books is collapsing — Jane Friedman's newsletter readership, ebook discovery, AI overview impact on book search, what the Bookwire/ElevenLabs partnership actually signals

- Piece 3: The copyright lawsuit scoreboard: who's winning, who's delaying, who's settling — tracking AI/content company lawsuits as a lens on how the industry is valuing human writing

Why this fits the positioning: Brandyn is building in the creator space while AI reshapes content economics. Publishing is the canary. Audiences who create newsletters, books, or video essays have immediate skin in these questions. He's reading about it already; he just hasn't made it public.

Cluster 3: Marketing, SEO & AI Search

What's in it: A small but highly specific cluster from Moz, SEMrush, and MarketingProfs tracking the breakdown of organic search under AI overviews, zero-click results, and AI-generated content.
Count: ~80 items (Moz 3, SEMrush 4, MarketingProfs 9, plus marketing-adjacent items from Inc/Fast Company)
Source pattern: Three professional trade tools feeding the same question: what happens to creators and marketers when Google is no longer a reliable distribution channel?
Annotation density: Zero.
Exemplars:

1. "How Marketers Win Visibility in the Age of Zero-Click Search and AI Overviews" — MarketingProfs

2. "How To Create a Defensive SEO Strategy for AI Search" — Moz

3. "What Do Tech Mad Cow Disease and AI Search Have in Common" — Moz

4. "WebMCP: What It Is, Why It Matters, and What to Do Now" — SEMrush

5. "Search Has Changed. And So Have We." — SEMrush

Proposed content series: "Search Is Dead. Now What?"

- Piece 1: The zero-click creator: what you lose when Google stops sending traffic — concrete data on AI overview adoption, what content categories are most exposed, how newsletter-first creators weathered Google before

- Piece 2: Defensive positioning for content creators in the AI search era — what Moz/SEMrush recommend, translated for a non-SEO creator audience; which distribution channels are AI-resistant

- Piece 3: WebMCP and the next layer of AI infrastructure — what Brandyn's SEO-fluent audience needs to know about how AI agents will cite and surface content going forward

Why this fits the positioning: This is professional-grade knowledge translated into creator language. Brandyn's audience builds content-dependent businesses. The SEO collapse is their problem too.

Cluster 4: Longform & Cultural Essays

What's in it: Longreads.com, Brain Pickings/The Marginalian, Open Culture, Fresh Air, Stuff You Should Know. Personal narrative essays, philosophical reflections, cultural history deep-dives, and audio-native longform.
Count: ~80 items across those feeds
Source pattern: Four distinct curated sources — all known for editorial selectivity, all publishing slowly and deliberately. This isn't news reading; these feeds push articles worth saving.
Annotation density: Zero. But the feed curation does the work here.
Exemplars:

1. "You Can Just Do Things" — Longreads.com

2. "Competitive Scrabble Is A Lexical Shitshow" — Longreads.com

3. "My Dad Made the Biggest Jewelled Egg in the World. The Obsession Would Destroy his Marriage, Family and Fortune" — Longreads.com

4. "When Things Fall Apart: Tibetan Buddhist Nun and Teacher Pema Chödrön on Transformation Through Difficult Times" — Brain Pickings

5. "Pi and the Seductions of Infinity" — Brain Pickings

Proposed content series: "How to Read Like You Mean It"

- Piece 1: Why you should subscribe to Longreads and three other things that publish slowly — a curator's case for slow reading as a competitive advantage for creators; what Brandyn actually saved and why

- Piece 2: The "you can just do things" essay is overquoted. Here's what people miss. — taking a viral idea and adding friction and specificity back; this is the curatorial positioning play

- Piece 3: Brain Pickings and the problem with beautiful thinking — what Maria Popova gets right and where the aesthetic-over-application trap lives for creator audiences

Why this fits the positioning: Brandyn saves both Longreads essays and zodiac readings. He's not purely a business optimization reader. A content series grounded in the actual reading life — messy, eclectic, private — is different from what everyone else builds out of business content.

Cluster 5: Photography & Camera

What's in it: PetaPixel and Fstoppers — gear reviews, technique tutorials, copyright and AI debates in photography, the business of photography as a profession, and how Instagram and AI tools are changing the visual economy.
Count: ~103 items across PetaPixel (61) and Fstoppers (42)
Source pattern: Two dedicated photography trade publications — not consumer reviews, not general tech. PetaPixel covers both technical and business-of-photography angles.
Annotation density: Zero.
Exemplars:

1. "Is the Camera Industry Pricing Out Beginners?" — PetaPixel

2. "How to Write a Photography Blog That Actually Drives Bookings (in About an Hour a Week)" — PetaPixel

3. "Stark and Grainy on Purpose: One Photographer's Case Against Straight Landscape Photos" — PetaPixel

4. "This Bill Would Force AI Companies to Disclose Copyrighted Works" — PetaPixel

5. "Instagram Begins Testing Clickable Links in Post Captions" — PetaPixel

Proposed content series: "The Visual Creator's Dilemma"

- Piece 1: When gear pricing gates out the next generation of visual creators — the PetaPixel framing, but applied to video/photo creators as a class rather than photographers specifically

- Piece 2: Intentionally imperfect: the aesthetic case against AI-polished visuals — "Stark and Grainy on Purpose" as a starting point for discussing what authenticity looks like in an era of generative polish

- Piece 3: AI copyright for creators: what the photographers are figuring out that everyone else isn't yet — photography is 5 years ahead of video on copyright fights; what creators should watch

Why this fits the positioning: Photography is Brandyn's most consistently tracked niche interest (2 dedicated feeds, 103 items). He's reading at the craft level, not the consumer level.

Low-signal clusters

These clusters had either too many sources, too much news-cycle content, or too little feed specificity to produce a reliable content series signal.

UK/World News & Politics (~7,200 items): Independent, Google News, Guardian World, BBC, NYT. General news reading across 5+ major outlets. No discernible topical concentration.
AI Industry & Policy (~800 items): 2,516 items in the full dataset contain AI in the title (23% of all saves). Almost entirely distributed across major news feeds — not research sources or technical publications. The AI saves are cultural tracking, not practitioner depth.
Film, TV & Entertainment (~550 items): /Film, BBC Entertainment, Guardian Culture, NYT Arts. Heavy Oscars 2026 coverage, streaming show discussions, franchise analysis. No Letterboxd-style depth. All March 2026.
Consumer Tech & Gadgets (~350 items): MakeUseOf, Engadget, Verge, WIRED. Mix of Windows tips, audio gear reviews, Android tweaks. Practical how-to reading.
Business Strategy & Entrepreneurship (~350 items): Inc, Fast Company, Entrepreneur. Inspirational-leaning business content with some AI ROI pieces. Little that anchors to a clear niche.
Financial Markets & Economy (~150 items): FT Markets. Iran war + oil prices dominate March 2026. Context-specific, not an investing series signal.
Reddit & Social Noise (~689 items): Reddit front page and hot topics search feed. Relationship advice, gaming megathreads, financial tickers, personal questions. Complete noise.
Design & Visual Arts (~111 items): Creative Bloq, Behance, Creative Review, Colossal. Visual interest is there but the content is industry news and project showcases — not practitioner-level debate material.
Science, Space & Climate (~130 items): Scientific American, New Scientist. Genuine intellectual reading, but generalist; no angle anchors it to Brandyn's professional positioning.
Health & Medicine (~70 items): Sub-cluster of the science feeds. RFK, vaccines, neuroscience. Current-events-driven.

Cross-cluster surprises

1. The spirituality cluster is the most anomalous signal in the dataset.

65 Thought Catalog items about zodiac signs and divine timing sit alongside 7,000 news items and trade publications. This isn't background noise — it's a specific subscription to a specific source. Everything else in the dataset has a professional or general-curiosity explanation. The Thought Catalog feed does not.

2. There is no Substack in this dataset.

For a creator who presumably operates in or around the newsletter/creator economy, the complete absence of Substack writers, independent newsletter subscriptions, or indie blogger feeds is notable. The feeds are entirely major media, trade publications, and Reddit. Either his Substack subscriptions live in email rather than RSS, or his reading diet is more traditional media than the creator economy aesthetic would suggest.

3. AI coverage is huge in volume and shallow in depth.

2,516 items — 23% of all saves — include AI in the title. Almost none of them are from AI-specific research sources, technical blogs, or practitioner publications. The AI reading is all cultural/business impact, filtered through mainstream media. He is tracking AI as a social force, not as a practitioner building with it.

4. The FT Markets feed (111 items) is the only financial-specialist feed.

Outside general business media (Inc, Fast Company), Financial Times: Markets is the only source with a specific financial lens. That's a relatively deliberate subscription. Most of the March 2026 content there was oil prices and the Iran war — but the subscription predates the news cycle, which suggests he was reading FT Markets before the crisis.

5. Photography and design read as actual craft interests, not trend-monitoring.

The camera gear saves are technical and opinionated ("Best Fujifilm Sensor for Black and White? One Photographer Tried Almost Every Single One"), not just product launches. The design saves include font discussions and branding debates, not just inspiration galleries. These are practitioner-level engagements, which is a different quality of attention than the news feeds.

What you couldn't see

The 0-annotation finding is the biggest limitation. In a normal Inoreader export with years of careful curation, annotations would distinguish "I saved this by accident" from "I starred this because I want to write about it." Here, both behaviors look identical. The 11,000 items might include a significant fraction that were auto-starred or saved in bulk during a feed setup.

I could not assess save rate per feed. Knowing that Brandyn saved 61 out of 300 PetaPixel articles published in March 2026 would be a different signal than 61 out of 61. I don't have the denominator.

I read every item from the small specialty feeds (Thought Catalog, Jane Friedman, Moz, SEMrush, Longreads, Brain Pickings) and sampled 20-50 items from feeds with 50-200 items. The major news feeds (Independent, Google News, Guardian, BBC, NYT) I sampled in batches of 30 and classified by topic pattern.

The date concentration in March 2026 makes time-pattern analysis meaningless — everything is bursty and everything is from the same month. Whether any of these interests are "steady over years" or "bursty research sprint" cannot be determined from this export.