Why the next card changes: sources & model

Open two pretend feeds. Watch a card, change a rule, and pull out the receipt for the next recommendation. Discover a feedback loop without handing over any real viewing history.

What supports the explanation?

What this model assumes

1. This is a finite authored content-based scorer and MMR/Jaccard adaptation, not a reconstruction of a named social platform.
2. No real account, personal viewing history, inferred interests, or user satisfaction model is used.
3. Watch is an explicit experiment action; no real video is played or completion measured.
4. The replay always chooses the middle suggestion by an authored rule, not a simulated human preference.
5. Tag diversity is not a measure of fairness, happiness, scientific quality or safety.
6. The separate source aggregate cannot evaluate our invented recommendation policy.
7. The phone teardown shows conceptual software stages, not physical device internals.
8. Changing a uniform starting-history preset resets its run; ordinary rule changes preserve past action records.
9. Content-based recommendation: Item features and history features are compared directly. Here the features are authored numeric tags, not extracted image content. Other systems can learn relationships across many users; this model does not perform collaborative filtering.
10. Normalization: All card vectors are nonnegative and sum to two. Counters stay positive, so dot(p,x)/(2 max(p)) lies between zero and one. The denominator is an authored scaling convention, not empirical probability calibration.
11. Candidate generation and ranking: These are separate steps. A dated 2016 YouTube engineering paper describes such a split, but it is not a complete specification of today’s system. Our finite shortlist illustrates the distinction with fully visible rules.
12. The MMR adaptation: Carbonell and Goldstein introduced Maximal Marginal Relevance in 1998 for retrieval/summarization. Our adaptation uses q = λ × match − (1 − λ) × maximum prior-selected tag overlap. The first penalty is zero. These teaching weights are not platform settings or universal recommendations.
13. Jaccard overlap: Count the shared nonzero topic coordinates and divide by the size of their union. A and B overlap by 1; A and C by 1/2; C and G by 1/3; A and D by 0. This is overlap in coarse authored tags, not a complete measure of semantic similarity.
14. Exact decisions, rounded labels: The implementation compares rational score numerators by cross multiplication. Rounded displayed values never decide ties. Negative selection scores remain valid; they do not mean dislike or harm. At λ = 0, first-slot scores all tie and the declared ID rule chooses the first card.
15. A visible tradeoff: Initially, slate A/B/C has mean match 11/12 and mean pairwise tag overlap 2/3. A/D/E has mean match 7/12 and overlap zero. Lower redundancy is not automatically higher satisfaction, fairness or truth. The intended task matters.
16. Implicit feedback has limits: A watch is a behavioral record. No watch may mean no exposure, no opportunity, a different task or an actual rejection. The source research on satisfaction and response bias illustrates why an engagement measure cannot simply be renamed satisfaction.
17. Feedback can influence later data: Chaney and colleagues’ 2018 simulations studied learning from behavior already influenced by recommendations. Their modeled homogeneity and utility results are bounded by their simulation. Our four-round example is separate and does not prove universal effects on real people.
18. Multiple objectives and history: Historical engineering papers describe multiple ranking objectives, not only one engagement counter. Content-based matching, collaborative approaches, learned ranking and reranking coexist; a timeline should not pretend each new method replaced everything before it.
19. A real log answers a particular question: The pinned Open Bandit small release contains 10,000 displayed-fashion-item records and 38 clicks. The click fraction is 0.38%. The records date from November 24–30, 2019. They are neither unique-person counts nor learning outcomes, and are not used to fit this lesson’s scorer.
20. Two different probabilities: The source file logs item-selection propensity_score = 0.0125 for these rows. That logging-policy probability is different from the observed click fraction 0.0038. A policy probability, a measured outcome fraction and our authored match score must retain their different meanings.

What has been checked

Analytical reference cases, conservation or transition invariants, finite drawing commands, bounded setup parsing, discovery and route integrity are checked automatically. These checks do not establish anatomical fidelity, learner outcomes or browser/device compatibility. Independent subject review, learner trials, comprehensive accessibility review and browser video encoding checks remain pending.

Each source supports the associated claim. Sources do not certify this implementation or its visuals.

About the cover illustration

Original unbranded 3D phones displaying the actual fictional recommendation slates and topic counters. A content-based scoring and reranking experiment, not a real account or physical software machine.