Side Quests: strong connections into genres a user has not rated highly

Decision 09

A side quest is a show a user would plausibly like but would not have reached for, and surprise only means anything measured against what someone has demonstrated. The row is built from that user’s own highly-rated shows, walks two hops out at a discount, and ranks by strength times novelty times how few of their favorites reach it.

Context

Side Quests shipped as a stub. It rendered for signed-in users only, it was always empty, and its copy promised it would “unlock once you’ve rated a handful of shows”. #10 asked for it to load even for a user who had rated nothing.

That was the easy half. What the row is for was harder.

I decided that a side quest is a surprise: a show the user would plausibly like, would not have reached for alone, and that lands in a genre they did not expect to like. The closest analogy is ordering an unfamiliar dish at a restaurant on the chance it is good. Not random, because random is noise. Not popular, because everyone already knows what is popular.

Two existing rules bound the answer. Nothing in TVLens is ever a popularity chart (ADR-05), which rules out the easy fillers. And ADR-08 fixed the shape of personalization: Layer 2 re-ranks Layer 1’s stored list per user and never scores shows a second time.

How this decision changed: surprise needs an expectation to violate

The first version of this row shipped with a cold-start path: a visitor with no ratings, or no account at all, saw the catalog’s strongest cross-genre edges. That path is now removed, for two reasons that only became visible with the row on the page.

A global row is surprising relative to nothing. Surprise is not a property of a show. It is a relation between a show and what someone expected, and a person who has rated nothing has demonstrated no expectation. What the cold-start row actually produced was one fixed list, identical for every visitor, with nothing in it derived from the person looking at it. That is the thing this project keeps refusing (ADR-05, ADR-07, ADR-08): not because it ranked by popularity, it did not, but because a single list served to everyone is a chart whatever it is sorted by.

“The catalog’s strongest cross-genre edges” is not the fringe. The strongest of them scores inside the top 10 percent of the whole graph. Unexpected in category, but the opposite of peripheral. Sorting the cross-genre subset by strength finds the loudest edges that happen to cross a line, not the quiet ones.

This also overturns the literal ask in #10, “they should load even if the user has not rated any shows”. Deliberately, and my own call. I wrote #10 during a demo when the row was an empty stub with no definition; once the definition is surprise measured against demonstrated taste, loading the row for a user who has demonstrated nothing is not a feature, it is a contradiction. The row is now gated, and the page says so in as many words.

The mechanism survived this. A side quest is still a strong Layer 1 edge reaching somewhere the user has not been, and it survived the second revision below too; what changed both times was who the row is for and how it is ordered. The file name stays across both, so every reference to this record from the code and from the other records still resolves.

How it changed again: the row was the recommendation row wearing a different title

The first amendment fixed who the row is for. It left the ordering untouched, and with the row on the page that turned out to be the larger problem: the row was returning the top recommendations of the shows I had rated, when what it should return is edge cases still bound by those recommendations, a second-degree connection rather than a first. The cause was structural rather than a tuning miss, and there were two faults in it.

The candidate pool was the recommendation pool. Walking ranks 0 through 5 out of the seeds is what a “more like this” row shows. Surprise was then applied only as a re-sort over that pool, so the row could re-order those shows but could never leave them. Measured on the development database with three seeds: 11 of 11 picks came from the first-degree pool. The overlap was 100 percent by construction, not by coincidence. Graph distance was not one of the row’s axes; it was a constant, pinned at its minimum.

Strength drowned novelty. The order was score x novelty, and the docstring claimed “both halves have to be there”. They did not have comparable range. Across one real candidate set, Layer 1 scores ran 0.42 to 5.76, about 14x, while novelty is a share and spans at most 5x in practice. The bigger term decided the row. The lead pick, Person of Interest, had a novelty of 0.25 and sat first purely on a 5.76 edge: the top “surprise” was the top recommendation.

What changed

Strength is log-compressed. log1p(score) puts strength on novelty’s order of magnitude, which is the only condition under which multiplying them means anything. This is the smallest of the three changes and it fixes the symptom that was actually visible.

The walk goes a second hop, at a discount. A two-hop path scores at its weakest link, multiplied by SIDE_QUEST_HOP_DECAY, so distance is earned rather than assumed. Shows the user has already watched stay in the walk as bridges even though they can never be picks: a show the user has seen is a real shared-people connection, and treating it as a dead end throws away the graph’s most reliable edges.

A show that many seeds reach sinks. Divided by (seeds that reached it) ** SIDE_QUEST_CENTRALITY_EXPONENT. This is the term that most directly answers “edge cases bound by our recommendations”: a candidate every one of a user’s favorites points at sits at the center of that taste, and centrality is measurable where peripherality was previously only implied.

What it did

Same user, same three seeds, before and after:

Three of twelve picks are now shows a recommendation row would never have surfaced, and the strongest edge in the pool no longer leads simply because it is the strongest edge.

What this change deliberately does not settle

I fitted every constant here against a 100-show catalog in which two hops already reach 43 percent of everything. SIDE_QUEST_HOP_DECAY = 0.5 is a round number chosen where hop-2 candidates were plentiful, and whether it is too harsh or too generous is not answerable from this data. Deferred to #20, together with a limitation the same measurement exposed: novelty is binary, so a user with no Crime rating sees every Crime-tagged show score as fully novel, and a seed whose neighborhood is one cluster can fill most of the row with that cluster. Genre affinity has more resolution than has-genre / has-not, and Layer 2 already keeps a signed number per genre.

Decision

Note, 2026-08-26. ADR-04’s rescoring tilted the hop decay toward novelty, and the constant stays at 0.5, see The hop decay is tilted, and left alone.

A side quest is a strong Layer 1 connection, out of a show this user rated highly, that lands in genres this user has never rated highly and that few of their favorites point at. Four parts, in the order the code applies them.

Seeds. The shows the user rated at or above 4.0. That is the same “high” line the home page already uses to glow a favorite genre and that ADR-08 personalizes from, so every feature on the page agrees about what liking a show means.

The gate: three seeds, or the row stays locked. Below three the row does not render at all, and a signed-in user sees one line of copy telling them how to unlock it. One seed says which show a person watched, not which genres they lean toward; nothing can be measured as surprising against a single point. Three is the smallest history in which a genre can be seen to recur, and therefore the smallest history in which its absence means something. Anonymous visitors get no row and no copy: they cannot rate anything without an account, so the unlock instruction would be a dead end.

The walk: the strong half of each seed’s stored list, and then the strong half of the lists belonging to what that reached. Ranks 0 through 5 of the 12 Layer 1 keeps per show (ADR-07), followed one hop further out. The rank cap holds at every hop, because a side quest has to be a confident connection and the weak tail of a list is mostly coincidence, so a strange genre found down there is noise wearing a surprise costume. But one hop is exactly the seed’s own recommendation list, so a one-hop row can re-order that pool and never leave it. The second hop is what gives distance somewhere to vary. A show the user has already watched is never a pick and always a bridge: it is a real connection, not a dead end.

Surprise: the share of a candidate’s genres the user has no positive history with. Collect the genres of the seeds; call that the demonstrated taste. For each candidate, novelty is the number of its genres outside that set divided by the number of genres it has. A candidate with novelty 0 is not a side quest, and the walk drops it however strong its edge.

The order is strength x novelty x centrality. Strength is log1p of the reaching edge’s Layer 1 score, taken at the weakest link of the path and decayed once per extra hop. Novelty is the share above. Centrality divides by the square root of how many seeds reached the candidate. All three have to be present: a blockbuster edge into more of the same sinks on novelty, a thin edge into a strange genre sinks on strength, and a show every favorite points at sinks on centrality. The log is not cosmetic. Raw Layer 1 scores span more than an order of magnitude while novelty is a share bounded at 1, so a plain product let the strongest edge in the pool win on strength alone, which is precisely how the row came to read as a recommendation list.

The only new number is a multiplier on Layer 1’s own score. Nothing scores a show twice and no second engine runs over the catalog, which keeps this inside ADR-08’s rule that Layer 2 re-ranks rather than re-scores.

Shows the user has already watched never appear, watched covering rated (ADR-08), and neither does anything the home page has already used in Top Picks.

Three render states, and the third one matters. Cards when the walk found something; the locked copy when the user is under three seeds; and nothing at all when an unlocked user’s own seeds happen to reach nothing new. Showing “rate three shows to unlock” to someone who has already rated three shows would be a lie, and it happens to about 3 percent of users in simulation, so it is a real state rather than a theoretical one.

The dedupe chain is unchanged: Top Picks, then Side Quests, then Recently added. One show, one row. The priority is by how personal a row is, and it deliberately does not match render order: Side Quests sits below Recently added on the page but claims its shows before it. A show only ever falls down that chain. Watch Next has since joined the chain (#24), claiming after Side Quests and before Recently added; the comment in views.index records why it yields to a row with the smaller pool.

What the numbers established, and what has since moved

Measured in August 2026 against 4000 synthetic three-seed users on a 100-show catalog, before the genre gate was graded and before the catalog was filtered and pruned to 248. The figures are gone; three findings from them still hold and are the reason this row exists:

• It is personal, not global. Two random users’ rows overlapped by a mean Jaccard of 0.08, median 0.00. Under the withdrawn cold-start path that number was 100% by construction.
• It is not a chart. Mean Spearman correlation between a row’s order and its members’ popularity rank was 0.156, and 0.028 against vote_average. Both noise.
• It is honestly short. Median 6 picks of a possible 12, and it filled completely for 3% of users. A half strip is the truthful answer when the catalog is small.

Re-measuring against the current rule and the current catalog is worth doing before the demo.

Tags exist now, and would still sharpen this

When this was written Tag and ShowTag held 0 rows, so genre was the only categorical signal. ADR-14 landed tags, which held 7 tags and 9 applications at the time, enough to exist and not enough to run novelty over. Genre remains coarse for this job. Once tags are populated, the same shape works with a finer vocabulary: novelty would run over tags as well as genres, letting the row tell “a workplace comedy you have not tried” from “a comedy”. Still a later decision.

One hazard this touched

Show.Meta.ordering was ["-popularity"] at the time, which meant any queryset that forgot an explicit order_by silently became the popularity ranking ADR-05 forbids. Every queryset on the Side Quests path is explicitly ordered and says so in a comment. The default has since changed to ["name"], deliberately; the comment on Show.Meta records the reasoning.

The hop decay is tilted, and left alone

ADR-04 raised each shared person’s episode share to INVOLVEMENT_EXPONENT = 1.375 before summing. That changes the input to math.log1p(score), and this note records the effect so it is not rediscovered later as a mystery.

log1p is concave, so it compresses large scores while the exponent crushes small ones. On the rank<=5 edges the walk actually uses, the maximum strength falls by 3.4% while the median falls by 43%. Strong edges keep their strength and weak edges lose theirs, so a two-hop path through two strong edges competes better against a one-hop path through a weak one than it used to.

Measured on the rebuilt catalog, sampling from that pool: the probability that a two-hop path out-scores a one-hop path rises from 22.5% to 26.1%, a factor of 1.16. Restoring the old balance would mean SIDE_QUEST_HOP_DECAY = 0.33 rather than 0.5.

Left at 0.5 deliberately. This ADR already flags the constant as needing a re-fit at real catalog scale, tracked in #20, and it was tuned against a 100-show catalog that no longer exists. Re-fitting it inside the rescoring change would mix two adjustments and make neither measurable. Side Quests was also expected to give way to Watch Next as the homepage row. Watch Next has since shipped (#24) and leads the page, but Side Quests stayed below it, so these constants are still live.

The direction of the tilt is toward novelty, which is the axis this row exists to serve, so 3.6 percentage points is not a regression. It is written down here so the next person to touch the constant knows part of the drift came from Layer 1 rather than from this row.

[ADR-04](/decisions/04-episode-weighted-people-recommender/) records the same measurement from the other side.

After Action Review

shows/tests.py::SideQuestsTests freezes the definition:

• the row locks below three high ratings
• anonymous visitors get neither the row nor the copy
• every pick lands in a genre the user has never rated highly
• the walk refuses the graph’s strongest edge when it is more of the same
• distance can beat a stronger edge and cannot win on its own
• the walk covers only the strong half of a seed’s list, and only the user’s own favorites
• an unlocked user with nothing new gets no section rather than the locked copy

SideQuestsRankingTests freezes the second revision:

• a blockbuster edge no longer outranks a novel one, while strength still separates two equally novel shows
• a show two hops out can be a pick, and loses to an identical show one hop in
• a watched show is never a pick and still carries the walk
• a show every seed reaches ranks below one that only a single seed found
• the surprise arithmetic appears once, written out against a known pick

Amendment, 2026-08-27: the genre gate is graded, not binary

demonstrated was a set, so one seed was enough to disqualify a genre forever. Measured against the live catalog with ten seeds:

One show tagged Western removed every Western in the catalog with the same force as nine Drama seeds. The cost was not marginal: 82 of 104 reachable candidates were dropped, and what survived was ten comedies out of twelve, not because comedy suited the reader but because comedy was what was left. The rule also got worse with use, since every genre a reader demonstrated was removed permanently.

Familiarity is now a share of the seeds rather than a membership test, and novelty is the mean of 1 - familiarity across the candidate’s genres, raised to SIDE_QUEST_GENRE_EXPONENT.

The hard drop is gone, but its behavior is not. A candidate whose every genre sits at familiarity 1.0 scores exactly 0.0 and can never be chosen, so the old cliff survives as the limit case of the formula rather than as a branch beside it.

Choosing the exponent

Grading the gate pulls this row toward Watch Next, which is what the exponent exists to control. Measured across seven values on the real catalog:

Below 1.5 the surprise row fills with the recommendation row. Above 2.0 it collapses back to the comedy list it started as. 1.5 was chosen: 7 comedies of 12 against 2.0’s 8, the same genre spread, and it is where a genre brushed once can reach the row again. Criminal Minds and The Boroughs, both Mystery at 1 of 10 seeds, enter at this value and at no higher one.

Rejected the same day: removing the gate entirely

Letting distance carry the row alone was measured and rejected. Without the genre term the top seven were MobLand, Lioness, Hawaii Five-0, Law & Order: Criminal Intent, Lanterns, Law & Order: SVU and Law & Order. Distance in this graph correlates with sharing people, and people cluster by genre, so walking further returns a different show rather than a different kind of show. Dropping the gate did not make the row surprising. It made it safe.

Read this record in the tvlens repo →

← All decisions