# Game mechanics and game-feel analysis

Review date: 10 November 2026.

## Scope and how to read the findings

This analysis covers **all 126 games in the frozen storefront inventory captured on 8–9 November 2026**. It builds on the repository's individually summarized rules, screenshot evidence and prior demo inspection notes. It is a new analysis of those records, **not a new playthrough**. The 89 catalogue-only identities are outside this scope.

The records contain terminal demo findings for 125 titles. Lykkerunden remains unavailable and its mechanics can only be discussed from its guide. A recorded terminal does not verify every manual input, rare effect, recovery path or bonus.

The [complete per-game catalogue](game-mechanics-catalogue.md) contains the rules, interpretation, selected prior observations, proposed feedback and limitations for each title. The [family index](game-mechanics-index.md) lists every game once, including its subtype. [Structured classifications](mechanics-classifications.json) preserve the same distinctions for further analysis. The [searchable mechanics catalogue](mechanics-review/index.html) filters games by family, mechanic tag, art style, provider, complexity, bonus and Auto, with screenshots and evidence links.

Three kinds of statements remain separate throughout:

- **Recorded observation:** an input or state change documented in the prior inspection, with linked evidence. This is not a new measurement.
- **Rule description / interpretation:** guide-described conditions and analysis of how the loop is structured. Unexercised branches remain provisional, particularly where guide identity is uncertain.
- **Game-feel proposal:** a possible improvement to test. It is not a claim that the effect is already present, absent or poorly implemented.

The game-feel skill and its feedback recipes inform the proposals: acknowledge actions, layer a few short responses, ease motion, scale intensity to importance, return to rest and preserve input responsiveness. The game implementations and their event hooks were not available for inspection. No polish was implemented.

## Main findings

The portfolio has much more mechanical variety than its repeated themes and reward tables suggest. However, **the most useful distinction is what the player does and what that action changes**, not whether the screen contains fruit, birds, jewels or pirates.

Many games are reveal systems with different qualification rules. Others present constrained spatial actions: swaps, group selections, merges, route drawing, placement, aiming or code hypotheses. A third layer consists of counters, resource budgets, recovery mechanisms and closing phases. This secondary layer often creates more of the game's identity than the basic input.

The presence of puzzle choices does not establish control over the lottery result. The records explicitly distinguish presentation-layer interaction from outcome determination; this report does not infer skill-dependent winnings from a puzzle-like interface.

### Primary families

Each title has one primary family for counting; subtypes and secondary mechanics remain in its entry. Broad families are not assertions of identical rules or shared implementation.

| Primary family | Games | Important distinction |
|---|---:|---|
| Symbol/number lines, counts and reveal collections | 32 | Lines, recipes, counts, automatic fusions and special stopping rules require different explanations |
| Amount matching and panel reveals | 27 | Most panels score independently; some titles match individual reveals across the ticket |
| Word and letter games | 14 | Crossword counts, independent words, name counts, whole-word search and word construction |
| Reference matching | 10 | Match a supplied reference, not another target cell |
| Adjacent-swap match-three | 8 | Collection, combat, excavation and obstacle removal differ despite shared swapping |
| Frontier pair removal | 7 | Removing a pair exposes local pieces; access and recovery are part of the loop |
| Guided routes, exploration and chains | 6 | Arrow following, energy-limited movement, row risk, excavation and tracing are separate subtypes |
| Group clearing and column refresh | 4 | Three select existing groups; Plisk & Plask replaces a column |
| Tile or creature upgrading | 4 | Three use pair upgrades; Himlens skatte uses adjacent triples |
| Exposed-tile or small-set pair matching | 2 | Mahjong accessibility differs from five-shell matching |
| Aimed matching projectiles | 2 | Kanonkuglen variants have different counter direction and secondary collections |
| Pegboard drops and landing collections | 2 | Kugleregn batches differ from Spladask's source reveals and row/column goals |
| Eight distinct single-title families | 8 | Minigame sequence, feeding, tray, stacking, code, wheel, routing and target/volley |
| **Total** | **126** | Includes guide-only Lykkerunden |

## 1. Reveal mechanics: small rules can make large differences

### Amount matching

Arme Riddere, Kys frøen, Pausefisk and several related titles repeat independent amount panels plus a separate symbol bonus. The central cognitive rule is **do not pool equal values from different panels**. A symbol bonus is an additional condition, not necessarily a separate playable level.

Millionjagten and Millionjagten XL increase the number of panels from 16 to 25. That increases reveal work and scanning, but does not by itself introduce a more sophisticated mechanic. Their recorded interaction also separates enlarging a panel from partially scratching it and from revealing everything.

Mini Quick and Verdens vildeste polarrejse need different wording: individual reveals are compared across their ticket rather than each being an independent three-value game. The latter also has a three-UniQ-symbol condition. This is why a blanket instruction such as “match within every panel” would be wrong across the whole family.

**Game-feel opportunity:** make the current unit of interaction obvious. Selection, zoom, scratch progress and resolution should each have their own short response. Repeated inputs that expose only a small patch should still visibly register. Routine scratching needs local texture and acknowledgement, not full-screen impact effects.

### Reference matching

Diamant Quick, Sprængfyldt, Vaffelisen, Vaniljekransen, Vejboden and Et godt par compare numbers with a supplied reference set. Blandede bolcher, Kaffepausen, Kaffepausen XL and Lykkesymboler do the equivalent with icons.

The important relationship is **reference → target → associated result**. A repeated number among target cells is insufficient unless it also matches a reference. Some games reveal a target number separately from its amount, creating two distinct input/result steps.

Et godt par belongs here despite its pair-oriented title. Sprængfyldt is a number-matching game; Sprængfarlige frugter is an area-coverage collection game. Similar names are unreliable classifiers.

### Lines, recipes and thresholds

Alfens skatte and Troldens skatte require four-symbol lines; Skovens mystik uses consecutive triples and multiplying wilds. Svampeskoven combines adjacent mushroom groups with a separate cluster-size condition. These are not interchangeable “match three” games, and the player is not necessarily moving the symbols.

Gyldne skinner combines mixed-symbol recipes, orientation-dependent lines and conditional train progress. A line can qualify while every recipe remains incomplete. Eldorado and Kort på stribe score separate boards, while Lykketræf can apply a new reference number retrospectively to earlier panels.

**Game-feel opportunity:** show which rule actually fired. Animate the qualifying cells together, then the corresponding row, recipe or counter. Do not use the same celebration for an exposed amount, partial progress and a completed condition.

### Familiar game imagery does not imply familiar decisions

| Game | What the recorded rules support | What should not be inferred |
|---|---|---|
| XO | Reveal assigned X/O lines and a separate amount panel | Adversarial tic-tac-toe placement |
| Yatzy | Reveal assigned five-die rows and evaluate the highest qualifying combination | Ordinary hold-and-reroll strategy |
| Fodboldfeber | Reveal scores and count qualifying fixture victories | Football movement or tactical control |
| Fuldtræffer | Weapon lines drive attacks and base-damage presentation | Direct aiming or a manual action-combat system |
| Spøgelsesjagt | Search-tile selection triggers a room-search animation | Free character movement through the room |

## 2. Word games: five distinct interaction structures

| Structure | Games | Key rule and feedback need |
|---|---|---|
| Supplied-letter crosswords | Bogstavjagt, rødt ord, stjerne, Dobbelt and Tripel | Count valid completed words separately per board; show reference-to-letter transfer and then word completion |
| Independently rewarded words | Bær og bogstaver, Sommerhaven, Strandfesten, Svampejagt, Snekugle, Tropefugle | Each whole word has its own result; partial coloured letters do not qualify |
| Completed-name count | Dyrevenner | Count completed animal names, not visible characters or partial names |
| Whole-word search | Find ord | Locate supplied whole words only in allowed directions; reversed strings do not qualify |
| Word construction | Ordet rundt | Construct target words; resulting letter colours then fill separate collection meters |

Tripel Bogstavjagt illustrates the main scoring trap: two completed words on each of three boards do not become a pooled six-word result. Ordet rundt illustrates a different separation: constructing all target words can still leave every reward meter incomplete.

**Game-feel opportunity:** keep text and boards stable while acknowledging changes. Highlight the newly supplied letter, its affected locations and completed words in a readable sequence. A single-letter hint, an automatic letter-transfer helper and continuous Auto need visibly different control states. Large shake, long freeze frames and repeated full-board flashes work against the reading task.

## 3. Matching puzzles: distinguish the player's verb

| Input | Examples | What changes |
|---|---|---|
| Swap adjacent pieces | Drageheltene, Guldåren, Ellens Eventyr | Two positions exchange; resulting lines resolve |
| Select an existing connected group | Boblebasserne, Fugle i Paradis, Magiske elementer | Chosen group clears and replacements enter |
| Replace a column | Plisk & Plask | A selected six-symbol column is replaced; new clusters resolve automatically |
| Merge equal items | Brikjagten, Havets Juveler, Skatkammeret | Sources combine into a higher value or level |
| Place beside equals to form a triple | Himlens skatte | An adjacent creature triple upgrades and can expand terrain |
| Select into a tray | Frugtflip | Exposed pieces occupy limited slots until a matching triple clears |
| Trace a chain | Tropefrugt | Connected equal fruits are selected along a hex-grid path |
| Reveal and let groups form automatically | Nattens drømme, Søde Sager | Reveals trigger automatic fusion/grouping; no manual swapping or merging is established |

### The eight adjacent-swap titles

| Game | Distinctive secondary system | Feedback sequence worth testing |
|---|---|---|
| Drageheltene | Colour-linked heroes, waves and chest effects | Match → relevant hero → attack → enemy/wave change |
| Ellens Eventyr | Cupboard locks, opening, contents and shelves | Adjacent match → lock removal → later opening → shelf transfer |
| Havenissens Baghave | Excavation layers, chests and charged character effects | Match → terrain change → chest → collection |
| Fuglevenner Drømmeland | Covered nests, waking birds and houses | Nearby match → cover removal → waking → house slot |
| Feen i skoven | Symbol collection and conditional heart bonus | Match → remaining-count decrease; explain qualifying hearts separately |
| Guldåren | Collection plus dynamite/stone and lantern rules | Normal collection versus special clear; show why a lantern qualifies |
| Finurlige frugter | Collection, bomb effects and changing double-count fruit | Clear → actual weighted contribution → remaining target |
| Rumekspressen | Collection plus turbo-route effects | Match → route/clear effect → counted symbols |

The first four expose additional spatial or staged objectives beyond reducing a colour count. This makes them useful comparison cases for richer mechanic design, but it does not prove they are better games or more responsive.

### Merging and trays

Brikjagten combines values, colour maxima and row clearing. Havets Juveler can resupply after a no-pair state. Skatkammeret upgrades groups with additional supplies. Himlens skatte combines triple placement with terrain and creature-pair reward conditions. A generic merge animation alone cannot explain all of these systems.

Frugtflip is especially distinctive because visible fruit controls tray matching while hidden colours control reward progress. The recorded apple triple advanced three different colour meters. The tray has seven slots, and board clearance leads into seven closing lucky tiles. This creates three separate questions for feedback: what can be selected, what clears the tray, and what advances the result?

**Game-feel opportunity:** use a short eased source-to-destination movement, a restrained squash/pop on a successful combination and a distinct meter transfer. Invalid interactions should receive immediate local feedback. The visual response must not obscure new pieces or make an unchanged meter look as though it advanced.

## 4. Accessibility frontiers and pair removal

Dybt farvand, Fuglevenner, Hajfyldt farvand, Hulens Hemmeligheder, Lykkeposen, Sommerfuglens Vingesus and Stenmaskernes Hemmelighed share a recognizable architecture: expose an accessible frontier, remove matches and unlock local pieces behind it.

This is not automatically a classic memory-card game. The important state is which objects are accessible, which remain exposed, which are blocked and what recovery can do when the frontier has no match. The records repeatedly show local unlocks rather than an entire new row becoming available.

Attached amounts are particularly easy to misread: a visible amount on an unmatched fish, butterfly or mask can remain unpaid. Recovery devices—reserves, special characters or transformations—also differ by title and many were not manually exercised.

Mahjong adds a different accessibility rule: an identical tile must be exposed and have a free side. Havets hemmeligheder is a small five-shell reveal set, without the same deep frontier structure.

**Game-feel opportunity:** pair selection → paired departure → local unlock → qualifying amount transfer. Make accessibility readable without relying on brightness or colour alone. Keep recovery activation distinguishable from an ordinary pair and from terminal disclosure.

## 5. Spatial and action-like systems

### Paths are not one mechanic

- **Regnbueguld:** follow revealed arrows from supplied starts until a goal or stop.
- **Fjollede kryb:** supplied coordinates start arrow chains, with bombs exposing neighbouring cells.
- **Den lille larve:** select reachable leaves under an energy budget; length and berry conditions add progression.
- **Det Skjulte Tempel:** choose one cell per row, track hearts and potentially reach a closing wheel.
- **Guldjægeren:** choose exposed excavation cells with a limited number of attempts and area effects.
- **Tropefrugt:** trace equal adjacent fruits; its nineteen-fruit bonus threshold is per turn, not a cumulative total.

These require different explanations of choice and cost. Reachability, a prescribed arrow route, and an equal-symbol chain should not share an undifferentiated “click a path” tutorial.

### Aiming, drops and placement

Kanonkuglen and Kanonkuglen 2 present aimed matching projectiles, but their counter semantics differ: the first tracks remaining-to-fifteen, the second ascending progress toward sixty, with lower collection bins providing another result layer.

Kugleregn releases one-to-five-ball batches, each requiring another input. Spladask combines launches with source-symbol reveals, locked areas and row/column collection. Its guide describes random landing; the player should not be described as selecting the landing location.

Sørøverskatten separates reticle placement from firing a volley. Jungletårnet separates rotation from placement and locked-piece settling. Rigtige Rør separates selecting an endpoint from drawing a non-crossing route. These are precisely the places where a control may look unresponsive if the first action changes selection but does not yet execute the move.

**Game-feel opportunity:** acknowledge selection immediately, preview the intended action and explain commitment. Preserve aim and route readability; any recoil or impact feedback should move visual elements rather than the actual target geometry. Strong effects should not mask whether a shot hit, a path crossed illegally or a piece settled.

## 6. Multi-stage loops and stopping rules

Kodeknækkeren has a code-solving stage followed by reward-ball reveals. The solved code assigns a multiplier, but that multiplier only matters if its particular reward qualifies. De gyldne nøgler switches among control modes, collects keys and finishes at mandatory Treasure Island chests. Keys are progress tokens, not already-awarded money.

Other games use secondary stages without changing the main control as dramatically. The following distinctions matter more than calling all of them “bonuses”:

| Stage type | Examples | Consequence for explanation |
|---|---|---|
| Inline additional condition | A symbol panel beside amount panels | No separate mode need be implied |
| Conditional interrupt and return | Guide-described Feen i skoven, Guldåren, Plisk & Plask branches | Explain what remains in the main round and what resumes |
| Conditional bonus board | Søde Sager | Explain the changed qualification rule, not merely the changed background |
| Mandatory closing stage | De gyldne nøgler; Frugtflip after board clearance | Completing the main task is not final completion |
| Deferred automatic resolution | Boblebasserne tour, Magiske elementer Starstorm, stored birds in Fugle i Paradis | Zero moves does not mean all effects have resolved |
| Loss-conditioned continuation | Bikuben maxi | The bonus trigger differs from collecting a positive bonus symbol |
| Failure-budget stop | Zombie Dans skulls; Det Skjulte Tempel hearts | Objects disclosed afterward must not look newly collected |
| No-available-action state | Pair frontiers and merging games | Resolve any genuine recovery/resupply before declaring completion |

**Game-feel opportunity:** explicitly show “main moves complete—resolving effects,” “bonus,” “returning to main game,” or “round complete” as appropriate. Terminal state is a first-class feedback event, not just the disappearance of the play button.

## 7. Game-feel assessment: what is supported and what remains unknown

The records support concrete state transitions: a swap consumes a move, a pair departs, a local frontier unlocks, a column is replaced, Auto changes to pause, a cupboard loses a lock, or a closing board appears. They do not establish frame timing, input latency, easing quality, sound design, hit-stop duration or shake decay.

Consequently, there is no defensible numerical “juice score” or ranking of the best-feeling games here. Even a detailed screenshot sequence cannot establish whether a click registered promptly or whether the audio reinforced the correct event.

### Proposed feedback architecture

For each mechanic, separate these events before choosing effects:

1. **Input acknowledged:** selected, targeted, held or scratched; no success implied yet.
2. **Action accepted:** valid swap, route, merge, shot or reveal committed.
3. **Local resolution:** symbol exposed, match cleared, hit resolved or object settled.
4. **Progress transfer:** the relevant counter, resource or objective changes.
5. **Condition qualified:** an actual threshold, line or recipe completes.
6. **Phase transition:** bonus, recovery, closing stage or return to main play.
7. **Round complete:** all pending effects and required stages have resolved.

These are proposed logical events, not verified hooks in the existing game code. Some titles combine events; others need several separate steps.

### Layer a small number of channels

| Event importance | Suggested response | Avoid |
|---|---|---|
| Routine selection or reveal | Local outline/press state, short sound if audio is enabled, brief eased settle | Full-screen shake, hit-stop or celebratory reward cues |
| Valid match, merge or pair | Source connection, small scale response, short particles, clear transfer to objective | Several overlapping effects that hide the resulting board |
| Special clear or physical impact | Stronger local burst, optional brief recoil or visual-only pause, followed by readable count changes | Delaying critical inputs or changing simulated target positions |
| Objective completion or stage change | Distinct objective confirmation, clear label and controlled transition | Treating an incomplete meter or exposed potential as completed |
| Final result | Stable, inspectable result and explicit completion state | Making an intermediate count-up or decorative ending look final |

Use ease-out for settling and a restrained overshoot for toy-like pops. Squash/stretch suits bubbles and soft characters better than dense crossword letters or metallic accounting panels. A painterly setting can use subtler movement and material-specific particles while keeping the same event structure; art style and mechanical feedback should be coordinated rather than conflated.

Possible prototype starting points—not observed timings or acceptance standards—are roughly 120–200 ms for a small local settle and 40–60 ms for an optional visual-only pause on an exceptional impact. Tune through playtests. Turn-based reveals often need no pause at all. Effects must return to rest, and overlapping effects must not leave a game slowed, scaled or displaced.

Provide reduced-motion and reduced-flash alternatives. Do not make sound, colour or a transient animation the only explanation of a match or error. These are usability proposals, not a compliance assessment.

### The highest-value clarity improvements to investigate

| Priority | Question to investigate | Why the evidence makes it relevant |
|---|---|---|
| 1 | Can players distinguish a potential amount from a qualified result? | Unmatched frontier objects, incomplete shelves and multiplier labels repeatedly remain unpaid |
| 2 | Is the control single-step, bulk reveal, batch release or continuous Auto? | Find, Fjern par, Skrab alt, Autoskrab and release batches behave differently |
| 3 | Does the player know whether a counter increases or decreases? | Collection totals and remaining targets use opposite directions, even between related titles |
| 4 | Is the first click a selection/zoom or a committed action? | Partial scratching, endpoint selection, reticle placement and pair selection need different acknowledgement |
| 5 | Are terminal and pending-resolution states distinct? | Zero moves can precede tours, storms, stored effects or closing reveals |
| 6 | Are local unlocks and recovery effects legible? | Pair frontiers, excavation and automatic reserves change available actions |
| 7 | Does feedback explain the actual scoring rule? | Candy type versus equal coin values, per-board word counts and per-turn chain counts can be confused |

These are prioritized test questions, not confirmed usability defects. The records show the underlying distinctions; they do not measure human misunderstanding.

## 8. Most informative games for deeper study

These are candidates for learning about mechanic design, not a quality ranking:

| Game | Why it is informative | Most useful next observation |
|---|---|---|
| Brikjagten | Merge values plus colour and row progress | Successful drag, invalid merge response and bonus-row transition |
| Frugtflip | Limited tray plus hidden reward mapping | Full-tray recovery and distinct helper behaviour |
| Drageheltene | Match-three translated into combat | Match-to-hero timing and a verified dragon-defeat branch |
| Ellens Eventyr | Multi-step environmental objectives | Lock/open/collect timing and a completed shelf |
| Havenissens Baghave | Layered excavation and charged effects | Visibility of each layer change and special-match rules |
| Den lille larve | Reachability, energy and growth | A successful manual move, invalid target and bonus trigger |
| Rigtige Rør | Spatial construction distinct from rewards | Manual drawing, redrawing and invalid crossing feedback |
| De gyldne nøgler | Multiple control vocabularies and closing phase | Manual examples of the unexercised tasks and their transitions |
| Kodeknækkeren | Logical guesses followed by separate reveal | Manual ordered guess, readable positional feedback and multiplier explanation |
| Jungletårnet | Rotation, placement and camera progression | Manual drag/drop, settling and guide mismatch resolution |
| Plisk & Plask | Column choice instead of ordinary matching input | Repeat column selection and conditional bonus return |
| Søde Sager | Automatic group formation and changed bonus qualification | Main five-group unlock and a bonus with unequal coin values |

## 9. Focused validation protocol

For a future authorized demo session, choose representative games from the families above rather than assuming one title verifies every variant. Retain state identity and record the screen with timing and audio if the test environment permits it.

For each representative mechanic:

1. Perform a valid manual action, an invalid or unavailable action where safely possible, and the relevant helper/Auto action separately.
2. Trace input acknowledgement → accepted action → local resolution → progress → qualification. Record where causality is unclear.
3. Repeat enough times to check whether feedback always returns to rest and whether queued/repeated inputs are lost, duplicated or buffered. Record measurements rather than inventing a universal timing target.
4. Observe a phase transition and the final pending-effect drain. Distinguish zero resources from explicit terminal completion.
5. Compare normal and reduced-motion feedback without changing the underlying state transition. Test with audio muted as well as enabled.
6. Verify each rare bonus or recovery only when it actually occurs. A losing ticket, visible bonus label or terminal disclosure does not prove an unexercised branch.

No fresh demo sessions or implementation changes were performed for this report. The per-game catalogue preserves the exact unresolved branches so future inspection can target them.

## Source maintenance and caveats

The source assessments contain some broad tags and stale prose. This analysis adds a separate interpretation layer rather than silently rewriting the historical record. Specific examples include Plisk & Plask's column-refresh input under an older group-clear tag, Fodboldfeber's score comparisons under number-bingo, and Mini Quick's old freshness sentence despite newer current-run evidence.

Guide identity mismatches, including Jungletårnet's configured help link, remain visible in each affected catalogue entry. High-level `demo-verified` or `demo-corroborated` flags do not override explicit limitations in the detailed record.

The catalogue and index are generated by `python tools/build-mechanics-analysis.py` from the frozen [assessments](game-assessments.json), with explicit interpretation overrides. Rebuild the searchable catalogue afterwards with `python tools/build-mechanics-site.py`; it also reads the art-style classifications. This synthesis is authored separately and should be reviewed if family mappings change. The [art-style report](game-art-styles.md) supplies the complementary rendering-style view.