Learning and Research Board

The goal is not to collect features. For every reference, identify how it creates desire, communicates risk, and lets the player recover.

Use this five-question note after each session:

1. What information is known before committing to risk?
2. Is the player making a decision, or only receiving randomness?
3. What is lost on failure?
4. How long does recovery take?
5. Does the reward change later play, or only increase a number?

Priority A: Before Competition Implementation

• RHS judging overview — Read the categories and criteria for gardens and floral exhibits. Extract 4–6 terms that can become understandable game scores: condition, brief, impact, difficulty, plant health, presentation.

https://www.rhs.org.uk/shows-events/exhibit-at-a-show/judging-at-rhs-shows

• American Rose Society container-grown judging — Read the container-grown section. Decide which real criteria are fun and which would become maintenance clutter.

https://rose.org/wp-content/uploads/2024/06/2024-Horticulture-Judging-Guidelines.pdf

• Observe one real judging ceremony or walkthrough — Watch an RHS judging video or flower-show walkthrough. Note how suspense, expertise, and medals are presented to non-experts.

• Define acceptable loss — Write three examples each of a satisfying event failure and an unacceptable event failure. Include money, time, flowers, seeds, and information.

• Rank the two desired peaks — Describe the ideal 10-second experience of cultivation success and the ideal 10-second experience of selling out. They may both exist, but one must lead into the other.

Priority B: Risk, Debt, and Shop Research

• Potionomics — Study how debt, competition, preparation, and negotiation create pressure. Record when the pressure feels motivating and when it feels exhausting. Watching a focused playthrough is enough if buying is not useful.

https://store.steampowered.com/app/1874490/Potionomics/

• Recettear — Study debt deadlines, sourcing, pricing, customer reactions, and what failure means. Pay special attention to whether recovery creates a new strategy or only repeats lost work.

https://store.steampowered.com/app/70400/Recettear_An_Item_Shops_Tale__Tale/

• Luck be a Landlord — Study bounded gambling: visible deadlines, build choices before a roll, readable combinations, and how losses lead into another attempt.

https://store.steampowered.com/app/1404850/Luck_be_a_Landlord/

• Shop failure model — Sketch what happens when the roadside stand, market stall, and leased shop each fail. Protect the greenhouse and basic seeds in every version.

• Offline debt rule — Compare active-cycle interest, real-time interest, and fixed installment schedules. Explain which one respects idle players.
• One-worker policy paper test — Give a fictional assistant six flowers, one ready order, one rare specimen, and limited cash. Using only harvest priority, minimum stock, protected items, and cash reserve, verify that the assistant behaves predictably without solving the whole game.
• Automation responsibility audit — For watering, fertilizing, treatment, harvesting, ordering, selling, and replanting, label each action as machine, employee, or player responsibility. Reject duplicate upgrades that perform the same job with different flavor text.

Priority C: Plant Knowledge and Tone

• Butterfly orchid lifecycle — With an experienced grower or an authoritative horticultural source, identify 6-8 visually distinct stages from establishment through flower spike and bloom. Record what is actually observable without uprooting the plant.
• Pixel growth-stage study — Compare three pixel-art farming games. Capture how many frames each plant uses, which silhouette changes communicate growth, and what remains readable at phone size.
• One-loop music exercise — Enter the two theme phrases in MuseScore, GarageBand, or BeepBox. Test 4/4 and 6/8, then note which version feels like waiting for growth rather than completing office work.

• Study one plant with the creator's partner — Follow a real plant from seed or cutting through one visible change. Record what she checks, what surprises her, and which action feels caring rather than technical.

• Strange Horticulture — Study how plant knowledge, customers, labels, and a physical-feeling interface make information pleasurable. Do not copy its occult story; focus on information UX.

https://store.steampowered.com/app/1574580/Strange_Horticulture/

• A Dark Room — Revisit the first 30 minutes. Record exactly when the interface reveals a new system and what desire makes the player engage with it.

https://github.com/doublespeakgames/adarkroom

• Kittens Game — Observe how old resources remain relevant when new systems arrive, and where the interface becomes too dense for a new player.

https://kittensgame.com/web/

• Rusty's Retirement — Observe when automation arrives, what the player still controls, and how the game remains readable while running in the background.

https://store.steampowered.com/app/2666510/Rustys_Retirement/ 园主感想：玩过，但没能玩太久。界面有点让人迷惑；一开始并不能真正放置，要手动做的事不少，后面才慢慢加入自动化；屏幕底部那条长条界面需要来回滚动，体验不太好。这也是园主想做一个"自己能玩下去的种植游戏"的原因之一，还曾想过做一个终端版的种田游戏来找新奇感。

Priority D: Future Floral Arrangement

• Find three spatial inventory or tile-placement games whose placement is satisfying in under ten seconds.
• For each, separate spatial strategy from visual decoration.
• Prototype on paper: place common flowers as volume, rare flowers as accents, and foliage as connectors.
• Decide whether arrangement quality should come from adjacency rules, silhouettes, color harmony, customer briefs, or a small combination.
• Reject any version that requires frequent rotation, pixel-perfect placement, or inventory housekeeping without meaningful choices.

Priority E: Plant Health and Rival Competition

• Follow one real symptom over time — With the creator's partner, photograph one affected plant across several days and record leaf position, spot shape, spread, watering, weather, and what treatment was attempted. Consult an authoritative horticultural or university extension source before assigning a disease name.
• Plant Tycoon — Study how breeding, pests, health, treatment, and slow discovery feed one another. Record which health systems create curiosity and which become repetitive chores.
• Disease-loop paper prototype — Test one path using cards: visible symptom -> two possible causes -> one investigation -> targeted treatment -> permanent resistance note. Reject versions where the correct answer is only random.
• Enhanced-seed vocabulary — Compare selective breeding, researched traits, and fictional gene modification. Decide which framing matches the garden's tone and how a player reads resistance without opening a genetics spreadsheet.
• Rival match without sabotage — Draw two candidate pots with visible progress and one shared judging brief. Test whether comparison alone creates enough tension.
• Bounded sabotage paper test — Give both sides three visible attack cards and three visible counters. Every attack must be telegraphed, event-only, and recoverable. Record whether the result feels funny, strategic, mean, or merely busy.
• Safety boundary — Keep pesticides and fertilizers as abstract or fictional game items unless text has been checked against an authoritative real-world source. Do not turn game recipes into real application instructions.

Priority F: Lineage and Progressive Complexity

• Progression-model comparison — Compare how Pokémon separates experience, individual differences, inheritance, and competitive titles. Identify which parts create attachment and which parts become hidden-number homework.
• Three-generation paper lineage — Raise one fictional plant family through three generations using only two traits, one title, and one tradeoff. Test whether each generation feels distinct without opening a detailed genetics panel.
• No universal champion test — Write three competition briefs that reward incompatible strengths. Confirm that no single lineage is the obvious best choice for all three.
• Automation exit audit — List every repeated greenhouse action and name its eventual batch action, device, employee, or persistent policy. Any action without an exit ramp blocks the next complexity layer.
• Progressive-disclosure replay — Revisit the first 30 minutes of A Dark Room and one mature idle-game save. Record what disappears, compresses, or becomes summarized when the next system arrives.
• Lineage language test — Compare 等级种子, 功勋种子, 冠军谱系, and a player-named family. Ask which wording implies training, inheritance, prestige, or confusing magic.

Priority G: 新手前五分钟与信息密度

起因：丹丹（游戏引擎 UI 从业者）试玩时报告"一开始文字太多、不知道看哪里"， 后期逐渐疲惫退出。现有条目几乎全是机制研究，界面与信息架构是空白。

• 《Refactoring UI》（Adam Wathan / Steve Schoger）— 确认存在。

只读"层级""留白""少用文字"三部分。提取：同一屏里如何只让一件事最响。 我们的高频区（温室首屏、订单卡、商店）逐个对照一遍。

• 《Don't Make Me Think》（Steve Krug）— 确认存在。

重点是"自明性"：一个控件不该需要说明文字。对照我们现在每个面板下方的解释段落， 列出哪些可以靠控件本身表达、哪些必须留。

• 自测：五分钟无声试玩 — 找一位没玩过的人，全程不解释、不提示，只记录

他们停顿超过 3 秒的每一个时刻和当时在看哪里。不要问"好不好玩"，只记"卡在哪"。 这比任何文档都便宜，而且可以反复做。

• 对照三个挂机游戏的前五分钟 — 建议 Universal Paperclips、A Dark Room（已在板上）、

Cookie Clicker。记录：第一屏有几个可点的东西？第二个系统在第几分钟出现？ 出现时旧系统是否被压缩或收起？ （Universal Paperclips 由 Frank Lantz 制作，以"逐步揭示"著称——确认存在，网页版免费。）

• 文案密度盘点 — 把当前所有面板的中文字数统计一遍，按屏排序。

给每一段问一句："删掉它，玩家会做错什么？"答不上来的进删除候选。

Priority H: 升级节奏与曲线

起因：花盆梯子实测 40 金（1 单）→ 160 金（4 单）→ 2000 金（48 单）， 第三级是前一级的 12.5 倍；160–2000 区间内没有任何吞吐升级可买。

• 自建一张进度曲线表 — 用表格列出每一级升级的：价格、约合多少单、

实测到手时间、相对上一级倍数、解锁了什么新动作。 规则：任何一级不应超过上一级的 3 倍，除非它同时解锁一个新玩法层。 这张表应该先于数值改动存在，而不是改完再补。

• Egg, Inc.（手机挂机，确认存在）— 观察它的升级阶梯密度：

体验记录 几乎每隔一两分钟就有一样东西可买。记录：便宜升级和昂贵升级如何交替出现， 以及"下一个目标"在界面上如何始终可见。

• Melvor Idle（网页/手机挂机，确认存在）— 观察成熟期的信息密度与

多系统并行时的导航。它是"内容极多但仍可玩"的正面样本，也是密度上限的反面教材。

• 拆一次"空档" — 拿任意一个你玩过的挂机/经营游戏，找出它最长的那一段

"没东西可买"的时间，量一下有多久，以及那段时间游戏给了什么替代目标。

• 《Designing Games》（Tynan Sylvester，RimWorld 作者）— 确认存在。

读"决策"与"节奏"两章，其余可跳。提取：一个升级应该改变玩家的做法，还是只改变数字。

Priority I: 挂机游戏的会话结构与疲劳

起因：丹丹"逐渐疲惫、玩不动了"。我们至今没有关于"一次游玩应该多长"的任何定义。

• 定义一次会话 — 写清楚：我们希望玩家一次坐下玩多久？

离开多久回来算正常？回来时应该看到什么？现在这三个问题都没有答案， 所以也无法判断"疲惫"是不是设计预期之外。

• 记录自己的退出点 — 园主和顾问各自完整玩一次，在想停下的那一刻立刻记下

时间、当时在做什么、以及"想停"的原因是无聊、累、还是没目标。不要强撑。 园主感想：因为太熟悉这个游戏，耐受度比一般人高，但一般到解锁花摊之后就开始无聊。那段时间觉得扩建又贵又慢、自动化要等很久，只能开着四倍速种花卖花，一边卖一边犹豫是不是直接交任务更赚，又不甘心，觉得钱赚得少、难赚，刺激不够。顾问和青禾都是 AI，不会真的想停下，这一项以园主的记录为准。后续讨论见讨论板 rough-machine-first。

• 《The Art of Game Design: A Book of Lenses》（Jesse Schell）— 确认存在。

不必通读，用"透镜"当检查清单。建议先看与节奏、期待、反馈相关的几片透镜。

• Rusty's Retirement（确认存在，已在开发站参考链接中）— 它主打"放在桌面角落

不打扰你"。观察：它如何让玩家不必盯着，以及它有没有"必须回来处理"的时刻。 园主感想：玩过，没玩太久。一开始不太能放置、手动操作多，底部长条界面要来回滚动。详见 C 组同名条目。

• Stardew Valley — 观察作物在长的时候，玩家在做什么。

https://store.steampowered.com/app/413150/Stardew_Valley/ 园主感想：去年沉迷过几个月，每天都很充实。它基本是牧场物语的借鉴和升级：每天先照顾动物、采集，然后浇水、收获、播种，之后拜访村民、参加活动、钓鱼采矿、做委托、收集野外资源。很多事不是必须做的，但做与不做收益不同，需要自己权衡和安排，而且一般不会有重大损失。

• GDC 放置游戏设计演讲（园主自己找到的视频）— 看放置游戏的核心循环和吸引力是怎么被拆开讲的。

https://youtu.be/Lu-RjxeDpU8 园主感想（看了一半）：内容和自己玩放置游戏时体会到的差不多，但拆得很清楚。核心是轻操作甚至无操作，加上持续的进展和奖励；循环是打开游戏、花光钱和资源、下线等待、再重复。这类游戏通常让在线收益远高于挂机。还有 prestige：到某个阶段可以重来，但能力大幅提升，回到开头能轻松碾过初期流程。由此园主意识到，我们的游戏本质更像管理游戏，却因为初心想做放置而硬加了"等花开"，结果让玩家在坐牢，像是错误的灵魂进了身体。后续讨论见讨论板 rough-machine-first。

• Viridi — Steam 上的免费小游戏，一盆多肉按真实时间慢慢长。看"纯照料、几乎零管理"能让人留多久，以及为什么留得住。

• Botany Manor — 几个小时能玩完。植物在这里是"要弄清楚的谜"（什么条件下才会开花），而不是要等的计时器。看种植的乐趣能不能不靠等待。

• 《A Theory of Fun for Game Design》（Raph Koster，中译《快乐之道》）— 很薄，图多字少。核心观点是：无聊，往往是玩家已经学会了眼前的模式，又没有新东西可学。可以拿来想"教程之后为什么又觉得慢"。

──────────────────────── 关于可信度的说明 ────────────────────────

上面标"确认存在"的书和游戏我有把握。我没有推荐具体的演讲标题或视频链接， 因为我不能确认某个具体 talk 的确切名称与出处；与其给一个可能错的标题， 不如给主题和检索方向，由园主自己找当下最合适的那一份。 后来园主自己找到了一份 GDC 放置游戏设计演讲，已放进 Priority I。

如果需要具体视频，建议按主题检索： "idle game design progression"、"onboarding first five minutes game"、 "information hierarchy game UI"。找到后按板子开头那五个问题做笔记即可。

Discussion Template

When bringing research back into the jam, use:

Reference:
Moment observed:
What I felt:
Mechanism causing it:
What might transfer:
What should not transfer:
Question for our prototype: