世人以人工智能之记忆，其谬在求之过矣。
大抵记忆之术，欲括往昔于方寸：为要言，为决断，为好恶，为修正之见。此于事已了，固善。然多事未定，或暂立之论，或争辩未决，或证尚不全，或待他日之示。
若君之记性不能存此境，则将疑虑化而为伪明。此物或似井然，然实承一净饰之谎.
此乃其败式：非忘，乃记之过净也。
谄媚者，记忆之过从也。早闭者，记忆之过决也。二者同源：皆出记忆之系统，重舒适与效率，轻于判断。
概要之困
要言非中。每要言择其要，去其非，传其声。
当模压缩纷乱之辩为"用户决择X"时，或可省令牌，而删决择所由之压：所拒之异，所疑之界，X或止真之状。
此乃长时AI系统以非由自信之故而自信之理。
非惟幻生事实，亦幻生决断。
其转：
There are three competing interpretations. One is currently stronger, but the evidence is incomplete.
入之
User believes interpretation one.
此甚为省事。然实乃失于决断。
例证简明。始创者云：“所献不效。”仓促之记忆录曰：“所献败。”然或所献非败也。或布散微弱也。或受众非宜也。或着陆页不明也。或所献虽佳而未验也。
Offer failed乃简明之要也。
Offer unproven; distribution and audience mismatch unresolved忆力为优。
首破其意，次存其疑。
修改非终层
修正之忆力强，盖存思变之由：尔信X，证易Y，将来之行宜调。
然非每值之忆悉合此形。
时无纠正之果，时怀紧张之感。时见模式频现，然证据未足，不可立为断言。时旧念非谬，惟其不备，囿于境，或待佳境以显。
是故，久存之忆，非但恃偏好、要言、纠谬而已。必需未决之忆，为未可归一之物，留其存焉。
未解之忆言曰：
Do not decide this yet.
Do not forget it either.
Keep the tension visible until the evidence improves.
患显矣：设问可化滞为勤，而形制愈美。故存忆须有结构，当有检视之机，立有决断之则。非为饰暧昧以浪漫，惟存不决之态，惟其尚能有所为耳。
| 忆之层级 | 其义在 | 其宜用之境 | 寻常寿数 | 代理当现之，若... | 失效之状 |
|---|---|---|---|---|---|
| 总括之忆 | "此乃所发之事。" | 迅疾之续 | 日数至旬日 | 任务唯需今时之态 | 删去疑虑 |
| 更正之忆 | "此乃所变之事。" | 杜犯重愆 | 久存，陈腐则替 | 今策复蹈已知之败 | 易改易为教条 |
| 未决之忆 | "此乃未竟者。" | 存活之问 | 数周至数月，非恒常之态 | 一决断触及活络之不定界。 | 若不分类，则成滞碍 |
连贯需总括，判断需修正，发现需未解之忆。
不确之构
一佳之未解记忆之条，非泛泛之备忘。当存推理之时知识之状：
核心领域：
- 疑問：何者实未解也？
- 标签/范围：何项目、何领域、或何决断，此触及之？
- 即兴诠释：何为其然也？
- 不確定之界未知者何？
- 尚需证物 何以使问句更切？
- 审策之规 / 期： 何时需收束，迁流，或息灭？
进阶之域，惟机宜可偿其重：
- 每解之信度： 弱 / 中 / 强，或若 20-40% 之概率区间。
- 真伪之验：何事可损可毙诸解？
- 联忆之忆：相关之纠、决断、要旨或门阈。
- 状：开、狭、移门、移纠、决、存档。
是构使疑虑不化怠惰。无之，则“持开放之心”成不决之辞。
更优之模範
於矜正錄旁增一檔案：
open_questions.md
用此核心模範：
## [date] — [question title]
Status:
open / narrowed / moved to gate / moved to correction / resolved / archived
Tags:
[project/domain/decision]
Question:
What is unresolved?
Live interpretations:
1. [Interpretation] — why this is plausible
2. [Interpretation] — why this is plausible
3. [Interpretation] — why this is plausible
Current strongest read:
Which interpretation is leading, and why?
Uncertainty boundary:
What do we not know yet?
Next evidence needed:
What would make this clearer?
Linked memories:
- corrections.md: [...]
- decisions.md: [...]
- gates.md: [...]
Review policy / TTL:
If no new evidence arrives by [date or condition], then [decide / move to gate / archive].
於重大之問，增確信範圍與虛假條件：
Interpretation: [...]
Confidence: weak / moderate / strong, or [20-40%]
Falsified if: [...]
Debiasing check: what would I believe if this interpretation were inconvenient?
確切之百分率，可製造虛假之精確。僅於實際校準預測時方可用之。於多數個人系統，範圍或帶域，較為安全.
具體之例證
編碼：
## 2026-05-24 — Is the slowdown algorithmic or data-shaped?
Tags:
search-api, performance, production
Question:
Is the latency spike caused by the algorithm, the data distribution, or the caching layer?
Live interpretations:
1. Algorithmic complexity — moderate — local profiling shows a slower path on larger inputs.
Falsified if: production traces show constant-time behavior after cache miss removal.
2. Data distribution — moderate — slow requests cluster around unusually large tenant records.
Falsified if: tenant size does not correlate with p95 latency.
3. Cache behavior — weak — recent cache-key change may be causing misses.
Falsified if: hit rate remains stable across the spike window.
Current strongest read:
Algorithmic complexity is leading, but production traces are missing.
Uncertainty boundary:
No production profiling sample yet.
Next evidence needed:
Trace p95 requests by tenant size and cache-hit status.
Linked memories:
- corrections.md: "Do not optimize generated assumptions before profiling."
- gates.md: "Performance fix accepted only after p95 improves on production-like data."
Review policy / TTL:
If traces are not collected by Friday, stop debating and instrument first.
Status:
open
策略：
## 2026-05-24 — Is the market wrong, or is the channel wrong?
Tags:
market-entry, distribution, conversion
Question:
Is weak early traction evidence that the market does not want the offer, or evidence that the current channel is wrong?
Live interpretations:
1. Offer weak — weak — no purchases yet, but the sample is small.
Falsified if: targeted readers save, reply, click, or buy after distribution.
2. Channel mismatch — moderate — the offer has not reached a meaningful targeted sample.
Falsified if: 100 target readers in the right channel produce no clicks, replies, saves, or buys.
3. Positioning weak — weak — the buyer may understand the topic but not the outcome.
Falsified if: interviews show the problem is clear and urgent but the offer still feels irrelevant.
Current strongest read:
Channel mismatch is leading, but the sample is not large enough.
Uncertainty boundary:
No reliable click-through, purchase intent, or target-reader sample yet.
Next evidence needed:
100 targeted readers or 14 days of deliberate distribution.
Review policy / TTL:
Review after sample threshold. If no signal, revise positioning before building a second offer.
Status:
open
要旨非永存诸问。要旨乃止蹩脚之要言，勿令活络之假说，未及证验而先毙.
取用之洁
未决之问，耗心神甚巨。勿悉载诸会话之中.
循此律：
- 将当前活跃之未决诸问，置入
state.md. - 诸开题，依项目、域、决断之类型而标记之。
- 唯取其标记合于事者而载之。
- 若用向量之储，则未决之项，当别立名域或元数据之域。
- 凡逾其时效而未生证验者，当衰之或归之档。
- 行周期之认知审计：何者存续，何者收窄，何者成阈，何者当绝？
于智能体系统，未解之忆当载明元数据：
epistemic_status: unresolved
confidence_range: [low / medium / high]
review_date: [...]
surface_when: [matching project/tag/decision]
若用嵌入或向量数据库，须使未解之物可滤。简法有二：唯当标签相合，且语义相似度足以为意时，方取之。确值因系而异，然其理恒常：未解之忆，当由相关而选，非尽倾于每境之窗。
若欲此顺畅映射于如 Obsidian、Logseq、Mem0、Zep、LangGraph 或定制向量库之工具，宜用前文题注或元数据。
epistemic_status: unresolved
tags: [market-entry, distribution]
status: open
confidence_range: moderate
review_date: 2026-06-07
surface_when: [market-entry, pricing, distribution]
否则，检索即成境染。未决之问众，则使器犹豫，喧扰，而运行费矣。
生息之序，实为要务。
文非别匣，条目迁徙：悬而未决者或狭或分，或为门径，或为勘正，或因新证重开。
迁徙之途：
open question
-> gate when the question becomes testable
-> correction when evidence changes behavior
-> decision when a path is chosen despite uncertainty
-> archived when no longer decision-relevant
-> reopened when new evidence changes the frame
例之生息：
open_questions.md
Question: Is the product weak, or has distribution not reached the right readers?
Status: open until 100 targeted readers or 14 days.
gates.md
Gate: If 100 targeted readers produce no clicks, saves, replies, or buys, revise the positioning.
corrections.md
Correction: "Shipping is not conversion." Publishing created an asset; distribution remained untested.
decisions.md
Decision: Keep the product live at $12 while testing distribution; reject building a second product until the gate resolves.
若百人目标读者反应强烈而无人购买，则问题可再议。
open_questions.md
New question: Is the article strong but the Gumroad page under-converting?
君非恒易旧念。时有 contextualize 之，狭之，或以新证重启之。
诘问之辞
权不在有三檔案，而在使其相爭。
用此提示：
Read state.md, corrections.md, gates.md, and open_questions.md.
Use only open questions whose tags match the current task.
For each relevant open question:
- Check whether it conflicts with a previous correction or active gate.
- Classify it as productive uncertainty, retreaded error, lingering task, or avoidance.
- Flag anything older than 30 days without new evidence or a reviewed TTL.
- Separate what is known from what is assumed.
Do not resolve the question unless the missing evidence is present.
此能捕获最大之失：以"未解决"为幌，不愿纳答。
反模式
未解决之记忆，亦能腐朽。
- 无穷开放/暧昧成瘾： 既得明证，复视不决为通脱。
- 感性朦胧： 守其微觉，而莫辨其可验者。
- 假衡： 视诸解若等，然有一者明证尤著。
- 护我之疑。悬而未决者，盖因结案伤及己心、沉没之资、固有之念或自我之象也。
- 无评语触发创发无返之环，永离其事。
- 无关决策存疑于不关将来之事者。
治之要，在於急务。凡悬而未决者，皆需有复审之机、证验之的、决断之链。若一问不能左右将来之决断，则或非存录之宜.
隐私与团队之境
启发性之问，常较之纠正更为敏感。纠正者，述其谬也；启发性之问，则述其或谬之可能：疑于策略、才干、关系、市场、架构或时机。使未决之记忆，默认存于私域。勿载入每云之代理。公例与实录，当分而治之。
团队或多智能体系统，未决之记忆亦需归属。
- 谁主此问？
- 孰能解之？
- 何证据之标准所求？
- 孰可睹之？
无所有权与裁决之权，共享之疑题，化而为政争之霭。
何知其效
以行為度系統，不以妍麗。追蹤：
- 迟至遗证至而后决者
- 设问由显而隐，遂设关隘。
- 所正之误，起于已解之问。
- 覆辙不重蹈。
- 时序预测之精准
- 审阅触发后之项目成果。
若悬而未决者，无易决断，则饰也；若迟正决而促当闭，则基也。凡手设之务，初则每二周稽之，既稳则一月一稽。一项目五悬而未决者，已足；其余皆当移于档、门、决、正之列。
稽时当问：
- 何问可易决？
- 何问无新证？
- 何问逾其TTL？
- 何问当为门，为正，为决，为藏？
- 何问吾持而开，盖因答不便？
二要义：其一日，启而未决或移之者，三十日之内其比；其二日，已决之问，后可防复谬者之比。
资源与相类之业
此文非谓不决之理为新。理查兹·休厄之《智力分析心理学》亦未尝言。 诸智囊之务，竞设之理，乃有定章。菲利普·泰洛克与善判之计，使校准、概率更迭、预测之术，为众所晓。科学有证伪之法、竞立之模、同侪之评。工程有事故之省、决断之录。律法有括题之术、证验之准、未决之实。
此间之旨较窄：个人AI记忆系统亦需同律。若不存认知之位、不确定之界、复检之契，则将未决之问压缩为自信之概。
值得研习之相关领域：
- 理查兹·休厄( Richards Heuer )智识分析之心理学
- 明断 / 超级预测
- 境迁意移与检索之偏
- 人环评估
- 科学证伪与竞相之假说
- 工技决断录与事后之省
今夕何以始
立open_questions.md之
书一问于萦思而不能诚决者
依四则之法：
- 请至少举出两种现场演绎之例。
- 为每项诠释设置信区间及证伪条件。
- 何证据之阙？
- 请指明TTL（Time To Live）或复审触发器之名。
乃问汝之代理人：
Read open_questions.md.
Tell me which current decision is being treated as settled even though the record says it is still unresolved.
Tell me which open question is productive uncertainty, and which one is avoidance.
Do not resolve a question unless the missing evidence is present.
若使者在适处滞汝，则文件有效矣.
校正之忆护汝不重蹈其败。未解之忆护汝不杀未解之事。
此乃校正记忆之框架，其第二层也：存知识之状于思辨之时，盖知者何、推者何、争者何、犹阙之证何也。