Method · Updated 2026-07-09
Method
World model is an overloaded term. This site does not try to declare absolute truth. It records claims, sources, evidence levels, review dates and boundaries so readers and agents can judge reliability.
世界模型正在被很多团队、产品和媒体使用。本站不做绝对真相认证，而是记录判断、来源、证据等级、复核日期和边界， 让读者和 agent 能判断一条说法有多可靠、应该在哪里被复查。
Short answer / 一句话结论
The site does trust layering, not truth certification. A claim can be useful and still be labeled as official, unverified, outdated, conflicting or bounded.
本站做的是可信分层，不是绝对真相认证。一个判断可以有参考价值，同时仍被标注为官方声明、未验证、已过期、 来源冲突或有适用边界。
Source priority / 来源优先级
Primary sources are preferred because world-model claims drift quickly. Secondary commentary can help discovery, but it cannot by itself turn a demo or product phrase into a capability conclusion.
世界模型相关说法变化很快，所以本站优先使用一手来源。二级解读可以帮助发现线索，但不能单独把一个演示或产品话术变成能力结论。
| Priority | Source type | How it is used |
|---|---|---|
| 1 | Papers, technical reports, official model pages and official product docs. | Verification base for definitions, dates, model framing and capability boundaries. |
| 2 | Company blogs, research posts and launch notes. | Useful for official positioning, but still labeled as official or product claims. |
| 3 | Reliable interviews, conference talks and demos. | Useful context, not enough by itself for durable claims. |
| 4 | Media coverage, social posts and secondary commentary. | Useful for discovery; not used as proof unless cross-checked. |
Claim types / 判断类型
Claim type describes where a statement comes from. It does not rank importance. A company product claim may be important, but it remains a product claim until stronger evidence appears.
判断类型说明一句话来自哪里，不等于重要性排序。一个公司产品声明可能很重要，但在更强证据出现前，它仍然只是产品声明。
| Type | Meaning | Boundary |
|---|---|---|
official_claim |
A company or lab says this about its own system. | Not independent validation. |
technical_report_claim |
A claim made in a paper, benchmark note or technical report. | May be task-specific or dependent on benchmark design. |
product_claim |
A capability or workflow exposed through a product page or launch. | Can change quickly as access, pricing and features change. |
editorial_judgment |
World Model Atlas synthesis based on one or more sources. | Must stay traceable to sources and open to correction. |
unverified / outdated / conflicting |
A claim with missing, stale or inconsistent support. | Should not be treated as settled. |
Evidence labels / 证据标签
- High: directly supported by a primary source, with clear wording and current review date.
- Medium-high: supported by official or report material, but downstream utility is still task-dependent.
- Medium: plausible synthesis or qualitative evidence, but not enough for a hard capability conclusion.
- Low / unverified: discovery-only material or claims that need rechecking before use.
这些标签是本站内部的编辑标签，不是通用 benchmark 分数。它们用来提醒读者：这句话的支撑材料到底有多强，边界在哪里。
Review protocol / 复核协议
A claim must be rechecked when it appears on the homepage, is repeated across articles, depends on a
fast-moving product, has an access warning, or reaches its reviewDue date.
如果一个判断出现在首页、被多篇文章重复使用、依赖快速变化的产品、有访问异常，或已经到达 reviewDue 日期，
就必须复核。
| Situation | Action | Public record |
|---|---|---|
| Stronger primary evidence appears. | Upgrade evidence level or confidence only if the claim wording still matches the source. | Update lastReviewed, reviewDue and source IDs. |
| Source wording changes or product behavior changes. | Downgrade, narrow or rewrite the claim. | Record the new boundary in boundary and Sources review notes. |
| Two public sources conflict. | Keep the conflict visible instead of blending them into a stronger statement. | Mark claim status as conflicting or add a conflict note. |
| Review date passes for a prominent claim. | Recheck before keeping it in homepage, article intro or metadata. | Mark as outdated if it cannot be checked in time. |
Source access failures / 来源访问异常
A 403 or 404 is not handled silently. A restricted official URL may remain as a source with an access note; a missing source needs replacement, archive recovery or downgrade before it supports a prominent claim.
403 或 404 不能静默忽略。官方 URL 如果因为访问策略返回 403，可以保留但必须写明访问说明；如果来源消失，则要替换、找归档， 或把相关判断降级。
ok_200_at_review: source was reachable during the latest check.restricted_403_at_review: access was blocked, but the source may still be official and valid.not_found_at_review: source should not support a major claim until replaced or archived.manual_review_needed: source requires browser, archive or publisher-side checking.
What this site does not claim / 本站不宣称什么
- It does not certify that a model has true physical understanding.
- It does not treat a polished demo as proof of long-horizon stability.
- It does not treat company marketing language as independent evidence.
- It does not attempt broad AI image/video forensics in the current phase.
- It does not sell access to basic explanations or source records.
换句话说，本站现阶段是可信资料入口，不是评测机构、产品排行榜、自动事实裁判，也不是把公开解释内容锁进付费墙的会员站。
Agent-readable layer / Agent 可读层
The structured layer is static JSON. It is meant to be easy to read, cache and cite; it is not an API, an account system or a permissioned service.
结构化层目前只是静态 JSON，目标是方便读取、缓存和引用，而不是 API、账号系统或权限服务。字段解释见公开 schema。