Pyth 与 Chainlink 预言机新鲜度：置信区间与 Heartbeat 披露雷达
本文正文暫以英文原文展示，翻譯版本需完成來源複核後再發布。
已验证事实边界
1. Pyth best-practices documentation discusses confidence intervals for price feeds. 2. Pyth price-data documentation explains price-data fields and feed update information. 3. Chainlink Data Feeds documentation describes Data Feeds as an official price-feed surface. 4. Chainlink Data Feeds API reference documents latestRoundData fields including timestamps used for freshness checks. 5. A price value without freshness and uncertainty context is not enough for collateral, liquidation, settlement or user-facing risk decisions. 6. Public oracle documentation does not prove a particular application selected the correct feed, threshold, fallback or emergency pause rule.
收起完整分析
决策矩阵
文章表格：可橫向滑動，或聚焦表格區域後使用方向鍵。
| Check | What it means | Practical action | Unknown to keep explicit |
|---|---|---|---|
| Freshness timestamp | publish time or updatedAt is the first control for deciding whether data can still be used | store feed timestamp and local read timestamp before acting on the value | chain latency, delayed publishers and application-specific freshness tolerance |
| Uncertainty field | confidence interval or deviation context explains uncertainty around the reported value | surface uncertainty alongside price instead of displaying price as a guarantee | how each app translates confidence into collateral or liquidation policy |
| Heartbeat threshold | feed update cadence can be threshold-driven rather than continuous | record the expected update trigger and stale-data cutoff for the feed | market closure, outage behavior and feed-specific heartbeat values |
| Fallback route | a fallback can protect users only if it is written before the primary source fails | define fallback source, pause rule and reconciliation requirement in advance | whether fallback data has equal rights, precision and manipulation resistance |
| Stop condition | oracle data must not be used when freshness, uncertainty or feed identity is unknown | pause pricing, liquidation or user action when timestamp or uncertainty evidence is missing | future feed migration, publisher set changes and app-level emergency governance |
来源地图
- Pyth Developer Hub：Best Practices（pyth-best-practices）
- Pyth Developer Hub：Understanding Price Data（pyth-understanding-price-data）
- Chainlink Documentation：What are Chainlink Data Feeds?（chainlink-data-feeds-overview）
- Chainlink Documentation：Data Feeds API Reference（chainlink-data-feeds-api）
控制说明
1. Freshness timestamp
这一项要求读者把公开来源翻译成操作规则。这里的含义是：publish time or updatedAt is the first control for deciding whether data can still be used。可执行动作是：store feed timestamp and local read timestamp before acting on the value。仍然未知的是：chain latency, delayed publishers and application-specific freshness tolerance。因此本文不会把未知部分包装成收益承诺、牌照承诺、审批保证、安全保证、恢复保证、平台排名或价格暗示。
2. Uncertainty field
这里的含义是：confidence interval or deviation context explains uncertainty around the reported value。可执行动作是：surface uncertainty alongside price instead of displaying price as a guarantee。仍然未知的是：how each app translates confidence into collateral or liquidation policy。
操作测试也必须很窄。这就是本文坚持以清单形式呈现的原因。
3. Heartbeat threshold
这里的含义是：feed update cadence can be threshold-driven rather than continuous。可执行动作是：record the expected update trigger and stale-data cutoff for the feed。仍然未知的是：market closure, outage behavior and feed-specific heartbeat values。
操作测试也必须很窄。这就是本文坚持以清单形式呈现的原因。
4. Fallback route
这里的含义是：a fallback can protect users only if it is written before the primary source fails。可执行动作是：define fallback source, pause rule and reconciliation requirement in advance。仍然未知的是：whether fallback data has equal rights, precision and manipulation resistance。
操作测试也必须很窄。这就是本文坚持以清单形式呈现的原因。
5. Stop condition
这里的含义是：oracle data must not be used when freshness, uncertainty or feed identity is unknown。可执行动作是：pause pricing, liquidation or user action when timestamp or uncertainty evidence is missing。仍然未知的是：future feed migration, publisher set changes and app-level emergency governance。
操作测试也必须很窄。这就是本文坚持以清单形式呈现的原因。
仍需明确的未知项
- chain latency, delayed publishers and application-specific freshness tolerance
- how each app translates confidence into collateral or liquidation policy
- market closure, outage behavior and feed-specific heartbeat values
- whether fallback data has equal rights, precision and manipulation resistance
- future feed migration, publisher set changes and app-level emergency governance
实用结论
预言机准备度取决于 freshness 与 uncertainty controls，而不只是价格端点是否返回数字。Pyth 记录了 confidence intervals 与 price-data fields；Chainlink 记录了 Data Feeds，并在 API reference 中说明 latestRoundData 包含 update timestamps。雷达输出是：在把 feed 用于抵押品、清算、定价或用户展示前，记录 publish time、confidence/deviation meaning、heartbeat/update threshold、stale-data rule、fallback route 与 action stop condition。 本文停在证据边界内，帮助读者决定下一步核验什么，而不是把公开文档改写成私人账户、钱包、API key、池子路径、稳定币余额、代币化资产或政策框架在任何未来条件下都能成功的承诺。