GSC Blog: Core Updates 与算法迭代演化(2010-2026)

23 篇 Google Search Central blog 官方公告的主题化提炼。不按时间排序,按"算法心智模型"组织 —— 想理解 Google 自己怎么解释算法演化,这一份比时间线更省力。

23 篇里有 2 篇是非英文 stub(2010-06 Caffeine 德文 / 2019-12 Events 西德文),无法直接引用其内容,但通过 stub 自身的元数据可参与时间线编排。

Part 1 — Core Update 的官方心智模型

Google 反复用同一段话定义 broad core update,几乎是模板化复用。关键 take-away:核心更新不是惩罚机制,不是 spam action,不针对具体页面或站点 —— 它是"重新评估全网内容相对价值"的过程。

1.1 标准定义(2019 年首篇 core-updates 公告 + 后续所有公告 reused)

来源:2019/08/core-updates.en.md —— 这一篇是后续所有 core update 公告的"圣经",所有后来的公告都在引用它。

Several times a year, we make significant, broad changes to our search algorithms and systems. We refer to these as "core updates." They're designed to ensure that overall, we're delivering on our mission to present relevant and authoritative content to searchers.

我们每年发几次广义核心更新,目的是确保整体上给用户呈现相关、权威的内容。

1.2 反惩罚叙事(Google 反复强调的"不要乱改")

There's nothing wrong with pages that may perform less well in a core update. They haven't violated our webmaster guidelines nor been subjected to a manual or algorithmic action... In fact, there's nothing in a core update that targets specific pages or sites. Instead, the changes are about improving how our systems assess content overall. These changes may cause some pages that were previously under-rewarded to do better.

—— 2019/08/core-updates,2022/05/may-2022-core-update 原样引用 这段

Why:这段话每次 core update 都几乎逐字复用(2019、2022/05、2024/03、2024/08 均含),目的是劝退 site owner 不要"修错东西"。Google 的潜台词是:你的下跌不是因为你做错了什么,而是别人做对了 —— 算法只是 reassess 了相对价值。

1.3 电影榜单类比(Google 最爱讲的故事)

One way to think of how a core update operates is to imagine you made a list of the top 100 movies in 2015. A few years later in 2019, you refresh the list. It's going to naturally change. Some new and wonderful movies that never existed before will now be candidates for inclusion. You might also reassess some films and realize they deserved a higher place on the list than they had before.

—— 2019/08/core-updates

这是 Google 解释 core update 最经典的隐喻:榜单变化 ≠ 老电影变烂,而是新电影更好 + 旧电影价值被重新认识。

1.4 Why 它存在:对抗"内容退化 + 用户期望升级"双向漂移

网页侧:从静态 HTML → 富媒体 → 实时流(Caffeine 2010 时代的核心驱动力);
用户侧:"Searchers want to find the latest relevant content and publishers expect to be found the instant they publish."(2010/06/caffeine);
算法侧:需要持续 re-weight 200+ 信号(Penguin 2016 公告 reused 这个数字);

所以 core update 不是"修 bug",是"持续校准 200+ 信号的相对权重"。

Part 2 — 时间线:从 Caffeine 到 2026 Discover

2.1 三大时代切分

时代	核心叙事	代表更新
2010 索引时代	把"批量索引"改成"连续增量索引"	Caffeine (2010/06)
2012-2016 算法浪潮	单点信号(Panda/Penguin)成为独立算法 → 后被吸收进 core	Penguin → core (2016/09)
2019-2022 Core Update 制度化	broad core updates 命名规范化,几乎每季一次	2019/08 first guidance, 2022/05
2022-2024 helpful content 时代	加 site-wide signal 反"为搜索引擎写的内容";最终融入 core	Helpful Content Update 2022/08 → 2024/03 吸收
2024-2026 收敛与简化	core update 越来越"复杂多系统融合";产品功能开始减法	March 2024 core(multi-system), 2025/11 简化 SERP, 2026/02 Discover core

2.2 关键里程碑表

日期	公告	性质	核心机制
2010-06-01	Caffeine/Mayday(德文 stub)	索引架构	长尾改动
2010-06-09	Caffeine 新索引系统正式宣告	索引架构	从分层批量索引 → 连续增量索引,50% fresher;100M GB 存储
2016-09-23	Penguin is now part of our core	算法集成	Penguin 从独立 refresh → 实时+granular,不再单独公告 refresh;并入 core ranking
2019-08-01	What site owners should know about core updates	制度化	首次系统性解释 broad core updates,引入 E-A-T 自评清单;后续所有 core update 公告的母版
2019-12-01	Events 在德/西上线(非英文 stub)	功能上线	跟 core update 无关,属于 SERP 功能扩展
2020-05-28	Evaluating page experience for a better web	信号引入预告	预告 Core Web Vitals(LCP/FID/CLS)+ 既有 page experience 信号将合并为一个"页面体验"ranking signal,推迟到 2021;移除 AMP requirement for Top Stories
2020-11-10	Timing for page experience	时间确认	锁定 2021/05 上线;Lighthouse/PageSpeed 使用涨 70%
2021-04-08	April 2021 Product Reviews Update	专题更新首发	仅英文,针对"thin content 评测";引入 in-depth/expert/原创性自评清单
2021-04-19	More details on page experience update	时间细化	从 2021/06 中开始 gradual rollout,2021/08 全量;Top Stories 摘除 AMP 强制
2021-07-26	Link spam update + qualifying links 提醒	spam 更新	提醒 rel="sponsored"; 推出新 link spam algorithm
2021-08-04	Simplifying the Page Experience report	报告/澄清	Safe Browsing 不是 ranking signal(关键澄清);Ad Experience 从 PE 报告剥离
2021-11-04	Bringing page experience to desktop	范围扩展	2022/02 起 desktop 也用 page experience;mobile-friendliness 仅 mobile
2021-12-01	December 2021 Product Reviews Update	专题迭代	加两条 best practice:实测证据(visuals/audio)+ 多卖家链接
2022-03-23	Improving Product Review ranking, one year on	专题迭代	引入"ranked lists/comparison reviews 也算"的 FAQ;为"best of"推荐加 first-hand evidence
2022-05-25	May 2022 core update	常规 core	1-2 周 rollout;原样引用 2019 模板话术
2022-08-18	Helpful Content Update(HCU)	新机制	新增 site-wide signal(非 spam action),ML classifier 持续运行;"people-first content" 概念正式登场
2022-12-14	December 2022 Link Spam Update	spam 更新	首次提到 SpamBrain AI 检测 link spam;识别买/卖链接的站点
2023-04-19	The role of page experience in creating helpful content	大转向	page experience 从"signal/system"退回为"指导原则";承认"没有单一的 page experience signal";Page Experience report 被拆
2023-10-04	October 2023 Spam Update	spam 更新	多语言覆盖(土耳其/越南/印尼/印地/中文等);针对 cloaking / hacked / auto-gen / scraped
2024-03-05	March 2024 core update + new spam policies	大融合	"more complex than usual,涉及多个 core systems";HCU 被吸收回 core;三条新 spam policy:expired domain abuse / scaled content abuse / site reputation abuse;后者 2024/05/05 生效
2024-08-15	August 2024 core update	修正性 core	明确"听到了 creators 反馈",更多 reward 小型独立站;捕捉网站的改进
2025-11-05	简化 search results page 更新	产品减法	不是 ranking 更新 —— 砍 SERP 上不常用的 features;2026/01 起 Search Console API 去掉对应 structured data type
2026-02-05	February 2026 Discover Core Update	Discover 专题 core	专门给 Discover 的 broad core update;reduce sensational/clickbait;topic-by-topic expertise(同一站点不同主题分开评估)

Part 3 — 专题更新:Spam / Product Reviews / Helpful Content 三条线如何收敛

3.1 Spam Update 线(从规则驱动 → AI 驱动)

2016 Penguin 并入 core 是分水岭。Penguin 公告(2016/09)宣布两件事:

Penguin is now real-time... we're not going to comment on future refreshes. Penguin is now more granular. Penguin now devalues spam by adjusting ranking based on spam signals, rather than affecting ranking of the whole site.

Why:从"惩罚整站 + 周期性公告 refresh"→"信号化、实时化、页面级别"。

之后 spam 公告的关键演化:

2021/07 Link Spam Update:还是基于规则 + 算法;
2022/12 December Link Spam Update:首次明确披露 SpamBrain(AI-based spam-prevention) 用于 link spam;
2023/10 October Spam Update:多语言覆盖,community-driven(community 报上来的 spam 类型反向驱动);
2024/03 spam policies 大改:从"反 auto-generated content"→"反 scaled content abuse"(无论人/机/混合,只要 scale + 操纵 ranking 就算)。

Our long-standing spam policy has been that use of automation, including generative AI, is spam if the primary purpose is manipulating ranking in Search results. The updated policy is in the same spirit of our previous policy and based on the same principle. It's been expanded to account for more sophisticated scaled content creation methods where it isn't always clear whether low quality content was created purely through automation.

—— 2024/03/core-update-spam-policies

Why:这是 Google 在 AI 生成内容大爆发后给的官方定调 —— 不反对 AI,反对 scale + manipulative intent。

3.2 Product Reviews 专题线(独立 → 并回 core)

2021/04 首发:仅英文,反"thin 评测";引入"in-depth/expert"清单;
2021/12 第 2 次:加"实测证据"+"多卖家链接";
2022/03 第 3 次:覆盖 ranked lists / "best of" 推荐;
后续:Product Reviews Update 在 2023 起逐步并入 core update,2024/03 之后官方公告里基本不再单独提"product reviews update"。

Why 单独搞一条线又并回 core:Google 实验性把"商品评测"做成单独信号,验证有效之后,把方法论(in-depth / first-hand / 实测)沉淀进 core ranking 的 helpful content 评估体系。

3.3 Helpful Content Update(HCU) 线 —— 最戏剧性的并回 core

2022/08 HCU 首次发布 是个新机制:

不是 spam action,不是 manual action;
是 site-wide ML classifier signal(关键 quote:This classifier process is entirely automated, using a machine-learning model);
持续运行,可监控新老站;
"去掉 unhelpful content 可以帮你其他内容排名" —— 第一次明确说低质量内容会拖累整站。

2024/03 公告 HCU 被吸收回 core:

The March 2024 core update is a more complex update than our usual core updates, involving changes to multiple core systems. It also marks an evolution in how we identify the helpfulness of content... There's no longer one signal or system used to do this.

Why 这是大事:Google 明确告诉 SEO 圈 ——不要再去研究"helpful content classifier"是什么、怎么过它了 —— 它已经被打散进多个 core systems。这把 HCU 从一个可单独"诊断/规避"的对象变成了"系统性的判断"。

Part 4 — Page Experience / Core Web Vitals 的位置:从"ranking signal"→"tie-breaker"→"guidance"

这是 23 篇里最跌宕起伏的故事线,值得单独追:

4.1 上升期(2020-2022):隆重引入

2020/05 evaluating-page-experience:预告将 CWV(LCP/FID/CLS)+ mobile-friendly + HTTPS + intrusive interstitials 合并为"a new signal";关键 quote:a holistic picture of the quality of a user's experience on a web page;
2020/11 timing:锁定 2021/05;
2021/04 more details:改为 2021/06 中-08 底 gradual rollout;预告 SXG / Top Stories 摘除 AMP;
2021/08 simplifying:关键澄清 Safe Browsing 不是 ranking signal(也即 PE signal 不包含 Safe Browsing);
2021/11 desktop:扩展到 desktop;

但即使在隆重期,Google 已经埋了对冲叙事:

While all of the components of page experience are important, we will prioritize pages with the best information overall, even if some aspects of page experience are subpar. A good page experience doesn't override having great, relevant content. However, in cases where there are multiple pages that have similar content, page experience becomes much more important for visibility in Search.

—— 2020/05/evaluating-page-experience(这段在 2020-2021 的多篇里反复出现)

Why:从一开始 Google 就把 PE 描述成 tie-breaker(同等内容下才有显著影响),不是 primary ranking factor。

4.2 转折(2023/04):悄悄"撤回"

2023/04/page-experience-in-search.en.md 是整个 PE 故事的关键转折点。表面是"把 PE 整合进 helpful content guidance",实质是 Google 在 walk back:

There is no single signal. Our core ranking systems look at a variety of signals that align with overall page experience.

While not all of these may be directly used to inform ranking, we do find that all of these aspects of page experience align with success in search ranking, and are worth attention.

The page experience update was a concept to describe a set of key page experience aspects for site owners to focus on... It was not a separate ranking system, and it did not combine all these signals into one single "page experience" signal.

Why 这是 walk-back:2020-2021 的话术都在说"a new signal that combines...",2023 直接重写历史 —— "it was a concept, not a signal"。

同时撤掉:

2023/12/01 起 Mobile Usability 报告 + Mobile-Friendly Test 工具 + API 全部 retire;
2023/11/08 起 "Good page experience" search appearance filter 也 retire;
Page Experience 报告改成"链接到 guidance 的一个 dashboard"。

4.3 INP 替代 FID(2024/03/12)

唯一的 PE 技术性更新:Interaction to Next Paint 替换 First Input Delay 进入 CWV。

4.4 当前位置(2023+):背景指导

PE 现在的官方定位:

不是 single signal;
是 holistic guidance,跟 helpful content 同源;
是 tie-breaker;
CWV 仍然"highly recommend"但"good stats don't guarantee good rankings"。

Part 5 — Google 反复强调的核心信念(从 23 篇里出现 3 次以上的金句)

5.1 "质量先于优化"

People-first content creators focus first on creating satisfying content, while also utilizing SEO best practices to bring searchers additional value.

—— 2022/08 HCU(后续多篇引用)

5.2 "没有什么要修的"

每次 core update 公告都重复:"There might not be anything to fix at all."

5.3 "改了不保证恢复"

Do keep in mind that improvements made by site owners aren't a guarantee of recovery, nor do pages have any static or guaranteed position in our search results. If there's more deserving content, that will continue to rank well with our systems.

—— 2019/08

5.4 "Google 不像人那样理解内容,只看信号"

It's also important to understand that search engines like Google do not understand content the way human beings do. Instead, we look for signals we can gather about content and understand how those correlate with how humans assess relevance.

—— 2019/08

5.5 "raters 不直接控制 ranking,只校准算法"

Rater data is not used directly in our ranking algorithms. Rather, we use them as a restaurant might get feedback cards from diners.

—— 2019/08

5.6 "信号是多个,不暴露具体哪些"

We use many more [signals], which we don't disclose to help protect the integrity of our results.

—— 2019/08

5.7 "Topic-by-topic expertise"(2026 新加)

Our systems are designed to identify expertise on a topic-by-topic basis. So whether a site has expertise in multiple areas or has a deep focus on a single topic, there's equal opportunity to show up in Discover.

—— 2026/02 Discover core(用本地新闻站 + 园艺 section 当例子)

这是对"site reputation abuse"政策的镜像 —— 既然能识别不同子主题的 expertise,就也能识别 host 借自身 reputation 寄生的第三方内容。

Part 6 — 反向指南:Google 明确说"不会做"的事

23 篇公告里 Google 用否定句明确划出的边界:

6.1 不会公开 200+ 信号的具体清单

our algorithms rely on more than 200 unique signals or "clues"... (2016 Penguin) We use many more, which we don't disclose to help protect the integrity of our results. (2019)

6.2 不再公告每次 spam algorithm refresh

we're not going to comment on future refreshes. (2016 Penguin 实时化时)

6.3 不存在"single page experience signal"

There is no single signal. (2023/04)

6.4 不接受"3rd party 的 site authority score"

some third-party services provide "reputation" or "authority" scores for sites. These don't correspond to any of Google's own signals nor come from Google. (2024/03)

(这是对 Moz DA / Ahrefs DR 这类指标的官方否认)

6.5 没有"preferred word count"

Are you writing to a particular word count because you've heard or read that Google has a preferred word count? (No, we don't). (2022/08 HCU)

6.6 E-A-T 不是 ranking factor

Since we originally wrote this post, we have been occasionally asked if E-A-T is a ranking factor. Our automated systems use a mix of many different signals to rank great content. We've tried to make this mix align what human beings would agree is great content as they would assess it according to E-A-T criteria. (2020/03 added to 2019/08 post)

Why:E-A-T 是 rater 评估框架,不是算法信号;算法对齐人类按 E-A-T 标准的判断,不实现 E-A-T。

6.7 改进不等于一定恢复

见 5.3。Google 拒绝给"做对了就一定回来"的承诺。

6.8 Safe Browsing / Ad Experience 不是 ranking signal

we're clarifying that Safe Browsing isn't used as a ranking signal (2021/08) the Ad Experience report was never used as a factor for page experience (2021/08)

6.9 AI 内容不是"自动等于 spam"

Our long-standing spam policy has been that use of automation, including generative AI, is spam if the primary purpose is manipulating ranking. (2024/03)

"如果"是关键词 —— 用 AI 写人本位的内容不违规;用任何手段(包括纯人工)规模化堆量则违规。

Part 7 — 交叉:23 篇之外的潜叙事

读完 23 篇浮现的几条 Google 没说但隐含的叙事:

7.1 "并回 core" 是 Google 的回收机制

Penguin(2012 独立 → 2016 并回)、Product Reviews(2021 独立 → 2024 默默并回)、HCU(2022 独立 → 2024 明确并回)。Google 的方法论:用独立系统快速实验 + 用户反馈,验证有效再吸收进 core 让 site owner 无法单点优化。

7.2 公告密度反映 SEO 圈舆论压力

2022/08 HCU 之后 → 持续公告 + FAQ + 2024/03 复杂 core + 2024/08 "听到 creators 反馈" → 说明 HCU 引发了 small publisher 的大规模 backlash,Google 在持续做公关。

7.3 时间线上的"扩展 → 简化"周期

2020-2022 是 PE 信号扩展期;
2023-2026 是简化期(retire Mobile Usability / Mobile-Friendly Test / Page Experience report / 各种 SERP features / 2026/01 起从 Search Console API 砍 structured data types)。

Why:Google 周期性把"多信号 + 多报告"塌缩回"少信号 + 少报告 + 多 holistic guidance",一是减少 site owner 的"optimize 表象"空间,二是降低自身的产品维护成本。

7.4 Discover 是新独立战场(2026/02)

第一次为 Discover 单独发 broad core update —— 把 Search 上验证过的方法论(reduce clickbait + topic expertise)迁到 Discover。预示 Discover 会成为下一波 ranking 演化的实验田。

附录 — 23 篇原档清单

按时间排序,标记 stub(非英文壳)和 substantive(有内容):

#	日期	slug	类型
1	2010-06-01	caffeine-oder-mayday	stub(德文)
2	2010-06-09	our-new-search-index-caffeine	substantive
3	2016-09-23	penguin-is-now-part-of-our-core	substantive
4	2019-08-01	core-updates	substantive(母版)
5	2019-12-01	events-launching-germany-spain	stub(西/德)
6	2020-05-28	evaluating-page-experience	substantive
7	2020-11-10	timing-for-page-experience	substantive
8	2021-04-08	product-reviews-update	substantive
9	2021-04-19	more-details-page-experience	substantive
10	2021-07-26	link-tagging-and-link-spam-update	substantive
11	2021-08-04	simplifying-the-page-experience-report	substantive
12	2021-11-04	bringing-page-experience-to-desktop	substantive
13	2021-12-01	product-reviews-update-and-your-site	substantive
14	2022-03-23	product-review-ranking-one-year-on	substantive
15	2022-05-25	may-2022-core-update	substantive
16	2022-08-18	helpful-content-update	substantive
17	2022-12-14	december-22-link-spam-update	substantive
18	2023-04-19	page-experience-in-search	substantive(walk-back 转折点)
19	2023-10-04	october-2023-spam-update	substantive
20	2024-03-05	core-update-spam-policies	substantive(大融合)
21	2024-08-15	august-2024-core-update	substantive
22	2025-11-05	update-on-our-efforts	substantive(SERP 简化)
23	2026-02-05	discover-core-update	substantive(Discover 首发)