Prevent unnecessary outbox activities from third-party plugin post updates

[pfefferle]

Problem

Third-party plugins (editorial calendars, social schedulers, content converters, hosting sync tools) frequently call wp_update_post() on published posts without meaningfully changing the content. Each call fires wp_after_insert_post, which creates an outbox entry — even when the post content hasn't actually changed.

This results in:

• Excessive Update activities sent to followers for unchanged posts
• Spurious Delete activities when plugins briefly transition post status
• Outbox flooding at rates of several activities per minute during bulk operations
• Old posts (years old) being re-federated unnecessarily

Contributing factors

1. Plugins calling wp_update_post() remotely or in batch — e.g., cloud-based editorial calendars syncing post metadata, content migration tools converting classic posts to blocks, search-and-replace operations
2. Non-deterministic the_content output — plugins injecting HTML with random values (nonces, embed secrets, random player IDs) into the_content filter output, making content appear different on every render even when the actual post content is identical. Examples:
   • WordPress core's get_post_embed_html() generates a fresh #?secret= on every call
   • Podcast player plugins inject random player_id values via wp_rand()
   • Subscription/paywall blocks rendering dynamic URLs
3. No deduplication gate — activity type is currently determined based solely on post status transitions, without checking whether the federated content actually changed

Proposed solution

Add a lightweight content-change detection step before creating outbox entries for Update activities. When an Update is determined, compare a hash of the relevant post properties against the previously federated version before writing to the outbox.

Considerations

• Latency budget is tight — this runs on every wp_after_insert_post, so the check must be fast (hash comparison, not full object transformation)
• Storage — could store a content hash in post meta (e.g., _activitypub_content_hash) that gets set when an activity is actually dispatched
• What to hash — should include the fields that matter for federation: post_content, post_title, post_excerpt, post_status, key meta fields. Should NOT include rendered the_content output (which is non-deterministic)
• Create vs Update — this check should only apply to Update activities; Create and Delete should proceed unconditionally
• Escape hatch — provide a filter so plugins can bypass the check if needed

Alternative approaches

• Rate-limiting outbox entries per post (e.g., max 1 Update per post per N minutes)
• Debouncing via short-lived transients before scheduling the outbox write
• Comparing post_modified_gmt against a stored "last federated" timestamp

[pfefferle]

FYI: @cagrimmett @jmdodd

[jeherve]

Add a lightweight content-change detection step before creating outbox entries for Update activities.

On WordPress.com and in wpcomsh, we have wpcom_post_has_changed_since_last_revision, which I think could be a good solution ; it limits the checks to the few fields we really care about, that would mean that there was a significant change:
https://github.com/Automattic/jetpack/blob/ad6c6463b0a060f38e46cecaaa3ddd9d6cc40bfd/projects/plugins/wpcomsh/feature-plugins/autosave-revision.php#L19

[pfefferle]

@jeherve I am not sure that this is enough, because it ignores tags, visibility changes and other specialties that comes with the plugin. I generally like the idea, but I am not sure it is so trivial.

I tend to favor #2932

[jeherve]

it ignores tags, visibility changes and other specialties that comes with the plugin

Ah yes, you'd have to account for that too.

I tend to favor #2932

👀

[pfefferle]

@jeherve I also have two other related PRs if you find some time:

• Fix purge methods to handle large collections without OOM or timeout #2929
• Add Site Health test to detect excessive outbox activity #2928