fep

FEP-0391: Special collection proofs

Summary

Some properties represent special collections, such as:

Verifying that any given object is part of a special collection is usually only possible by resolving that collection and checking its items one-by-one until the current object is found. This can be inefficient to verify. It would be easier if there was an inverse claim for each claim made about an object being part of a special collection. This FEP aims to define some properties that can be used to make those inverse claims.

Mapping claims and inverse claims

Given the special collections above, we can map the following claims and inverse claims. The last two inverse claims are particularly salient, since inclusion in replies and/or context carries additional semantic meaning and is a socially loaded relation used to establish conversational constructs.

Verifying presence in outbox

Claim
Activity <A>’s actor (<A>.actor) is actor <B> Inverse claim
Actor <B>’s outbox collection (<B>.outbox) contains activity <A>

This is not particularly useful to prove.

Verifying presence in inbox

Claim
Activity <A> has to/cc/audience including actor <B> Inverse claim
Actor <B>’s inbox collection (<B>.inbox) contains activity <A>

This is not particularly useful to prove, and at best can only be implied if <B> is addressed directly and not through some collection that triggers inbox forwarding.

Verifying presence in followers

Claim
Actor <B>’s following collection (<B>.following) contains actor <A>
Equivalent claim
Actor <B> claims that <B> is following <A>
Another equivalent claim
Actor <B> claims that <A> is followed by <B>
Inverse claim
Actor <A>’s followers collection (<A>.followers) contains actor <B>
Equivalent inverse claim
Actor <A> claims that <A> is followed by <B>
Another equivalent inverse claim
Actor <A> claims that <B> is following <A>

This can be verified by showing one of the following:

Verifying presence in following

Claim
Actor <B>’s followers collection (<B>.followers) contains actor <A>
Equivalent claim
Actor <B> claims that <B> is followed by <A>
Another equivalent claim
Actor <B> claims that <A> is following <B>
Inverse claim
Actor <A>’s following collection (<A>.following) contains actor <B>
Equivalent inverse claim
Actor <A> claims that <A> is following <B>
Another equivalent inverse claim
Actor <A> claims that <B> is followed by <A>

This can be verified by showing one of the following:

Verifying presence in liked

Claim
Actor <A>’s liked collection (<A>.liked) contains object <O>

This can be verified by showing a trusted activity exists in <O>.likes where:

Note that there is an issue that may occur if <A> issues multiple Like activities for the same object <O>, and then issues any Undo Like activities at a later point in time. The most recent activity will have its side-effects carried out. It is possible for some of these functionally duplicate Like activites to remain in <O>.likes even though the object <O> is no longer in <A>.liked. See https://github.com/w3c/activitypub/issues/381 for more information.

Verifying presence in likes

Claim
Object <O>’s likes collection (<O>.likes) contains a Like activity <L> from actor <A>

If <L> is not already trusted via some other mechanism, this can be verified by showing that <A>.liked includes <O>. Note the duplication issue from the previous section.

The Like activity <L> MAY result in an Add activity <R> targeting <O>.likes. This result can be used by <A> as proof of the claim, if it can be shown to be a trusted activity.

Verifying presence in shares

Claim
Object <O>’s shares collection (<O>.shares) contains an Announce activity <S> from actor <A>

The Announce activity <S> MAY result in an Add activity <R> targeting <O>.shares. This result can be used by <A> as proof of the claim, if it can be shown to be a trusted activity.

Verifying presence in replies

Claim
Object <R> is inReplyTo object <O>
Inverse claim
Object <O>’s replies collection (<O>.replies) contains object <R> attributed to actor <A>

The inverse claim can be verified by showing that an activity exists where:

Verifying presence in context

Claim
Object <O> has a context pointing to context collection <C>
Inverse claim
Context collection <C> contains object <O>

The inverse claim can be verified by showing that an activity exists where:

Types of proofs

We can map proofs onto the result property provided that its subject is an activity and its referent is a trusted activity. We may also define some extension properties for non-activity objects.

Showing inclusion in replies collections via a reply stamp

Provides proof that some object was added to some replies collection.

On an activity where inReplyTo has been set, the proof can be verified if all of the following are satisfied:

On a Create activity where the object has inReplyTo set, the proof can be verified if all of the following are satisfied:

On a non-activity Object where inReplyTo has been set, we define the extension property inReplyToProof since the use of result is invalid on non-Activity types. The proof is valid if all of the following are satisfied:

Example of an activity with inReplyTo set to some activity with a replies collection:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-activity",
	"actor": "https://example.com/actors/2",
	"type": "Activity",
	"object": "https://example.com/some-object",
	"inReplyTo": {
		"id": "https://example.com/some-other-create",
		"actor": "https://example.com/actors/1",
		"type": "Create",
		"object": "https://example.com/some-other-object",
		"content": "I am accepting replies to this activity.",
		"replies": "https://example.com/some-other-create/replies"
	},
	"result": {
		"id": "https://example.com/some-proof",
		"actor": "https://example.com/actors/1",
		"type": "Add",
		"object": "https://example.com/some-activity",
		"target": "https://example.com/some-other-create/replies",
		"attributedTo": "https://example.com/some-activity"
	}
}

Example of a Create activity with object.inReplyTo set to some non-activity object with a replies collection:

{
	"@context": ["https://www.w3.org/ns/activitystreams", "https://w3id.org/fep/0391"],
	"id": "https://example.com/create-some-reply",
	"actor": "https://example.com/actors/2",
	"type": "Create",
	"object": {
		"id": "https://example.com/some-reply",
		"type": "Note",
		"attributedTo": "https://example.com/actors/2",
		"content": "This is a reply, and I can prove it was added to the replies collection.",
		"inReplyTo": {
			"id": "https://example.com/some-object",
			"type": "Note",
			"attributedTo": "https://example.com/actors/1",
			"content": "I am accepting replies to this object.",
			"replies": "https://example.com/some-object/replies"
		},
		"inReplyToProof": "https://example.com/some-proof"
	},
	"result": {
		"id": "https://example.com/some-proof",
		"actor": "https://example.com/actors/1",
		"type": "Add",
		"object": "https://example.com/some-reply",
		"target": "https://example.com/some-object/replies",
		"attributedTo": "https://example.com/create-some-reply"
	}
}

Showing inclusion in context collections via a context stamp

Provides proof that some object was added to some context collection.

On an activity where the context has been set to a collection, the proof can be verified if all of the following are satisfied:

On a Create activity where the object has context set, the proof can be verified if all of the following are satisfied:

On a non-activity Object where context has been set, we define the extension property contextProof since the use of result is invalid on non-Activity types. The proof is valid if all of the following are satisfied:

Example of an activity with context set to some owned collection:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-activity",
	"actor": "https://example.com/some-actor",
	"type": "Activity",
	"object": "https://example.com/some-object",
	"context": {
		"id": "https://example.com/some-context",
		"type": "Collection",
		"attributedTo": "https://example.com/some-context-moderator"
	},
	"result": {
		"id": "https://example.com/some-proof",
		"actor": "https://example.com/some-context-moderator",
		"type": "Add",
		"object": "https://example.com/some-activity",
		"target": "https://example.com/some-context",
		"attributedTo": "https://example.com/some-activity"
	}
}

Example of a Create activity with object.context set to some owned collection:

{
	"@context": ["https://www.w3.org/ns/activitystreams", "https://w3id.org/fep/0391"],
	"id": "https://example.com/create-some-object",
	"actor": "https://example.com/some-actor",
	"type": "Create",
	"object": {
		"id": "https://example.com/some-object",
		"type": "Note",
		"attributedTo": "https://example.com/some-actor",
		"content": "This object is part of some context, and I can prove it was added to the context collection.",
		"context": {
			"id": "https://example.com/some-context",
			"type": "Collection",
			"attributedTo": "https://example.com/some-context-moderator"
		},
		"contextProof": "https://example.com/some-proof"
	},
	"result": {
		"id": "https://example.com/some-proof",
		"actor": "https://example.com/some-context-moderator",
		"type": "Add",
		"object": "https://example.com/some-object",
		"target": "https://example.com/some-context",
		"attributedTo": "https://example.com/create-some-object"
	}
}

Showing inclusion in likes collections via a like stamp

Provides proof that the current activity was added to the object.likes collection.

On a Like activity where the object has a likes collection, the proof can be verified if all of the following are satisfied:

Example:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-like",
	"summary": "A Like activity, with proof that it was added to the likes collection.",
	"actor": "https://example.com/actors/2",
	"type": "Like",
	"object": {
		"id": "https://example.com/some-object",
		"type": "Note",
		"content": "I am accepting likes of this object.",
		"likes": "https://example.com/some-object/likes",
		"attributedTo": "https://example.com/actors/1"
	},
	"result": {
		"id": "https://example.com/some-proof",
		"actor": "https://example.com/actors/1",
		"type": "Add",
		"object": "https://example.com/some-like",
		"target": "https://example.com/some-object/likes",
		"attributedTo": "https://example.com/some-like"
	}
}

Showing inclusion in shares collections via a share stamp

Provides proof that the current activity was added to the object.shares collection.

On an Announce activity where the object has a shares collection, the proof can be verified if all of the following are satisfied:

Example:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-announce",
	"summary": "An Announce activity, with proof that it was added to the shares collection.",
	"actor": "https://example.com/actors/2",
	"type": "Announce",
	"object": {
		"id": "https://example.com/some-object",
		"type": "Note",
		"content": "I am accepting shares of this object.",
		"shares": "https://example.com/some-object/shares",
		"attributedTo": "https://example.com/actors/1"
	},
	"result": {
		"id": "https://example.com/some-proof",
		"actor": "https://example.com/actors/1",
		"type": "Add",
		"object": "https://example.com/some-announce",
		"target": "https://example.com/some-object/shares",
		"attributedTo": "https://example.com/some-announce"
	}
}

Relationship proofs

[!WARNING] Experimental, requires further thought.

[!WARNING] Currently bugged. See https://github.com/w3c/activitystreams/issues/593 for more details.

Provides proof that the current relationship is reciprocally claimed.

Relationships other than following or being a follower may be proved using this property, but the requirements for such a proof are out of scope of this FEP.

We define the extension property relationshipProof since the use of result is invalid on non-Activity types. The proof is valid if all of the following are satisfied:

Example that proves a user is following another user:

{
	"@context": ["https://www.w3.org/ns/activitystreams", "https://w3id.org/fep/0391"],
	"id": "https://example.com/some-relationship",
	"type": "Relationship",
	"attributedTo": "https://example.com/actors/1",
	"subject": {
		"id": "https://example.com/actors/1",
		"following": "https://example.com/actors/1/following"
	},
	"relationship": "IsFollowing",
	"object": {
		"id": "https://example.com/actors/2",
		"followers": "https://example.com/actors/2/followers"
	},
	"relationshipProof": [
		{
			"id": "https://example.com/not-enough-proof",
			"actor": "https://example.com/actors/1",
			"type": "Add",
			"object": "https://example.com/actors/2",
			"target": "https://example.com/actors/1/following"
		},
		{
			"id": "https://example.com/proof-by-inverse-relationship",
			"type": "Relationship",
			"attributedTo": "https://example.com/actors/2",
			"subject": "https://example.com/actors/2",
			"relationship": "IsFollowedBy",
			"object": "https://example.com/actors/1"
		},
		{
			"id": "https://example.com/proof-by-being-added-to-followers",
			"actor": "https://example.com/actors/2",
			"type": "Add",
			"object": "https://example.com/actors/1",
			"target": "https://example.com/actors/2/followers"
		},
		{
			"id": "https://example.com/proof-by-having-follow-accepted",
			"actor": "https://example.com/actors/2",
			"type": "Accept",
			"object": {
				"actor": "https://example.com/actors/1",
				"type": "Follow",
				"object": "https://example.com/actors/2"
			}
		}
	]
}

Example that proves a user is followed by another user:

{
	"@context": ["https://www.w3.org/ns/activitystreams", "https://w3id.org/fep/0391"],
	"id": "https://example.com/some-relationship",
	"type": "Relationship",
	"attributedTo": "https://example.com/actors/1",
	"subject": {
		"id": "https://example.com/actors/1",
		"followers": "https://example.com/actors/1/following"
	},
	"relationship": "IsFollowedBy",
	"object": {
		"id": "https://example.com/actors/2",
		"following": "https://example.com/actors/2/followers"
	},
	"relationshipProof": [
		{
			"id": "https://example.com/not-enough-proof",
			"actor": "https://example.com/actors/1",
			"type": "Add",
			"object": "https://example.com/actors/2",
			"target": "https://example.com/actors/1/followers"
		},
		{
			"id": "https://example.com/proof-by-inverse-relationship",
			"type": "Relationship",
			"attributedTo": "https://example.com/actors/2",
			"subject": "https://example.com/actors/2",
			"relationship": "IsFollowing",
			"object": "https://example.com/actors/1"
		},
		{
			"id": "https://example.com/proof-by-being-added-to-following",
			"actor": "https://example.com/actors/2",
			"type": "Add",
			"object": "https://example.com/actors/1",
			"target": "https://example.com/actors/2/following"
		},
		{
			"id": "https://example.com/proof-by-having-follow",
			"actor": "https://example.com/actors/2",
			"type": "Follow",
			"object": "https://example.com/actors/1"
		}
	]
}

Miscellaneous examples:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-follow",
	"actor": {
		"id": "https://example.com/actors/2",
		"following": "https://example.com/actors/2/following"
	},
	"type": "Follow",
	"object": {
		"id": "https://example.com/actors/1",
		"followers": "https://example.com/actors/1/followers"
	},
	"result": {
		"id": "https://example.com/accept-follow",
		"actor": "https://example.com/actor/1",
		"type": "Accept",
		"object": "https://example.com/some-follow",
		"result": [
			{
				"id": "https://example.com/resulting-add-to-followers",
				"actor": "https://example.com/actors/1",
				"type": "Add",
				"object": "https://example.com/actors/2",
				"target": "https://example.com/actors/1/followers",
				"attributedTo": "https://example.com/accept-follow"
			},
			{
				"id": "https://example.com/resulting-add-to-following",
				"actor": "https://example.com/actors/2",
				"type": "Add",
				"object": "https://example.com/actors/1",
				"target": "https://example.com/actors/2/following",
				"attributedTo": "https://example.com/accept-follow"
			}
		]
	}
}

Existing mechanisms of trust

Verifying stamps has an issue with bootstrapping trust. In addition to verifying stamps via logical inference, ActivityPub clients SHOULD also consider the following:

Obtaining and revoking stamps

Upon receiving an activity with a side effect of adding something to a special collection, ActivityPub servers SHOULD generate and deliver an Add activity representing this side-effect. The ActivityPub server MAY require manual action by a user. The resulting stamp activity SHOULD be attributedTo the activity that caused the side-effect, delivered to its actor, and additionally SHOULD either be resolvable (to allow direct same-origin checking) or otherwise include a cryptographic proof. If a cryptographic proof is included, the proof SHOULD expire after some reasonable window, beyond which point a new proof should be issued. If the resulting stamp activity is resolvable, it MAY become unresolvable after some time; HTTP caching SHOULD be used to indicate a time-to-live for ActivityPub clients to re-check cached stamps. The time-to-live MAY be used as the duration of the cryptographic proof, if one exists.

The recipient of a stamp SHOULD Update their object to include the appropriate property for the stamp – result if it is an activity, inReplyToProof if it is a non-activity object that declares inReplyTo, contextProof if it is a non-activity object that declares context. [TODO: how to handle relationship proofs?]

Revoking a stamp can be done by:

A simpler example flow for issuing and revoking a like stamp

This flow is similar for any stamp that uses result on an activity – Like, Announce, Activity with inReplyTo, Activity with context.

Actor 2 likes a post by actor 1:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-like",
	"summary": "Actor 2 liked a Note by actor 1",
	"actor": {
		"id": "https://example.com/actors/2",
		"followers": "https://example.com/actors/2/followers"
	},
	"type": "Like",
	"object": {
		"id": "https://example.com/some-object",
		"type": "Note",
		"content": "I am accepting likes of this object.",
		"likes": "https://example.com/some-object/likes",
		"attributedTo": {
			"id": "https://example.com/actors/1",
			"followers": "https://example.com/actors/1/followers"
		},
		"cc": [
			"https://example.com/actors/1/followers",
			"as:Public"
		],
		"audience": "https://example.com/some-object/audience"
	},
	"to": "https://example.com/actors/1",
	"cc": [
		"https://example.com/actors/2/followers",
		"https://example.com/some-object/audience",
		"as:Public"
	],
	"audience": "https://example.com/some-like/audience"
}

Actor 1 sends a like stamp:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-proof",
	"summary": "Actor 1 approved a like",
	"actor": "https://example.com/actors/1",
	"type": "Add",
	"object": "https://example.com/some-like",
	"target": "https://example.com/some-object/likes",
	"attributedTo": "https://example.com/some-like",
	"to": "https://example.com/actors/2",
	"cc": [
		"https://example.com/actors/2/followers",
		"https://example.com/some-like/audience",
		"as:Public"
	]
}

Actor 2 updates their Like activity with proof:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-update",
	"summary": "Actor 2 updated their Like with proof",
	"actor": {
		"id": "https://example.com/actors/2",
		"followers": "https://example.com/actors/2/followers"
	},
	"type": "Update",
	"object": {
		"id": "https://example.com/some-like",
		"summary": "Actor 2 liked a Note by actor 1, with proof",
		"actor": "https://example.com/actors/2",
		"type": "Like",
		"object": "https://example.com/some-object",
		"to": "https://example.com/actors/1",
		"cc": [
			"https://example.com/actors/2/followers",
			"https://example.com/some-object/audience",
			"as:Public"
		],
		"audience": "https://example.com/some-like/audience",
		"result": "https://example.com/some-proof"
	},
	"cc": [
		"https://example.com/actors/2/followers",
		"https://example.com/some-like/audience",
		"as:Public"
	]
}

If actor 1 had provided cryptographic proof, actor 2 would be able to authoritatively serve the stamp until the proof expired. Actor 1 can silently revoke the stamp by making it no longer resolve, or actively revoke the stamp by issuing a revocation activity [TODO: Update with “undoneBy” pointing to a Remove?]

A more complex flow for stamps involving embedded objects

This flow uses inReplyToProof or contextProof on the object of a Create instead of using only result on the activity itself.

Some actor creates some object that is part of some context, owned by some context moderator:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/create-some-object",
	"actor": "https://example.com/some-actor",
	"type": "Create",
	"object": {
		"id": "https://example.com/some-object",
		"type": "Note",
		"attributedTo": {
			"id": "https://example.com/some-actor",
			"followers": "https://example.com/some-actor/followers"
		},
		"content": "This object is part of some context, and I can prove it was added to the context collection.",
		"context": {
			"id": "https://example.com/some-context",
			"type": "Collection",
			"attributedTo": "https://example.com/some-context-moderator",
			"audience": "https://example.com/some-context/audience"
		},
		"to": "https://example.com/some-context-moderator",
		"cc": "as:Public"
	},
	"to": [
		"https://example.com/some-context-moderator",
		"https://example.com/some-context/audience"
	],
	"cc": "as:Public",
	"audience": "https://example.com/create-some-object/audience"
}

The context moderator adds the object to the context:

{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/some-proof",
	"summary": "A new post was added to the context",
	"actor": "https://example.com/some-context-moderator",
	"type": "Add",
	"object": "https://example.com/some-object",
	"target": "https://example.com/some-context",
	"attributedTo": "https://example.com/create-some-object",
	"cc": [
		"https://example.com/some-context/audience",
		"https://example.com/some-actor",
		"https://example.com/create-some-object/audience",
		"as:Public"
	]
}

The actor who created the object now can issue two updates: one for the Create activity to add a result, and one for the object of that activity to add object.contextProof:

{
	"@context": ["https://www.w3.org/ns/activitystreams", "https://w3id.org/fep/0391"],
	"id": "https://example.com/update-some-object",
	"actor": {
		"id": "https://example.com/some-actor",
		"followers": "https://example.com/some-actor/followers"
	},
	"type": "Update",
	"object": {
		"id": "https://example.com/some-object",
		"type": "Note",
		"attributedTo": "https://example.com/some-actor",
		"content": "This object is part of some context, and I can prove it was added to the context collection.",
		"context": "https://example.com/some-context",
		"contextProof": "https://example.com/some-proof",
		"to": "https://example.com/some-context-moderator",
		"cc": "as:Public"
	},
	"to": [
		"https://example.com/some-actor/followers",
		"as:Public"
	]
}
{
	"@context": "https://www.w3.org/ns/activitystreams",
	"id": "https://example.com/update-create",
	"actor": "https://example.com/some-actor",
	"type": "Update",
	"object": {
		"id": "https://example.com/create-some-object",
		"actor": "https://example.com/some-actor",
		"type": "Create",
		"object": "https://example.com/some-object",
		"to": "https://example.com/some-context-moderator",
		"cc": "as:Public",
		"result": "https://example.com/some-proof"
	}
}

References

CC0 1.0 Universal (CC0 1.0) Public Domain Dedication

To the extent possible under law, the authors of this Fediverse Enhancement Proposal have waived all copyright and related or neighboring rights to this work.