Webmentions, Privacy and Consent
Discourse between a few IndieWeb bloggers about the different ways to use webmentions recently caught my interest.
Webmentions is the name given to tooling coded into a blog such as this one or another web site, that is typically aligned to the concept of the IndieWeb or fediverse. The tooling enables a user to participate in a small social layer on the web without recourse to any other platform.
A webmention is either sent or received from a site that has the tooling baked in. They are sent from one such site to another via a link on the sending site to content on the receiving site. When a link like this is created the sending site sends out a signal to the receiving site that carries a notification of the nature and, where applicable, any content that’s related to that link. This could for instance be a reply, a like, a repost, an RSVP, bookmark or mention. It’s down to the receiver what they do with that notification. They could pay it no mind, they could publish it on their site - partially or in whole, attributed or not, or they could keep it private.
The tooling can be implemented in various different ways but whatever means is used it must adhere to the Webmention Framework. The framework essentially sets out protocols that either must or should be followed for the tooling to work as intended.
The framework is not prescriptive about how received webmentions are used - on that front decisions are down to each recipient. Consequently how they are used differs between sites and that has inevitably led to the discourse mentioned.
I have shared a link to this blog post on a Mastodon instance. Mastodon itself has no webmention tooling. Instead, a bridging service sits between the fediverse and the IndieWeb. The bridging service notices when someone interacts with a post that links here, translates that interaction into a webmention, and delivers it to webmention.io on my behalf. My blog then picks it up when it next builds and polls the inbox. Someone may follow that link to this post and decide to comment on, like, share or bookmark it, and in doing so set that relay in motion without knowing it exists. The person interacting with my post may have no idea or understanding of any of that. They chose to interact with a post on a social platform whose norms they signed up to. They did not choose to have that interaction relayed across a bridge and published on somebody’s personal website.
If my blog had no webmention tooling, the relayed signal would disappear into the ether. Since I do, this raises points of contention.
Not discounting the interaction relayed from Mastodon by the bridging service, picked up by webmention.io and passed on to me without so much as a by your leave, if I then publish said interaction on this blog I am doing so without the explicit consent of the person that made it. I do not think there is much if any argument to say that consent can be implied here. Whether the person cares or not what I do with their interaction is not the point in question. The question here is about consent.
In practice I am to all intents and purposes ignoring that question since I am currently publishing all webmentions I receive. Since I have dismissed concerns about consent I am also quite effectively disregarding the notion of privacy. I am publicly sharing the interactions of others in whatever way I choose to use received webmentions. I display links and author but omit content. Regardless as I have ignored consent it would be facile to suggest I am paying anything more than lip service to rights for privacy.
As I understand it, the core purpose of webmentions is to enable a user to participate in a small social layer on the web without recourse to any other platform. Webmentions don’t forbid platform-sourced signals, and bridges carrying them across are a legitimate part of the system. However, I think the spirit of the thing is person to person. In the Mastodon example above, the person doing the interacting is one step removed from all of that, and that feels just a little bit off to me.
What though would using webmentions with consent, and which respect privacy look like?
An exchange between sites where the respective owners have either baked in the webmention tooling themselves or consent to the sending of webmentions on a site they have chosen to use? Keeping received webmentions private? Displaying reactions and replies without attribution or links? Publicly noting a reply with attribution and a link back to the source but without publishing the content of that reply?
These are all issues considered and discussed by the likes of Quirk, Moreale, Fink, Belvadi, and Ben.
They all appear to have decided on the matter and followed a course of action that aligns with each of their respective decisions. I on the other hand have not. Instead I have gone on a voyage of discovery and learning, implemented webmentions on this site and not yet been guided by the points articulated in this post. Indulgence over self mastery for the time being.
If you’re still reading and interested here’s how this site currently sends, receives and uses webmentions. I think it shows there is some hope.
No promises but things may change in the future now I’ve thought things through a bit more. I intend to continue displaying replies, quotes and mentions where, despite what I’ve said above, there is perhaps at minimum a reasonable implied-consent argument. Likes, bookmarks and shares carry no conversational content and given I have attributed them are where consent is thinnest. A filter for these is pending. In the meantime, anyone who would rather not appear here for any reason should let me know and I will take the mention down at the next build.