Skip to main content

Replied to a post on werd.io :

That could work, but I dislike the centralisation + it presents other problems (exposes people's social graph, which is arguably private information, and I still think vouch in it's current form is gamable, but that might be a misunderstanding on my part - I only glanced through it).

What we're really talking about here is WoT; where you can assert that people you actively select as friends are at the highest level of trust (since you've explicitly asserted that they are people you want to hear from), all the way down to complete strangers, which if they've passed a captcha we could make some statement as to they're (probably) human, but can't say whether they're an asshole or not.

For the system to work, I think there needs to be more than just a binary score, which means more than a binary handling of the mention... perhaps friends are automatically accepted, while others the webmention endpoint responds with a 302 redirect to some sort of human verification page.

Perhaps a form of vouch / vouch server could be brought in as a reputation score, by way of asserting that the person sending the webmention isn't a troll/sealion/gamergater etc. Up/down votes might work by whether the recipient subsequently moderates it as abusive, these signals could be correlated across multiple feeds to provide an aggregated score that could be used as a signal for future webmentions from that domain.