Check if domain(s) in a post are valid, preferably before posting. If the domain check fails flag the post for moderator review.
This is not a link checker.
This is not an email address checker.
Recently had a post containing an email address. As the replies added up, the OPs facts were not adding up. In a check of the domain name, could not find a registered domain name, e.g. ICANN lookup
As such the post was unlisted.
This might have very specific utility, but there are also legitimate reasons to provide invalid domain names in posts.
For example when sharing configuration here on meta you may not wish to publicly share the address of your instance. It’s also pretty common to share examples of URL structure against generic or nonexistent domains.
That sounds like a very specific set of circumstances, could you share any more?
This is bad practice, though. Those domains could exist and be something else — or maybe even worse could exist in the future. Examples should use example.com, example.org, or example.net. These are real, resolve, and are intended for this purpose.
The thing about best practice is that the uninitiated aren’t familiar. It’s borderline ivory tower stuff to create friction when posting in the context of the above. Remember that the relative ease of the standard install is such that sometimes installing Discourse might be a users first ‘sysadmin’-type task.
We can’t possibly go about verifying domain ownership before someone posts a link. If they share a config with an invalid domain how do we differentiate between a typo, improperly configured DNS, and a domain which has yet to replicate?
Not meant to be snarky.
If you mean as a moderator how do you check it manually. If the ICANN lookup fails then send a private message the person asking for details, that is what I do.
Here is the specific post
It looks like the OP left the site right after the last post when the topic was unlisted. Most users who get their topic unlisted tend to respond much sooner rather than later.
The user joined a few hours before posting and also received a few flags during that time.
Could you maybe elaborate a little on how a domain checker would have helped with the topic linked above?
The OP doesn’t include a domain, only an email address. It was your response in post 11 which referred to a .com, which incidentally is a valid registered domain - it’s parked at sedo. Had the user specified it and the domain been checked it would have returned valid whois information.
I’m not clear on how your proposed solution would have reduced moderation burden in the topic that you linked.
I see my mistake; thanks for making me do Rubber duck debugging.
As a human I recognized plausibleai as a domain but it is part of the user name. But still it would be nice to have a domain checker as users still try to pass bogus domains along.
Honestly this all strikes me as a bit more xy problem than rdd.
Nobody wants to deal with the fallout of malicious behavior, trying to entice users off-site could definitely indicate that. Discourse has a lot of protections against bulk spambot behavior, but this kind of targeted stuff still falls to community flags and responsive moderators.
If it were a deterministic problem then XY problem makes sense however with LLM and training it is worth trying. Maybe it should be a Discourse AI suggestion instead of a feature suggestion.
Still a lot of spam gets through. The spam detector mostly catches the new users who type to fast sort of spam.
As this is getting off topic will stop at this point.