Allow filtering out medium.com sites with custom domains

knazarov

medium.com requires a sign-up to read its contents, and as such I'm filtering out all medium.com results.
However, some sites use the "own domain" feature of medium, where they just point the DNS "CNAME" records to medium.com. This still means that they are "login-walled", but bypass kagi filters (and you have to cherry-pick every time).

I wonder whether it's possible to create a slightly more flexible blocklist that is based on what the website actually "is". At the minimum, being able to block based on a "real domain" would be nice.

Vlad

knazarov How would an interface for this look like?

buteo

I wonder if a working solution to your request might also help with nailing down general bothers like Pinterest and its small army of domains: https://www.netify.ai/resources/applications/pinterest

Recast

I'd like this too. I generally find things published via Medium to be low-quality, so I have that domain de-prioritised lower (but still present at the bottom, on the off chance that it turns out to be a good article). As an example, a Medium custom domain I have lowered is javascript[.]plainenglish[.]io.

knazarov

Vlad

Ideally, the functionality shouldn't "clobber" the normal flow and introduce additional controls if the site doesn't have aliases. So here's how I see it:

The user clicks on the "shield" icon to the right of the item in the result set
If the site has less than a certain threshold number of aliases/CNAMEs (say, 10), the UI is not changed in any way
If there are more than the threshold number of aliases:
- Under the normal "Website ranking adjustment for <X>", there appears a toggle to "apply the rule to aliases as well"
- Near the toggle there's an "information" icon on what an alias is
- An estimated number of known aliases is shown. For instance "apply the rule to <n> known aliases"

The same can be done via the profile settings dialog as well if the user wishes to change their ruleset.

Recast

Another network of sites that would benefit from this "alias filtering" is hubpages.com and its sub-sites like pethelpful.com and turbofuture.com, which largely appear to be AI-generated sludge. In "view source" on any of the sub-sites you can see it includes scripts from hubpages, so it should be technically possible to detect these and filter them out.