search

searxng / searx-space

Statistics of the public SearX(NG) instances
https://searx.space/
GNU Affero General Public License v3.0
170 stars 27 forks source link

Rework ranking of SearXNG instances. #99

Open mrpaulblack opened 2 years ago

mrpaulblack commented 2 years ago

Currently we rank instances by TLS grade from cryptcheck, CSP grade from mozilla observatory and then the response time for a query measured by searx.space.

This does not guarantee that the best possible instance is at the top and we should think about reworking the ranking system. Since SearXNG is a fork we can completely redo this list and do for example a list with instances and not rank them at all. Also we can require more things for an instance to do, for example to not use analytics and have a constant uptime or to somehow filter instances by the quality of results.

This is just an idea, comments are welcome...

return42 commented 2 years ago

IMO it is better not rank the instances / sort engines by their location makes more sense.

The ranking has absolutely no meaning in practice, by example: a bad TLS grade (from cryptocheck, what is questionable strict) weights more than a manipulated SearXNG engine with analytics or (and this is funny) a SearXNG running on Cloudflare ..

We have users who do not know what TLS or cloudflare is, they just want to have an up-to-date running instance in their neighborhood .. with the ranking we have, we give a wrong illusion of privacy to our users.

return42 commented 2 years ago

I want to rebreathe this issue .. are there any opinions on this ... should we rework the ranking or not?