Sounds to me like this is the kind of abuse blocking any site would use not just cloudflair. Do you have any evidence that Cloudflair is unique in any way in this?
That’s not a meaningful comparison. Blocking sites do indeed block differently in various different circumstances & discriminate against different groups of people. There are patterns (like Tor blocking) but the meaningful comparison is CF to inclusive sites. E.g. gnucash.org. Gnucash demonstrates how a website can be deployed in an inclusive manner that respects user’s rights.
Cloudflare is unique in how it deceives its users (e.g. tells its users they have a “zero trust” model when in fact you must trust CF with visibility on all traffic payloads). CF holds the SSL keys, unlike other implementations. The recommendation to anti-feature tag CF sites would cover the vast majority of exclusive access-restricted projects. But if a link leads to a rare Siteground site, that should also get an anti-feature tag for being exclusive.
I mention this because I am not sure not using Cloudflair would change much.
Of course it would. Cloudflare brings in a long list of problems. Not using CF (like gnucash.org does) solves all those problems of exclusivity and privacy.
You would have to use another CDN or build your own solution. Abuse is a real thing and is the reason we cannot have nice things.
The Gnucash project disproves this. Furthermore, a CF link can often be replaced with an archive.org link.
Bruce Schneiere has frequently covered data sharing between US tech giants and intelligence agencies in his blog. It’s widely accepted. To call that a “conspiracy theory” is severely out of touch, post-Snowden revelations. At best, it’s only true as a technicality (that is, the US does not admit that the Snowden leaks are real so the official narrative still differs). Bruce Schneiere concurs with Snowden’s revelations & often acknowledges in his blog that that info sharing is going on.
That said, I do not see your specific claim about the NSA in the document that I linked, which is well cited. Which paragraph number are you referring to?