Global intelligence - in other words: picking tags and title words (either in the negative or positive) for all feeds. For example, I’m a python developer. So I’d like to be able to say that all articles with the python tag or python in the title are thumbs up. Conversely, I don’t want to see any articles about Ruby or Rails, so I’d like to be able to do a universal thumbs down for those keywords, instead of having to do it with every feed where they may appear.
10 Likes
I has considered this, but I feel that there are very, very few tags/title phrases where this would work. I greatly dislike the idea of not knowing exactly why stories get classified the way they do. I like for it to be as transparent as possible. Global classifiers would muddy this up. On top of the fact that it’s quite a bit more work to get working, while there are so few good examples of phrases where this works. I noticed that if there are multiple sites that have a tag that I dislike, I can just train the few sites to dislike that tag. And that’s to say nothing of new sites that you may not have considered when adding a global tag.
There’s issues around building a good UI to make global tags, surfacing which tags were global and which were site-specific while looking at them, and clicking a tag in the Feed view and having it turn from green to red, but not knowing if it’s global or specific.
It’s a complicated feature and since there’s far less of a benefit than the UX headache is worth, I decided not to pursue it.
That’s a reasonable answer. 
Thanks.
When I started using NewsBlur, I expected training to work globally from the get-go. For example, training a tag “audi” I’d expected would highlight in all of my feeds. It took a little while to realize the training was feed-by-feed, and I’d have to tag “audi” in 5 different feeds to get them all.
Great program - definitely fills the hole in my life left when Google ripped my Reader heart out.