A couple of potential bugs:
- Training on a text that is a URL does not appear to remove the story from the feed
- When clicking on a page with trained text, a large thumb is disabled and clicking makes it multiply.
(text view, Firefox)
Regarding item #1 specifically; I am seeing the same behavior on other sites without it being a link.
Thanks for reporting this, @B111 (is that Bill in 1337speak?). Just deployed a fix for the multiplying thumb issue.
There were actually three things going wrong:
- The thumb icons inside story content were rendered large because the story content CSS (width: auto !important) was overriding the icon’s intended 10x10px size. Fixed by adding a more specific CSS rule for score icons inside story content.
- When mark.js removes its highlight wrappers (unmark()), it preserves all child elements, including the appended thumb icons. So each time highlights were re-applied, another icon got added on top of the orphaned ones. Fixed by cleaning up those orphaned icons after unmarking.
- Clicking on the thumb icon itself (rather than the highlighted text around it) wasn’t recognized as a click inside a classifier highlight, so it triggered an unnecessary re-mark cycle, which made the duplication happen on every click.
For #1 (trained text not removing the story), training text as negative will hide stories only when you’re viewing in Focus mode (the green circle). In All Stories or Unread mode, negatively scored stories still appear but with a red bar on the left. Could you check whether that matches what you’re seeing?
Thanks Sam!
For my username, I think it was just random button mashing generated - however has been years 
It does appear these items items that you noted are fixed, as cannot reproduce the icon issue or it regenerating when clicked. Thanks!
Regarding the training text not removing from the story, it sounds like maybe this is my confusion with the feature. I’m use to title based training automatically hiding, but it sounds like maybe a feature difference with text based. Here is a visual:
That’s because you’re looking at the text view, not the feed view. I can only train on the feed view. I don’t actually get the text view until you are actually requesting it in real time.