So, I went to read this feed today: https://newsblur.com/site/6885443/awol-the-ancient-world-online
The first thing I notice is, there are 10 unread items. For the past few months, this has become the “Dreaded Ten” for me. One would hope it means “there are exactly 10 unread items”, but nope, most of the time it means “Newsblur has automatically marked everything as read except for the most recent 10, and possibly deleted the older stories altogether”.
This list of “Dreaded Ten” feeds has seemed to grow for months now. At first it seemed reasonable, because it was only very busy feeds nobody could possibly keep up with, like Al Jazeera  or Map Porn  … but then it started happening to more typical feeds such as the Greenlining Institute , Richmond Confidential , Dawn News Blog  and Open Democracy .
These are not very busy feeds, so the impact is very disproportionate to whatever performance gains you’re getting from this. Take Dawn News for example: the oldest story in Newsblur is from September 24. This can’t be possible - they update every few days. If I switch to “all stories” the oldest one is STILL from September 24.
So, I thought, maybe it’s based on interaction. Maybe I don’t interact with them enough. Maybe I pressed “mark all as read” just a few too many times after only skimming the titles. I’d still object to this metric since for me, skimming titles is a valid and important way of interacting with a feed, but at least I’d understand it.
But today I find my AWOL feed has joined the Dreaded Ten, and now I’m truly baffled. I have been CONSISTENTLY opening items from this feed in new tabs, and saving them. I currently have 75 items saved in this folder. But as of today, the oldest item in this feed is from Oct 3, three days ago! How many items have I missed? Who knows.
Could you please address the algorithms involved here, perhaps even point me to the code controlling it and the git commits where you started enforcing it? Thanks for your help.