About Feedscrub
Do you constantly have 1000+ unread posts in your newsreader? We've created Feedscrub to help you out.
Feedscrub is like a spam filter for your news feeds. It filters out the posts that are least interesting to you and leaves only the stuff that is relevant.
How Does It Work?
- First, tell Feedscrub which feeds you'd like to filter. In return we'll give you one feed with all your content aggregated.
- Next, for each post you're not interested in, click the "Scrub" button. The more posts you scrub, the more accurate your filter will get.
- Check your "junk feed" periodically to catch any relevant posts that we auto-scrubbed. If you find any, click the "Save" button.
Frequently Asked Questions
Q: Which feeds should I filter? All of them?!
A: We recommend you only filter feeds for which:
1) only a subset of the posts are interesting to you, and
2) consistently build up in your feed reader
For instance, you probably wouldn't filter your brother's blog because you probably want to read all of what he has to say. A global news feed, on the other hand, might be a good target for filtering if you're only interested in, say, Europe.
Q: How is it different from other news filters?
A: Other news filters tackle information overload as a quality problem. Their philosophy is that a post is only interesting if people are commenting on it or linking to it.
We believe that people are different, and that you have a unique interest in some topics that others aren't commenting on. Also we understand that you aren't interested in all the comment-worthy posts on the internet, only content that is relevant to you!
That's why you train FeedScrub with your own interests. Every time you hit the Scrub button you're giving the filter information about what topics interest you most, so it will be smarter about filtering out irrelevant posts.
Q: How does the filter work?
A: Very similarly to a spam filter, actually. When you Scrub a post about "politics", we take the words in that post and use them to train your filter. After a while your filter learns that you're not interested in "politics" and filters out posts about "politics".
For more technical information about how filters like FeedScrub's work, see this post and .
Q: Do my scrub/save ratings affect all my feeds or just the one I'm processing?
A: Your save/scrub ratings tell us about what you are interested, so they affect all your feeds. We do it this way for a few reasons, first because we believe users' interest is primarily topical, second because it reduces the amount of overall training data you need (otherwise you'd need to train 100 posts per feed!), and third it makes our database much happier since otherwise we'd have to store several distinct filters per user. That's a lot of data!
Q: Can I have feed-specific rules?
A: It's not currently possible, but if it's something you'd like to see let us know by clicking on the Feedback link on any page.
Q: Can I get some kind of view of my current filter to see if it has learned correctly?
A: We'd recommend you check out the setup filter page. It'll show you what the filter predicts you'll choose for each post so you see whether the filter is learning correctly. Got another idea for how to see this? We'd love to hear about it! Just click the feedback link on the side of any page.
Q: Can I throw a test feed at it and see what the filter decides?
A: Absolutely, just add a feed, then head to the setup filter page and look for the blue backgrounds. These tell you what your filter predicts you'll choose. Feel free to temporarily delete a feed from the manage feeds page in order to test another feed; you won't lose any data unless you explicitly reset your filter (available on the setup filter page).
Q: Is there any way to import/export ratings?
A: It's not currently possible, but if you let us know what format you'd like I'm sure we can find a way to make this happen. Please use the "feedback" button on the side of any page.
Q: How often is each feed refreshed in the Feedscrub cache?
A: It depends on the popularity of the feed, and this will change as we scale and receive feedback. If a feed has just one subscriber in Feedscrub then we currently fetch it every 3 hours. Having just two subscribers knocks that down to every hour, and when a feed has a very significant number of subscribers we fetch it as often as every 15 minutes.
If you're using a web-based reader such as Google Reader, then you might notice it takes longer than that to receive new feeds--that's because Google Reader and others have their own caches and fetch from us periodically. Let us know if you'd like a particular feed fetched more often and we'll see what we can do.
Q: I saved a post after it showed up in my junk feed; why isn't it now showing up in my saved feed?
A: Feedscrub categorizes each post you receive exactly once, as they come in from the original source. When you save something from the junk feed we take that information into account when categorizing future posts, but we don't move that item to your saved feed.
The reason we don't move saved/scrubbed items to the appropriate feed is because then you would get duplicate copies of that post. Since we're trying to help you overcome information overload, we don't want to inundate you with duplicates.
Q: What happens if I both "save" and "scrub" the same post?
A: It's easiest to answer that question with a parable... The short answer is, they cancel each other out.
Let's say the entire content of the post in question were "Green pants". Let's also say that before you save or scrub that article your filter knows you like the word "green" and doesn't know anything about "pants". When you save the article, it learns that you like "green" two times, and you like "pants" one time. Then when you scrub the article, its collective knowledge is that you: like green*2, like pants*1, dislike green*1, and dislike pants*1. So if you add those up, you're back to liking green*1 and it doesn't know anything about pants, the same as it was before.
Q: Why doesn't the "more" button work on the Train Your Filter page?
A: It's very likely that Feedscrub only has 20 or so posts cached for this feed. This happens when the feed was recently added because most sources only publish the 20 latest items to their RSS feed. The more buttons should start working as soon as we get more posts for your feed.
Didn't see your question here? , , or use this form.