Decide Which RSS Feeds to Follow

RSS is a great way to eas­ily fol­low a lot of web­sites. This arti­cle isn’t an intro­duc­tion to RSS—that’ll have to wait for another time—but if you’re new to RSS or just a lit­tle hazy on the details, I highly rec­om­mend you watch the 3-minute video RSS in Plain Eng­lish.

Even though RSS is awfully con­ve­nient, even­tu­ally even the con­ve­nience of hav­ing your favorite sites con­densed into one place can get overwhelming.

The prob­lem

For a while you’ll have the illu­sion that it doesn’t mat­ter how many feeds you fol­low since the head­lines are con­ve­niently stacked up for you in a sin­gle place. If you reach the point where you’re fol­low­ing dozens of feeds though, the wheat-to-chaff ratio will start to wear on you. You’ll find your­self scan­ning through pages of head­lines just to find a hand­ful that mat­ter to you.

And here’s what bugged me: I was pretty sure a small num­ber of feeds accounted for a large per­cent­age of the valu­able items, and another set of feeds were essen­tially spam­ming me with head­lines that I could largely do with­out. But how to tell which?

Sure, you prob­a­bly remem­ber the 3–4 feeds you’re con­stantly going back to, but it’s hard to notice and remem­ber the absence of value: off the top of my head I couldn’t eas­ily dis­tin­guish between sites that gave me good stuff infre­quently (but often enough), vs. those that I could really send to the land­fill with­out remorse.

A Failed Solution

If only I could run an exper­i­ment for a few weeks, track­ing each feed’s signal-to-noise ratio, I could chuck the losers. And while I could just keep some tally marks in a spread­sheet, that sounded awfully tedious.

So I decided to use my feed aggregator’s fea­tures to do the track­ing for me.

Solu­tion 1.0: NetVibes

Orig­i­nally I used NetVibes: each time a feed gave me some­thing I cared about, I increased the size of its win­dow by one. You can see here that NetWit’s Think Tank gave me 5 valu­able head­lines in the course of my test, which put it in the top tier of feeds.

But the NetVibes approach had draw­backs. First, it would only let me make a win­dow as large as the num­ber of items in a feed. So if NetWit’s RSS feed only included five items, I couldn’t expand the NetVibes win­dow to 6 slots. Sec­ond, if you hand NetVibes a lot of feeds to man­age it gets turtle-slow. What’s worse, over my years as a NetVibes user I’d expe­ri­enced numer­ous ser­vice out­ages, or times when the site was so slow that it might as well have been down.

So I switched to Google Reader, which lacked NetVibes’ eye candy and some fea­tures I cared about, but was blaz­ing fast and com­pletely reliable.

Google’s Offer­ing: Read­ing Trends

Google Reader pro­vides an extremely slick fea­ture that helps up solve the prob­lem. In the upper left of the page you’ll see a “Trends” menu option:

Click­ing this option gives you a table show­ing your read­ing trends orga­nized by feed—one for the items you’ve read:

and one for the items you’ve starred:

Which is very infor­ma­tive besides being highly swell. So what’s the catch?

The Catch: “Read Items” Statistics

I need to read an arti­cle to decide whether it was valu­able. Let’s say I see an item that looks promis­ing and click the title to read it. Now if I quickly decide it’s not for me… Google Reader still con­sid­ers it a Read item. The fact that I’ve “read” an item (i.e., clicked on its title) really just means it had a tempt­ing title, not that it included beloved content.

No Catch: “Starred Items” Statistics

Assum­ing you use the star to mean “I read this item and it was use­ful”, the Starred Items trends give us exactly (mostly) what we need.

Here’s the approach.

1. Star things you care about

Every time a feed includes an item that you’re glad you’ve read, star it.

2. Tally the results

When you’ve been run­ning this for a while—at least 4 weeks, but 8 would be bet­ter in case a par­tic­u­lar feed is in a slump—check the Trends option using the Starred view.

3. Sift and Winnow

Notice those trash can icons on the right side of the fig­ure above? They’re great for this step: they drop a feed from your Reader account.

Pick some thresh­olds num­bers: the num­ber below which you’ll send a feed into the hop­per, and the num­ber above which you con­sider a feed to be golden.

For exam­ple, my num­bers were 1 and 6 for a 6-week run. If a feed gave me 6 or more items I val­ued in a 6-week period, it got spe­cial con­sid­er­a­tion (see below). On the other hand, if a feed only had 0 or 1 items of inter­est in a 6-week period, I sent it pack­ing, but I gave a stay of exe­cu­tion to any feed where:

• there was just one post I cared about, but it was awe­some, or
• the feed had very few posts, so it gen­er­ated very lit­tle noise for me to wade through, or
• I just liked the site behind the feed.

So using those rules, A List Apart got a stay of exe­cu­tion: it’s an author­i­ta­tive site; it pub­lishes infre­quently (2 posts per month); I like the site a lot; and even though their two posts dur­ing my test weren’t rel­e­vant to me, pre­vi­ous posts have been excellent.

4. Ele­vate Uberfeeds

Some feeds were so con­sis­tently use­ful that I wanted to put them in a spe­cial group: if I had just a few min­utes to check feeds, I’d check that group­ing first. (See the “Where’s the Beef” side­bar for a list.) Here, Wild Apri­cot makes the move from “Other”:

…to “Uber”:

5. Main­tain

A cou­ple of times a year run through the exer­cise again. Of course feeds you’ve added more recently will be at a dis­ad­van­tage since they’ll have fewer starred items, but you can take that into account when you weed out the chaff.

Wrap­ping Up

I’ve done this exer­cise three times over the last cou­ple of years and I always find a few feeds that were more use­ful than I’d expected (and vice versa), so I always end up drop­ping a few and ele­vat­ing a few. I hope it’ll make your feed read­ing eas­ier too.