A person leans on the counter of a carnival booth for a game where you throw darts to win posters. There's a wide array of posters behind them including Prince, Katie Perry, and Scooby Doo

I like Feedly as an RSS back-end a lot. There are other RSS services that offer keyword filtering, but Feedly goes beyond that.

I’ve seen people dis Feedly because they don’t like its attempt to popularize RSS as a research tool for work as opposed to a convenient way to aggregate sources of information for a personal interest. Others have a reaction to its focus on marketing research (even though it is moving beyond that niche).

When I come at it from the perspective of a former tech journalist and former marketing content lead, its purpose in life is clarified. Rather than seeing it as a weird way to make a niche, personal technology popular, it’s better to see it as a way to use automation to bring the benefits of a clipping service to people who don’t have the departmental or personal budget to pay for one. It is also happy to make money off people who just love RSS.

I stopped using it when I went through a period where I was trying to cut down on inputs a little more, and its mobile client was always frustrating and a bit buggy. Lately, though, I’ve been sticking more stuff in my RSS reader, especially as things like ooh.directory come back around.

Feedly’s filtering is better than anyone else’s, because it goes past keywords. There is some interesting stuff going on behind the scenes, including something that taxonomizes every article that passes through their system. That’s a boon for someone using it as a souped-up clipping service, because the world changes from “this is a list of sources I know about, look for my keywords” to “I’m interested in these topics, bring me anything related to them from across the breadth of the feeds you, service, know about.”

The people who write most RSS readers tend to treat RSS as a way to save yourself a bunch of visiting sites every day or as a way to avoid the bad design and ads of the sites' layouts. They’re not interested in writing or maintaining a back-end. In the Apple ecosystem, iCloud backend syncing is the most interesting thing in RSS readers because it gives people the benefit of having the read/unread state of all their feeds in sync. It “frees” them from the RSS backend services (e.g. Feedly, the Old Reader, Inoreader, etc.) and allows them to focus on the age-old RSS use case of quietly hoarding feeds in a reader and maybe sharing OPML files with others.

One annoying side effect of this approach to RSS is the recreation of the “blog roll” pattern, which RSS app authors recreate in the form of pre-populated feed lists, meant to “help get you up and running with RSS.”

Like blogrolls, they’re a proponent of homogeneity and an aggregation of the safest opinions to have within a certain niche among tech obsessives. If once upon a time nobody ever got fired for buying IBM, nobody will ever go wrong quoting something they read from the pre-populated list of feeds in their RSS reader.

This is stultifying, and it steers me back to what I like about Feedly:

If the workflow of the modern, individualist RSS consumer is a sort of hunt-and-gather trudge across the plain that is the Internet, Feedly is a reasonable run at an industrial information age.

Yes, it has a very safe, very non-controversial list of initial feeds you can use or browse through and pick from. But it also has a pretty big store of feeds you never see, and it is continually operating on them: The articles they contain are analyzed and categorized, providing a secondary stream of content you can dip into that probably transcends your own list, and that curates at the article level, not the feed level. You can follow topics, not individuals or individual entities.

Is there a risk of some kind of monoculture or slant finding its way into Feedly’s approach? Absolutely. To be truly engaged in a topic is to ultimately really only trust yourself when it comes to assessing and vetting information sources.

That’s where the whole “Iron Man vs. robots” thing comes in (thank you Luke Kanies for the metaphor.)

Feedly offers you a way to curate what it brings you:

  • Was the article properly categorized?
  • Do you want to see this source again?
  • Within this broad category, do you want to see this topic again?

You’re always free to bring in your own sources, you’re always free to recategorize, but you have a system that augments and supports. You still have to operate it and guide it back.

There are more prosaic benefits to Feedly’s approach, as well. Because it is constantly taxonomizing the content that passes through it, you can filter at a topical as opposed to keyword level, and that has some nice advantages for getting rid of annoyances. For instance:

If you follow many mainstream sites with paid staff you can’t unsee the number of sponsored and affiliate link posts they put up. They try to frame it like they’re doing some sort of journalism (“the lowest price we’ve seen”). If you follow the product segment found in a given deal post, though, you also can’t unsee how much of the stuff they’re pushing is stock the manufacturer is probably trying to clear to make way for the next thing, and you definitely can’t unsee the way these deals are only coming from sites with affiliate programs. No, this is not me discovering a conspiracy. This is me pointing out that deal posts are self-serving, presented as “research,” and inherently limited. They’re noise.

Feedly has a conception of “Deal Posts” and categorizes them as such. You can eliminate most of them with a single generic filter instead of painstakingly gathering the textual characteristics of each kind of deal post on each site you follow. Feedly’s not perfect, but I’ll take an 80 or 90 percent success rate and dial in a few outliers over trying to build a bullet-proof keyword list. That’s very useful automation.

The Client Problem

So, earlier I threw a little shade at RSS reader developers. It’s true. Something like Feedly needs back-end infra and people working on the problem of automated taxonomizing. The consumer RSS reader market doesn’t support that on $5 an app store purchase, so there’s no realistic way to move past the sole proprietor model of RSS curation/consumption.

As pure reading tools, though, the clients are pretty good! Plenty of ways to save and share content, flexibility on how you read the full article, simple ways to quickly import a feed into the reader while you’re out browsing, and (some, limited) filtering, at least on keywords.

Feedly, on the other hand, does not have a good client. The iOS client is buggy and the web client doesn’t feel very clean. There are some weird language things going on because Feedly is trying to turn streams of information composed of RSS feeds and other sources into a uniform, consistent river.

At the same time, you can get Feedly’s output mediated through a good reader. Personally, I like Reeder. It’s clean, pleasant, (Apple) cross-platform, has its own built-in read-it-later service (a good use of iCloud back-end syncing), and generally stays out of your way. Like other Apple readers it syncs feeds on iCloud if you wish, but it can also talk to many other services, including Feedly. Feedly and Reeder may represent the harmonic convergence of front-end and back-end.