A new rss strategy

For the last several years, i have been using feedmail to view posts from sites i care about. I consider feedmail to be almost perfect software. It does what it claims to do, has a fast and simple (although not beautiful) interface, a reasonable pricing strategy, and a friendly and helpful developer. Receiving posts in my mail client has lots of advantages, because in many ways posts are like emails, my email client is comfortable, and email has lots of built in systems for filtering things. Furthermore, having posts loaded and rendered on a server means i don’t need to worry about enabling a vpn in order to get updates from sites that don’t make it over the great firewall.

But there are two things about this setup which make me want to change it. The first is that i’m not hugely enamoured with the long list of posts that scream to be read. This kind of timeline overwhelms me, and i would rather enjoy the time i spend poking around on people’s person websites. The second is that it requires an external account, and i try to make a point of having as few accounts as possible, for various irrational reasons.

So there are two things that i can do to solve these two issues. The first is to use a local rss reader. The second is to run something on a server that i already have access to. This would probably be some kind of script that generates a listing of recent posts.

I won’t beat around the bush: the path i chose is to run a local rss reader, with the thought that at some point i will again replace this setup with a static page generated on a server. Running something locally is simpler, and lets me first get a feel for what i want my rss experience to be like. Once i’ve got a more solid idea, i might make a move towards something hosted. Or maybe not—i might end up liking this system so much that i never want it to change.

Briefly, it seems that there are approximately one and a half styles of rss reader. The classic style, as exemplified by netnewswire, has three columns: one for all channels, one for posts in a channel, and one for post content. The channel list may or may not allow categorisation using subfolders or tags. It also may or may not allow clicking further up the folder hierarchy in order to populate the posts list with posts from multiple channels. Which leads to the first variation, in which all channels are present in a single list, a design exemplified by elfeed. Elfeed’s use of search and tags enables it to approximate the behaviour of individual categories. For a low volume of channels, or a casual reader, or a very diligent reader, this approach works quite well, but i personally find it quite overwhelming. Another variation is the river of news, which is usually a web page with all posts from all streams listed chronologically, with the full content available. This works quite well as a hub for blogs on a related topic, such as planet emacs, but not so well for personal channels, as it makes it easy for one channel to crowd others out, and encourages skimming here and now rather than reading in depth. Finally, there is another crowd containing feedly, inoreader, and others, which takes this fundamental formula and makes the interface look a bit more like a digital newspaper or social media website.

There is a lot of talk around rss about it not having an algorithm. This is not quite true, but the algorithm is usually very simple: take all posts and sort them in reverse chronological order. The nice thing about this is that it doesn’t hide posts you might find interesting just because you may not get angry and engage aggressively with them. But as mentioned, this comes with the downside that more frequent posters drown at the less frequent posters, even if the value of each post is equal.

Thinking about it, i first wondered if an alternative algorithm that could be implemented could be similar to a music player’s shuffle algorithm. See, it turns out that shuffling music is hard, because true randomness often doesn’t feel random. A truly random shuffle could end up with a whole album of songs in order and in a row in the middle of a queue, but to the listener, that is clearly not random, whatever the algorithm says. So now the shuffle button implements a different algorithm, perhaps prioritising songs that have not been recently played, and spacing out songs from the same artist or album so that they never play consecutively. Applied to rss, while this would introduce some variety into the post listing, a high output channel would still make up the majority of posts.

Perhaps more advanced algorithms could further improve the situation, but this feels a little suspicious, because of the perceived simplicity and honesty of the rss system. So we need to find another way to avoid this issue, and it seems that the interface could be the way to do that.

On his website, Ton Zijlstra describes his ideal rss reading strategy. This involves having folders for channels based on how well he knows the person the channel belongs to, as well as renaming every channel to be the name of the person it corresponds to. In this way, he takes steps to make his reading rss less of another endless feed, and more an intentional activity through which he can learn about what people he is familiar with have been doing recently. This is an interesting thought, and i would like to try it for myself, and see if it has any effect on the way that i think about rss.

The other thing that i want to try is to move from a long list of posts to a system that puts the spotlight on each channel in turn. One of the only pieces of software that seems to be experimenting in that direction is fraidycat, designed by its developer to comfortable accommodate very large numbers of rss channels. The way it attempts that is to make every channel take up exactly the same amount of space on the screen: each channel gets a title line, and a single line horizontally listing the most recent posts. A channel can be expanded to show more posts temporarily. Channels can be given tags, and also sorted within tags based on the desired check frequency, from all the time to very rarely. While my volume is much lower, i value this alternative interface as a viable replacement for the status quo.

While i haven’t felt out exactly what i want to change here, i know i want to explore further. Rss is a valuable system, but like any system it has better and worse ways to use it. I stopped using social media for a reason, and i don’t want rss to become another black whole that i can fall into to kill time and grumble about the state of the world; rather, i want to highlight people saying and making exciting and interesting things.