Chaos is not the same as randomness, for there is always order hidden in the former.
I want to talk about a tale of the digital age. Of now, as well as of then.
The early Internet was, in a certain sense, isolating. Finding interesting things to read or websites to visit was cumbersome.
There were different attempts to cut through the clutter: web directories, mailing lists, portals. But those who identified this crying need also saw in it an opportunity to exert control and power. Do not imagine these people as mustachioed villains in subvolcanic bunkers, because they weren't that. Still, they were lured by the idea of being the gatekeepers of the entry point to the Internet most people used every day.
Their mission was not to bring people search results, but to bring them recommendations. That's what Yahoo! was most successful at, in its heyday: putting news, articles, games, weather and everything else in one easily accessible place.
The winners in the long struggle to be the people's worldwide curator turned out to be the social networks: Facebook, Twitter, Instagram, Reddit. They won for two reasons, one to do with people, the other to do with content. The first is well understood: people created accounts with these services because all their friends were there too. The second reason, though, is less obvious: for all their privacy faults, systems that meticulously followed people's interests to recommend them stuff they'd be likely to enjoy were tremendously efficient. The personalized social media feed became the machete in the Internet rainforest.
The entrepreneurs got glory, money and power, and the public got shiny, convenient, pleasant services to find new things to read, watch and listen to. The situation has stabilized. Or, if you're of a nasty turn of mind as I am, you may say it has metastasized.
Indeed, I have come to realize that the dominion of the personalized feed must crumble and fuck off. For me, as for many others, it fails to fulfil its core purpose: bringing new, interesting, exciting recommendations that I would not hear of otherwise.
In fact, it appears that it has become increasingly useless under the pressure to turn a profit, interspersing semi-irrelevant sponsored distractions like paid posts and sponsored tweets among the increasingly sparse meat I'm actually after. Yet even unadulterated by money, the algorithmic social feed following one's every action to understand what new content to bring them would still fail, because it is predicated on a false assumption: that people want more stuff just like the one they've previously enjoyed. (That's the reason YouTube won't stop pestering you with a popular channel if you've liked one of their videos.) While this makes sense sometimes for some people, it becomes nauseating if this assumption underlies the main avenue for the discovery of the new, fun, important or interesting. Every day, for everyone.
That's where we are today. (How bad this is for being a well informed citizen in a democracy is a whole separate barrel of dung.)
I don't think people ever really want more of the same. Like that one dude at Apple put it, people won't know what they want until you show it to them. They want the novel, the exotic and the excellent. Not complacent sameyness, but exciting serendipity.
The sterile personalized algorithmic feeds of today aren't too great at this, particularly when it comes to interesting articles or news. I suspect that people don't necessarily like the feed, so much as they are fine with it.
I'm not fine with it anymore. I want being on the Web to feel like anything could be around the corner, and that it will be excellent. So in May I started an experiment called Headlines.
It's not technically impressive at all. Headlines is a hastily hacked together, simplistic, clean RSS aggregator website. Each page draws on a couple dozen high-quality sources from around the Web to create feeds of articles on different topics (it started with news, but I added three other categories).
Where it's different from an RSS reader is that articles aren't a chronological timeline. Instead, they're continually sourced to make sure they're recent, but they appear in a random order each time the feed is refreshed. That way, big publishers don't get to dominate smaller ones as much. In the future, I'd like it to limit the number of articles it draws from any given source.
It's a work in progress. It's sometimes sluggish, and there's the occasional bug. It's unintelligent, because it's limited and random.
But here's the thing: it's better than any recommendation website, social media feed or curation service I've ever used. We've been testing it out for the past three months, and the combination of a large number of trusted, high-quality sources along with the randomness factor has resulted in an extraordinarily fun, serendipitous and insightful recommendation engine. As far as the news section goes, it's better at offering an objective, diversely sourced view of current world events than anything I've found. It's simultaneously stupidly simple and freakishly useful.
It's curated chaos, and it has beaten the algorithm.
That it has been possible for such a simple project to be more useful than sophisticated feeds built by teams of engineers over the span of years may tell us two things. It may mean I (and others who've tested it) are a minority of weirdos. Or, perhaps, it may tell us that tech companies and big websites are chasing entirely the wrong idea, and they either haven't noticed, don't care or refuse it for financial reasons.
Because it's useful, Headlines will remain online. It's just an experiment, so I don't plan on making further major changes. If you like it, I invite you to use it too. If you have thoughts on the topic, I'd be happy to hear from you.
This project is not the final solution to good recommendations, or to making the Web an exciting open plain to explore. It's a small first step. Maybe just a tiptoe.
I think we should strive to build things that make our lives excellent, not mediocre. Sometimes, social media algorithms don't even manage to do the latter.