Why are tech platforms obsessed with audio all of a sudden? Radio’s been around forever.

Here’s some context: In the last few years, improvements in machine learning have powered automatic transcription that’s cheap, quick, and accurate. That makes it possible, with enough computing power, to treat MP3 files like text. Before machine learning, a podcast’s MP3 files were like black boxes in an RSS feed. Now it’s possible for computers to know what an MP3 file is talking about without the use of metatags. That’s huge.

If we can treat MP3 files like text, we can link between them. What might it be like to fall into an audio rabbit hole? What would it look like to have programming that expands and contracts depending on how much time a listener has or where they are in their day? With better personalization, what’s news to me might be slightly different than what’s news to you. You used to have one giant antenna for your entire audience. Now, as a broadcaster, you can have many antennas for many audiences.

In a future where you don’t just have one big audience, but many, having a diverse newsroom is more important than ever. Look around your newsrooms right now: What talent are you overlooking? Who could you start developing right now? I think you could futureproof your newsroom by hiring more people that look like your audience and training them today.

In the radio of the future, the role of editors might change. They’ll still be really important; editors are the ones who decide what’s news. But it used to be that one of their main functions was to choose which stories went into the circular file — the trash can. You only had so much air time. But with dynamic on-demand radio, they won’t have that problem. In this version of the future, I think the best editors won’t be the gatekeepers. They’ll be the champions. The advocates. And they’ll have an incredible tool in their tool belt: data.

With data, we can overcome our personal biases to make better informed editorial decisions. It’s no longer “I like this, I don’t like that.” A savvy editor could learn about their audience and what works for them. We can use data to tell better stories. With second-by-second analytics, we’ll make better choices about our craft. We could A/B test a lede, for example. We can question everything!