Netflix CEO Reed Hastings made a bold keynote speech at a recent press event in Paris, in which he predicted the death of linear television.
"Internet television is the new wave in entertainment," he said. "It's no longer 'What do they show [on that channel]?' but 'What do I want to watch?' - and some day the idea of a show 'being on at 8 o'clock' will seem very foreign."
But if scheduled television really is set to become extinct, how will we stumble across new programming and forms of entertainment?
According to Carlos Gomez Uribe, Netflix's vice president of product innovation, the algorithms employed by the streaming service will eventually reach such an advanced state that they'll do all of the hard work for us.
How does Netflix know what I want to watch?
Whenever a new film or TV show is added to Netflix, each piece of content is classified by human taggers: "But then as soon as people start to play it, the statistical algorithms based on play behaviour take over and that's the dominant way in which we produce recommendations."
What to Read Next
So how exactly do these algorithms - "about 15" working in unison, according to Gomez Uribe - work?
"The main strategy is to find other people that have a similar taste in movies and TV shows as you - then we group them together and analyse what those people have been watching.
"We then recommend to you the popular shows that this community's watching. Through A/B testing, we continue to refine the clusters and the global communities that we create."
Is Netflix the first true global TV network?
Since January 2016, recommendations have been global, with Netflix expanding to over 190 countries worldwide, so these algorithms are no longer dictated by your location.
"Many different types of algorithms come together - and before last year, all these algorithms were regional," says Gomez Uribe.
"We would run each algorithm on the data for a single country, meaning that if you were a member in the United States, your recommendations were the product of analysing play behaviour data from other people in the US.
"But when you do that, stories don't travel as easily - so now what we have are completely global algorithms that use the data from the entire world to create the recommendations."
In short, Netflix is now far more likely to recommend foreign language content to its users than it ever was before. Though since each Netflix territory has different content, it's not a seamless crossover - something Gomez Uribe says the company is working to counter.
"Our content is changing all the time - and the consequence of analysing the data is that we change our programming, so all the time we're adding new content and dropping some content."
Netflix could soon predict what you want to watch
Having opened up the world of recommendations - or, as Gomez Uribe puts it, "helping regional content travel to the rest of the world" - the next stage in Netflix recommendations could be even more dramatic.
"We know that it's still hard for people to find something great to watch in every session," he says. "One ambition that we have is: you log into Netflix and a video starts to play immediately - and 80% of the time, you're quite happy with that video and you don't have to make any decision."
Yes - in the future, Netflix could take its predictive algorithms to a whole new level and start auto-playing what it thinks you want to watch, based on your behaviour and play-data from other users.
Amazing? Scary? With Gomez Uribe and his 100-strong team working day-in, day-out, to develop their algorithms, one thing's for certain - the changes to Netflix are only just beginning.
