Video autoplay on the web has been around in different forms for years, but its embrace by Facebook, Twitter, Snapchat, Instagram, and other platforms in the past year is changing its economics and user’s expectations. Many people dislike autoplay — quite a few hate it — but it’s a powerful tool to capture and hold viewers’ attention, it’s a perfect medium for advertising, and it generates huge impression numbers for platforms and publishers. It’s not going away. For video, it’s the vector to the future.

That said, there are huge problems, and plenty of work to be done to try to solve them. How do you handle sensitive content — shootings, beatings, beheadings? Platforms, publishers, and users uploading videos need fine-grained tools to set these preferences video by video, and users likewise should to be able to set their viewing preferences. This isn’t just about protecting your data plan — it’s also about protecting your psyche.

How do you come up with metrics that properly track meaningful attention, not just playcounts? (Probably no impression is completely meaningless, but some are definitely more meaningful than others.) What do you do about audio? The best implementations I’ve seen start autoplay in silent mode with captions. Unexpected autoplay audio is a deep annoyance — motion and text, not so much. And there’s a lot of experimentation to be done with text-heavy video, particularly on the news side. I believe silent with captions will become an emerging standard, both on big platforms and on publishers’ sites.

One weird thing is that right now, video autoplay is actually more common in web browsers and in mobile apps than it is on digital TV platforms. Part of the power of traditional television — part of the reason so much on the web is converging towards TV as an ideal — is that as soon as you turn it on, the programming just starts. Until and unless you turn it off, it keeps going forever. You can just sit back and let it happen — or, change the channel and get a completely new autoplay stream. Most of us, conceptually, don’t like this passive quality of television, because of politics and probably some psychosexual hangups we haven’t fully explored, but it is a super-powerful quality of media. To give you what you want without you having to ask for it — before you even were certain that you wanted it — is the condition to which all media aspires. It’s why social media has been so successful — it’s a subset of the web, lightly tailored to our preferences, that never stops.

Weirdly, digital TV platforms — Apple TV, Roku, Chromecast, etc. — don’t really work this way. They’re built more around the model of DVDs and CDs, with a smidgen of smartphone apps, than they are television and radio. If you turn on your Apple TV, video doesn’t start. You get a big menu. Open most of the apps, you get another menu. Click “Live TV” in these apps, you’re equally likely to get either another menu or instructions to enter a code into your computer to verify your right to watch a cable channel. And if anything, news publishers — The Wall Street Journal, Mashable, the just-launched Washington Post — have made their video apps even more boring than the cable channels’. There’s no good way to do discovery, to drill into and out of or between applications in the way you can follow a trail of links in a web browser or flip channels on a television set. It’s all completely backwards.

And — this is the point — it means all the incentives are pushing publishers to develop video content for social and video platforms on people’s phones and laptops rather than their television sets. Even though the crummiest cable television channel can get way more revenue than the vast majority of websites and newspapers. Video is difficult and expensive to produce, but the potential rewards are huge — if you have a good way to get to the audience.

There are some exceptions (not for wasting opportunity — everybody wastes opportunities) when it comes to good implementations of autoplay. Let’s take the case of Apple TV, where the most potential is going unrealized right now. ESPN is one of the few apps that as soon as you open it, starts a string of video clips — news and highlights of the day — in a box at the top of the screen with all of the menu options. Newsy, the Scripps’ aggregator that’s one of the early non-TV entrants to the Apple TV platform, begins with a presenter summary, and then a series of clips with top stories.

Reuters TV is maybe the most innovative. Instead of a choice of clips or topics (politics, sports, entertainment, etc.), your first menu choice is whether you have 10, 15, or 30 minutes to spend watching news. It then prepares a series of news clips accordingly, which will either play in sequence or the user can jump from one to the next. Reuters TV started as a way to deliver news on mobile phones, but it works beautifully on the television as well.

That sort of cross-platform cross-pollination is where the power really lies. All of these experiments are going to tell us what we can do on television sets, on phones and tablets, on the home page. They’re going to change publishers’ relationships with new and ongoing users. Autoplay is currently the most powerful form of media distribution. In 2016, we’re either going to figure it out, or it’s going to run right over us.