I know why Mark Zuckerberg risked live demo failure
On Wednesday evening, I had a profound sense of déjà vu. When I watched Mark Zuckerberg open his Meta Connect keynote by giving the world a live backstage tour from his new glasses, I was transported back to 2012.
I was in the live audience at Google I/O when Sergey Brin introduced the world to some friends about to jump out of an airship, high above the event. I watched, agape, as athletes dove through the sky, launched bikes across the roof, rappelled down the side of the Moscone Center in San Francisco, then strode onto the stage right in front of me — all while streaming the whole thing live from their Google Glass headsets.
We’d never seen anything like it before — or since. “Google Wins The Internet With A Live Skydiving Demo Of Google Glass,” declared TechCrunch. “Google One-Ups Apple’s Famed Keynotes,” wrote The Atlantic. (Our headline was a bit more reserved.)
It’s an energy that’s been conspicuously missing from product launches ever since the covid-19 pandemic, when Apple began turning its once-exciting keynotes into highly polished 60- or 90-minute prerecorded videos. There’s excitement in knowing that something is real, that a demo can fail. Even if a demo does fail, it isn’t always bad! It reinforces the idea that everything else you’re seeing is real.
But if there’s a chance you’re going to fail, you need to be ready to fail. The demos need to be cool enough to genuinely wow if you actually pull them off. Or at the very least, the reward and risk should be aligned.
At Meta Connect, it was immediately clear Zuck was trying to bring at least a smidge of the same anything-can-happen energy that Google offered in 2012, now that he was introducing the first high-profile set of glasses with a screen since Google itself.
With Apple having largely whiffed on AI for the past few years, and Google having somewhat retreated from live risk after its AI made factual errors in demos, wouldn’t it be wonderful if Meta could look braver and smarter than its competitors? I get the logic.
But Zuck didn’t come ready to fail.
Zuck tried again and again and again and again to answer a glasses-to-glasses video call from Meta CTO Andrew Bosworth — something that can be notoriously unreliable even with fully released video calling solutions, much less a demo on stage. That’s even though the reward, if successful, would only be a simple demo of wireless video calling tech we’ve had for ages.
Earlier, he brought on chef Jack Mancuso to demonstrate how Meta’s glasses could help him whip up a “Korean-inspired steak sauce” by analyzing the ingredients in front of him and walking him through a recipe — only for the disembodied AI voice to inform him he was already partway through the recipe and leave him dumbly repeating the line “What do I do first?” again and again.
Then, he and Zuck had the gall to blame the fail on the Wi-Fi, as if an utter lack of connectivity could have tricked the glasses into thinking he’d already chopped up green onions and put them in a bowl. (It was not the Wi-Fi, and every techie watching the keynote knew it; usually, voice assistants stop talking at all when they lose connection.)
It later became clear that Zuck had been slinging some additional bullshit by pretending to think up a “Korean-inspired steak sauce” for Chef Cuso to attempt. Kotaku’s John Walker:
The hilariously clearly-labeled bottles of “Sesame Oil” and “Soy Sauce,” with the words squarely facing the camera, sit next to—oh my goodness, would you look at that—a jar of Cuso-branded seasoning! There’s also some spring onions, a couple of lemons, two garlic cloves, salt and pepper and maybe a potato and a bottle of honey? It’s all next to a very sad-looking steak sandwich. So, Meta AI, what can we possibly do?
Even so, the demo fails weren’t necessarily product failures for Meta, at least if you were watching the whole keynote. To me, the hiccups made it more impressive when the demos actually worked. I spent less time thinking about how Meta could have totally faked that bit where Zuck scribbles out an invisible message with his new neural wristband than I would have otherwise, and more time marveling at a demo where the glasses caption what people are saying in real time, right in front of your eyes.
When Boz apologized for fumbling the controls during that demo, it made me think: Oh, this one is real! And there, the result was cool enough that risk and reward were aligned.
Zuck blaming the Wi-Fi, and trying again and again to get a demo working, reminded me of that time Steve Jobs blamed the Wi-Fi for his own Apple demo fail in 2010. He was trying to browse The New York Times on the new iPhone 4, but it just wouldn’t load.
Like Zuck, Jobs also didn’t plan for failure that time: instead, he soon paused the entire keynote and told a room full of journalists they had to turn off the 570 Wi-Fi hotspots they’d apparently brought into the room.
“All you bloggers need to turn off your base stations, turn off your Wi-Fi, every notebook, put ’em down on the floor,” he decreed (and then asked a bit more politely).
But there was a method to the madness. Jobs really did have a good reason to disrupt his entire keynote.
Twenty minutes later, after additional wireless demos, Steve Jobs said he had one more thing to share — and made the first video call between two iPhones, the first public demonstration of FaceTime. That’s when video calling technology was still novel (though it did exist). The reward was worth the risk.
Skydive if you’re going to get the world to say you’ve got the best tech demo ever; tell a room full of journalists to turn off their hotspots if you’re going to make the first iPhone video call. Don’t worry about little hiccups: because of Jobs’ earlier fail, there was no question in 2010 that FaceTime was real. But if the best you’ve got is a half-canned cooking demo and yet another video call, be ready to move on.
I do hope we can return to live demos again. Most people who are willing to sit through tech keynotes understand that software is hard, hardware is harder, and fails happen! And for many of us, the failures show your product is real. In a time when generative AI and smoke-and-mirrors have become the norm, knowing something is real is more valuable than ever before.
Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.
