Mark Zuckerberg is weirdly talented at launching clunkers. I say that with love. And also, with the haunted look of someone who lived through Facebook Home.
But here’s the twist: those “terrible” products are kind of the point. They’re how Meta learns in public, at industrial scale, until one day the awkward V1 grows up, gets contacts, and eats a market.
Let’s unpack how that works plain English, short sentences, no condescension, a little self-roast where appropriate.
The pattern: ship, trip, iterate, dominate
Zuck’s track record is a highlight reel of misses that taught Meta exactly what to build next.
- The Facebook Phone era. In 2013, the HTC First (the “Facebook phone”) face-planted so hard AT&T slashed the price to 99 cents within a month. Facebook Home li…
Mark Zuckerberg is weirdly talented at launching clunkers. I say that with love. And also, with the haunted look of someone who lived through Facebook Home.
But here’s the twist: those “terrible” products are kind of the point. They’re how Meta learns in public, at industrial scale, until one day the awkward V1 grows up, gets contacts, and eats a market.
Let’s unpack how that works plain English, short sentences, no condescension, a little self-roast where appropriate.
The pattern: ship, trip, iterate, dominate
Zuck’s track record is a highlight reel of misses that taught Meta exactly what to build next.
- The Facebook Phone era. In 2013, the HTC First (the “Facebook phone”) face-planted so hard AT&T slashed the price to 99 cents within a month. Facebook Home limped along, its team disbanded, and the UK launch was scrapped. Ouch.
- Portal. A surprisingly nifty video-calling appliance with exactly the wrong brand name on the bezel. Meta killed consumer Portals in 2022 and quietly sunset the line. Great product, wrong company.
- Libra/Diem. Remember when Facebook tried to reinvent money? Regulators did too. The project sold its assets to Silvergate in 2022 and rode off into the crypto sunset.
- Horizon Worlds. The metaverse party where most people didn’t show. Internal docs leaked to the WSJ showed monthly users under 200k by late 2022, with poor retention. (VR diehards remain; the rest… wandered off.)
- Poke/Slingshot and friends. A whole zoo of Snapchat clones launched, flopped, and quietly vanished. Then… Stories arrived, and Instagram’s copy hit 400M daily users double Snapchat’s total user base at the time. Reels looked wobbly at first too, but it now drives a huge chunk of Instagram time. Iteration is a dish best served… shameless.
This is the Zuck cycle: copy, compress, carpet-bomb distribution, learn, then keep going until the metrics say, “you did it.”
Why Zuck can keep doing this (and you can’t fire him)
1) Founder control means long runway. Meta has dual-class shares. Zuck owns a minority of the stock but controls a majority of the voting power, which officially makes Meta a “controlled company.” Translation: he can green-light giant, weird bets without worrying about a shareholder coup. (You may exhale, activist investors.)
2) The ad money machine funds the experiments. When your core business throws off rivers of cash, you can afford to be wrong loudly. That’s why Meta can lift capex from ~$40B in 2024 to $60–65B in 2025… and then raise guidance again to $70–72B as the AI buildout accelerates. Most companies get one moonshot. Meta orders a tasting menu.
3) Distribution is the cheat code. Even a “meh” feature gets a billion-user audition when it shows up in Facebook, Instagram, and WhatsApp. That’s how Reels went from copycat to habit; Meta says AI-powered recommendations boosted time spent on Instagram by 24% once Reels took hold. Stories did the same trick years earlier.
“Terrible” as a strategy
Zuck’s superpower isn’t perfect taste. It’s industrial-grade tolerance for being wrong:
- Public pivots at CEO speed. From “all-in on the metaverse” to “personal superintelligence” in just a few news cycles, with team structures and budgets moving accordingly. That July 2025 memo lays it out: an AI that knows you your context, your stuff living in glasses and apps. Whether you love or fear that vision, it’s a clear north star.
- Metrics over mythology. Meta will A/B test everything that isn’t welded down. Sometimes that veers into “did they really do that?” territory (the company has been accused of stress-testing user tolerance on Android), but the culture is unapologetically experimental. Results beat vibes.
- Ruthless reuse. The best parts of failed projects survive inside winners. Facebook Home died; its ideas lived in Messenger and the main app. Poke vanished; Stories thrived. Horizon sputtered; the wearable work kept going.
When it finally works
Here’s the fun part: after enough misses, the puzzle pieces click.
Ray-Ban Meta smart glasses are the first genuinely sticky Meta gadget. They started simple camera, audio, voice but momentum is real: over 2 million pairs sold by early 2025, new models with tiny displays shipping, and a whole category bending toward “AI in your face but make it stylish.” Analysts peg Meta’s share of smart-glasses shipments at 70%+ in 2025. That’s not niche; that’s a beachhead.
Does this make the metaverse retroactively smart? Not exactly. But it proves the bet behind the bet: glasses might be how AI gets out of your pocket and into your life. And Meta can keep swinging until it finds the right version of that future. The bill is… substantial. Reality Labs is still losing billions. But the willingness to be “wrong” longer than competitors is starting to buy real advantage.
What builders can steal from Zuck (ethically, please)
- Ship the embarrassing V1. If your first version isn’t a little cringey, you waited too long. Just don’t staple your logo to a phone. We’ve tried that. It was 99 cents.
- Exploit your distribution. Don’t have a billion-user feed? Cool. Use the distribution you do have: your newsletter, your Discord, your five best customers.
- Measure what matters. Copycats are fine if you can out-learn the original. Stories and Reels weren’t new; they were instrumented better.
- Hold a strong opinion, loosely. Be metaverse for breakfast and personal AI for lunch if that’s where the evidence points. Just leave room for dessert.
I’ll end on this: it’s easy to dunk on the flops. It’s harder and way more useful to see the method in the mess. Zuck’s genius isn’t clairvoyance; it’s endurance. He’s built a machine that can be publicly wrong, repeatedly, and still get to right.
That’s oddly inspiring. Make something a little terrible today. Learn fast, laugh at your own V1, and keep going. The future belongs to the folks who iterate with a grin.