Sam: The board of Automattic just put Matt Mullenweg—the guy who started WordPress—on indefinite leave. This isn't a vacation.
Kai: And… wow. What a way to start the day. This is Kai.
Sam: And I'm Sam.
Kai: Welcome to Open Source, Trending. It's Thursday, September 10th, and we have a packed show for you today.
Kai: On the docket: we're digging into that bombshell—a massive shakeup at Automattic, the company behind WordPress.com and Tumblr.
Sam: Then, Alibaba's new open-source model is turning heads… because it looks like it copied someone's homework.
Kai: And we're looking at the growing mountain of proof that self-driving cars really are saving lives.
Kai: Okay, let's start with that Automattic news. TechCrunch is reporting the board forced CEO and founder Matt Mullenweg into a leave of absence.
Sam: 'Forced' is the operative word. This isn't a sabbatical. The comments on the story are going nuclear. This is a boardroom coup.
Kai: 'Coup' feels a little strong. I mean, the pressure of running a giant that powers, what, 40% of the internet? Maybe the guy just needs a break!
Sam: Kai, founders don't get 'forced' into breaks when things are going well. A 'leave of absence' is corporate speak for 'we're firing you, but we're letting you save face.'
Kai: Okay, but why? What could be so bad? Is it Tumblr bleeding money again? Competition from Ghost and the whole headless CMS movement?
Sam: The speculation is all over the map. Stagnant growth, a failed AI strategy, maybe internal culture clashes. Remember, Automattic is fully distributed—that's hard to manage at scale.
Kai: So if you're a developer building on WordPress... this has to be unsettling.
Sam: It's pure uncertainty. A new CEO could change everything: double down on the block editor, spin off or kill Tumblr, or chase enterprise clients and alienate the open-source community.
Kai: So the future of a huge chunk of the internet is suddenly up in the air. Wow.
Sam: Suddenly very, very up in the air.
Sam: Alright, next up… let's talk about AI model 'originality'—or maybe the lack of it. A new Gist is making the rounds about Alibaba's latest model, Qwen 3.8.
Kai: Right, and it's supposedly incredible! It's using this new 'reasoning pre-fills' technique, which is what everyone's expecting from the unreleased GPT-5.5 Pro. Basically, the model starts with the chain-of-thought path already laid out.
Sam: Right. Except... it might not be 'using the technique' so much as 'copying the answers.' The Gist shows Qwen 3.8 outputting the exact same internal pre-fill tokens as OpenAI's closed-source model. Word for word.
Kai: Wait—what? How is that even possible if GPT-5.5 Pro isn't public?
Sam: That's the billion-dollar question. The most likely answer is that Qwen was trained, at least in part, on synthetic data from an early or leaked version of the OpenAI model. This isn't innovation, Kai. This is mimicry.
Kai: Hold on. Even if that's true... isn't that a good thing? Cutting-edge tech gets into open-source models faster. Who cares how it gets there?
Sam: I care! It's intellectual property laundering! And it's dangerous for the ecosystem. If every new model is just trained on the output of the last one, we get model collapse—a downward spiral of quality. It's an ouroboros of mediocrity.
Kai: 'Ouroboros of mediocrity' is a little dramatic, don't you think? For a developer, this just means you get to play with reasoning pre-fills now, in an open model, instead of waiting for OpenAI's blessing. It's a pragmatic win.
Sam: It's a short-term win that undermines real research. We're celebrating a photocopy while the Mona Lisa is locked in a vault.
Kai: Hey, sometimes the copy is good enough. We'll see what the community does with it.
Kai: Alright, let's switch from software to hardware. An article in IEEE Spectrum is making a bold claim: we have growing proof that autonomous cars are saving lives.
Sam: I've heard this one before. What's the 'proof' this time?
Kai: Actual hard data. The studies they cite show Waymo and Cruise vehicles have way fewer injury-causing crashes per million miles than humans. We're talking something like 76% fewer crashes with any injury.
Sam: Okay, but let's look at those miles. Are we comparing apples to apples? The article itself notes that most of those AV miles are on sun-drenched highways in Arizona. They aren't dealing with black ice in Boston at 2 AM.
Kai: Fair point, but that gap is still massive. And remember, the 'human driver' stats include everyone—the drunk, the texters, the tired. An AV is never any of those things.
Sam: True, but they have their own issues: edge cases, sensor failures… and the fact that one splashy AV crash gets more hate than ten thousand human fender-benders. The comments section on this story has over 500 posts. Public trust is everything, and it's tissue-paper thin.
Kai: So it feels like the tech is getting there, but public perception is just stuck in first gear.
Sam: It means the biggest challenge isn't engineering anymore. It's psychology and law. It's about who gets sued when things go wrong, and how you convince people that one weird robot mistake is better than the constant, predictable chaos of human error.
Kai: Which is a much harder problem to solve than just writing better code.
Sam: Infinitely harder.
Kai: So, to recap: a boardroom earthquake at Automattic puts the future of WordPress and Tumblr in doubt.
Sam: A new open-source AI model looks like it cheated off OpenAI's test paper, and self-driving cars are probably safer than you, even if you don't want to admit it.
Kai: And before we go, a fun one: today, September 10th, is apparently 'Swap Ideas Day'.
Sam: Let me guess. You've already swapped out your entire VS Code setup for a terminal-based editor you found on GitHub five minutes ago.
Kai: Not yet. But it is a good reminder to look outside your bubble. If you're a React dev, check out some Svelte code. If you're all Python, go look at some Rust. You never know.
Kai: That's our show for today. We'll be back tomorrow with the latest in open source and dev.
Sam: Until then, try not to get forced into a 'leave of absence.' See you tomorrow.
This show is made with AI: the hosts’ voices are synthetic and the scripts are AI-assisted. Every story links to its original source.