🤔The one where I ask questions and have no answers
One nice thing about writing a newsletter is that it’s a break from the day-to-day hustle to get stories out of the door. It’s also a chance to go back to some older stories and see how they played out – and sometimes it only takes a week for those stories to change entirely. Here is a good example:
What’s next for Mozilla?
When I first wrote about the leadership changes at Mozilla a few weeks ago, I admit that I was skeptical. Replacing Mitchell Baker with Laura Chambers, who despite Mozilla’s claims, never really held a product role and doesn’t have a background in open source, seemed at best like an odd move and, for quite a few pundits, it looked like another Mozilla blunder. Now, I think it may have been the right move. Sometimes you need a new leader to come in to make the hard decisions.
Today’s newsletter was written on the Keychron Q8.
About two weeks ago, a week after the leadership change, I got a tip that there would be layoffs at Mozilla, with about 60 people impacted. Mark Gurman over at Bloomberg beat me to that story, but luckily I also got the internal memo about what that meant for Mozilla’s product portfolio. Since Bloomberg didn’t have that context, I quickly retooled my story to focus on that instead.
As it turns out, Mozilla is shutting down projects like its 3D virtual world Hubs, scaling back its investments in its Mastodon instance and some of its security and privacy tools (which are often just repackaged and re-branded versions from partners anyway). Instead, Mozilla wants to focus on bringing trustworthy AI to Firefox and it’s combining the Pocket, Content and AI teams with the Firefox organization.
Mozilla is great at getting distracted. Refocusing on Firefox is a necessity before it loses what is left of its market- and mind-share.
Frederic Lardinois
And let’s be real. I have no idea if there’s a future for Firefox, but Mozilla’s future is tied to the success of its browser, despite its best efforts to change that. Sure, the Google deal will get renewed for the foreseeable future. Google needs it as much as Mozilla to keep it out of yet another anti-trust lawsuit (even as everyone knows that it’s just playing a game here by propping up a competitor to Chrome). Yet, once Firefox becomes completely irrelevant, there's no value in keeping up the charade for Google anymore either.
Mozilla is betting heavily on (small) local AI models and the natural place for the organization to bring that to market is within Firefox. It’s unclear to me what that’ll look like, but the organization regularly points to Fakespot as an example.
Either way, I do feel more optimistic about the future of Mozilla today than I did a month ago.
What's next for digital media news organization?
Talking about layoffs... The media bloodbath continues: Vice is essentially shutting down, BuzzFeed is cutting 16% of its employees, and our Yahoo cousins at Engadget laid off 10 people, including its top leadership team on the editorial side.
The Engadget layoffs hit close to home, of course, especially after we had our round of layoffs at TechCrunch a few weeks ago. We do both belong to the same private equity-owned parent organization, after all. Nobody has ever accused a private equity firm of doing what’s in the best interest of employees (of the future of journalism).
Going forward, Engadget is going to be split into a News and Features team, that will focus on news and “traffic growth” and a Reviews and Buying Advice team, that will focus on evergreen content and guides. The second part is code for commerce posts – the current bane of the Internet.
When advertising rates fall and subscription businesses don’t work out, what’s left? The current answer: commerce posts that generate affiliate fees from Amazon and the like. The logical result of that is what’s left of Cnet today or (because Apple News keeps recommending this to me), the Travel Products section of Travel+Leisure. Some of the smarter publishers hide this stuff on their homepages but know exactly how to perform the necessary SEO voodoo to have it rank high on Google.
For a while, Wirecutter got this model right and felt trustworthy – and maybe it still is in some areas – but the fact that every random site now features product recommendations that are clearly not based on actual reviews makes every other site that uses this same model feel less trustworthy. It also devalues the journalism on those sites.
One thing I do know: if you're a journalist and your leadership starts unironically referring to your work as 'content,' start thinking about a plan B.
What's next for open-source in the age of AI?
Let’s get back to tech, though. Google launched Gemma this week, a set of open models that perform quite well against Meta, Mistral and others competitors.
That is interesting by itself, but it got me thinking once again about what this means in the context of open source. Because these models are ‘open weight models’ that may have permissive licenses but are explicitly not open source. Google stressed that multiple times in its press conference before it launched the Gemma models and noted that it did so to ensure it could release these models in a responsible way (Google’s own Gemini model also showed what happens when it’s done in a way that’s maybe a bit too responsible). To me, that feels like a bit of a cop-out. It allows you to claim to be ‘open’ without having to go all the way and, among other things, disclose your training data and other details.
I know the folks over at the Open Source Initiative are deep in the weeds on what an open-source definition would look like. One of their core tenets, best I can tell, is that to be open source, a model’s creator would have to allow users to study how the system works and be able to inspect its components.
Right now, though, there’s still a discussion around what the word ‘system’ actually refers to in this context. I know. I know. Fascinating stuff, but also of vital importance to what AI development will look like in the future. More about that in a future newsletter.open-weight
Frederic Lardinois
One lesson to learn from this edition: I'm much better at asking questions than providing answers...
Note: This newsletter is a personal project that is not affiliated with Yahoo and TechCrunch.
