We are still in the cathedral model era of Open Source AI. The Bazaar begins only when the process itself opens up, when outsiders can help build the model, not just use what's handed to them. 5/5
The Cathedral and the Bazaar
There's been a lot of discussions going on lately about opensource AI, as tech giants like Nvidia, Google, and Microsoft have signed an open letter in favor of open-weight AI models. It's important that we understand what "open source" means. 1/5
Here’s the catch, most of what's being called "open" AI today is just the finished product handed out for free. The weights are public, but the process that produced them isn't. 4/5
Moral of the story: don't become a feature factory. Don't ship something just because you can build it (or your coding agent can generate it). Every feature should make your product more useful to your users, not just add to the bloat. (4/4)
Your coding agent can't empathize with your customers. You can.
Building new features used to be expensive. That expense was quietly filtering out features that weren't worth building.
(1/4)
AI-generated text can be more polished and structured, yet people still prefer writing that feels human. The same is true for features. AI-built ones can look more robust and complete on the surface, yet users still struggle to actually use them. (3/4)
The noise is finally being separated from the signal.
We are witnessing a software engineering renaissance. Instead of destroying the field, every attempt to automate the profession out of existence is only making it stronger, sharper, and more resilient.
3/3
What doesn't kill you makes you stronger
AI is being positioned as a threat to software engineering, a force that would make developers obsolete and commoditize the software engineering entirely.
Yet, instead of killing the software engineering, AI has started to refine it
1/3
The craft is being stripped of its busywork like memorizing syntax nuances, boilerplate code, and drafting mechanical documentation. Now engineers can focus on what machines still can't replicate: systems thinking, creative architecture, and deep problem-solving.
2/3
So what’s the solution? Build bridges.
→ Allow people with knowledge to borrow authority.
→ Help people with formal authority earn trust of the team.
Alignment between knowledge and power isn’t automatic, It’s something leaders must continuously reinforce.
4/4
Don’t let the promotion cycle turn into a cold war inside your team
You think you’ve done the right thing. You promoted fairly. You rewarded performance.
And then you hear it: “We now have two power centers in the team.”
1/4
but it lived under permanent tension and suspicion. We called it the Cold War.
Organizations behave the same way. Two power centers, two narratives, two sources of truth. You end up with a confused team, delayed decisions and quiet frustration.
3/4
Not recognizing luck brings you success. Recognizing it makes you great.
Recently, I watched a Veritasium video, “Why Being Delusional Is a Superpower.” In it he shows that luck plays a far greater role in success than most people are willing to admit.
(1/5)
Great leaders don’t inflate themselves, they look beyond themselves. They recognize luck, timing, and circumstance, and that humility is what allows them to keep growing.
(5/5)
#goodtogreat#leadership#success