In the ever-evolving world of technology, few voices carry as much weight as that of John Carmack, the legendary programmer behind id Software’s groundbreaking games like Doom and Quake. Recently, Carmack took to X (formerly Twitter) to share his candid thoughts on the futility of building entirely new operating systems in today’s tech environment. His post, accessible at https://twitter.com/ID_AA_Carmack/status/1961172409920491849, responds to a reflection by game developer Jonathan Blow on why modern engineers shy away from conceiving fresh OSes. Carmack’s commentary draws from his own experiences, including a pitch he received two years ago for LIBBA, a custom OS designed for smart glasses.

What makes Carmack’s perspective particularly resonant is his deep admiration for minimalist, efficient systems like Oberon, Plan 9, and even the unconventional TempleOS. He confesses a love for “clear, efficient programs that do their job without baggage.” Yet, he firmly argues that pursuing a new OS today lacks practical product sense, citing high costs, limited shelf life, and the burden on developers as outweighing any potential benefits.

The Meta Debacle and Custom OS Pitfalls

This stance isn’t theoretical for Carmack; it’s rooted in real-world battles. During his tenure at Meta, he “strenuously” objected to the company’s heavy investment in XROS, a fully custom operating system for extended reality (XR) devices. As detailed in discussions on Hacker News, Carmack’s warnings highlighted the inefficiencies of such endeavors in large corporations, where internal politics and resource drains can sap innovation. Meta poured significant resources into XROS, only to face the harsh realities Carmack predicted.

Echoing this, the email Carmack referenced in his post admitted he was right about LIBBA—the custom OS pitch failed to justify itself. This anecdote underscores a broader industry trend: while bespoke systems promise optimization, they often falter against established platforms like Android or Linux, which offer vast ecosystems and developer familiarity.

AI’s Role in Rethinking Software Foundations

Looking ahead, Carmack’s views intersect with emerging challenges in artificial intelligence development, as seen in various posts found on X where he discusses AI’s computational demands. For instance, he has noted how AI models, despite their power, struggle with real-time processing akin to self-driving systems, which are far from achieving artificial general intelligence (AGI). This ties into why reinventing OSes might be misguided—modern AI thrives on scalable, existing infrastructures rather than isolated custom builds.

Furthermore, Carmack has pondered AI’s potential to enhance codebases, suggesting tools could “make codebases more beautiful” by cleaning up inefficiencies, as mentioned in X posts from 2025. This implies that instead of building new OSes, the focus should shift to leveraging AI to refine and optimize current systems, reducing the “hairball” of legacy code without starting from scratch.

Lessons from Historical Ideals and Future Realities

Carmack’s reverence for past OS experiments like Plan 9, developed at Bell Labs, serves as a cautionary tale. As explored in a Quora thread explaining his Twitter handle’s nod to id Software and Armadillo Aerospace, his career has always balanced idealism with pragmatism. He argues that custom OSes rarely deliver on their promises in consumer products, where compatibility and rapid iteration trump purity.

In the context of XR and AI, this means companies like Meta might better invest in adapting proven OSes rather than custom ones. Carmack’s post also hints at deeper challenges for 2025, such as AI’s need for textual interfaces over GUIs, as he has discussed on X, potentially forcing a reevaluation of how software architectures support intelligent systems.

Balancing Innovation with Practicality

Ultimately, Carmack’s message is a call for grounded innovation. While the allure of a “dedicated OS for smart glasses” like LIBBA persists, his experiences at Meta reveal the opportunity costs. As Hacker News users have debated in relation to his other posts, the complexity of modern hardware— from chip fabrication limits to AI-driven compression—demands ecosystems that evolve incrementally.

For industry insiders, this serves as a reminder: true progress often lies in enhancing what’s already there, not in quixotic quests for reinvention. Carmack’s insights, drawn from decades of pushing technological boundaries, suggest that in 2025, the real breakthroughs will come from AI-augmented refinements, not entirely new foundations.