Software is breaking down at a systemic level
Points and comments are a snapshot, not live.
Software reliability is declining due to systemic complexity, not isolated incidents.
The author argues that 2026's major outages (Microsoft Azure's two-day global failure, an FAA ground-stop, the PHP2Shell vulnerability affecting 70% of websites, and the eMudhra SSL key incident) reveal a systemic breakdown, not coincidental failures. Key causes include excessive reliance on open-source dependencies (e.g., NPM's vulnerability to supply-chain attacks), cloud computing's complexity, and AI adoption that accelerates "decoherence"-accumulation of locally valid but incompatible code. The essay warns tech companies are losing "organizational cognition" as de-skilled engineers use AI to generate code they cannot fully understand, while cost-cutting and corner-cutting erode load-bearing infrastructure.
What commenters are saying
Several commenters push back against the article's novelty, arguing software has always been unreliable and citing historical parallels like Windows crashes or the Log4j vulnerability. A split emerges: some claim the current scale and interconnectedness create a "Kessler syndrome" effect where cascading failures are more likely, while others counter that cloud computing and open source have improved uptime compared to earlier eras. Embedded engineers note that their traditionally stable domain is now being affected by the same trends. One commenter criticizes the article's claim that NPM was revolutionary in 2014, pointing to much older package managers like Maven and apt.