complexity

一篇提出「去相干（decoherence）」概念的评论文章，论述 AI 大量生成的冗余内容如何让系统的技术债指数级膨胀，并认为这既排除了智能爆炸的可能，也带来了「去相干崩溃」的新风险。

A reflective post re-examining Brooks's classic 'The Mythical Man-Month' in the AI era, arguing that while AI can generate runnable code quickly, the real bottlenecks shift from human coordination to specification and verification of large-scale software systems.

The article proposes that consciousness is a spectrum based on active interconnected complexity and speculates on implications for AI systems like chatgpt.

The article explores how the unique characteristics of software development, such as its speed, flexibility, and hidden costs, can lead to mental strain and irrational behavior among developers and stakeholders.

Engineers are losing grip on their systems due to rapid delivery and higher-order abstractions, which speed up happy paths but prolong failure paths without enough discussion.

Zach Kehs observes that unlike physical buildings, software faces no structural ceiling and can always degrade further through additional layers of indirection and reduced performance.

The article explores how organizations face coordination challenges similar to slime molds, drawing metaphors from biological systems to understand adaptive behaviors in complex environments.

This article announces the release of Sloc Cloc and Code (scc) version 4.0.0, featuring new hotspot analysis to identify code files requiring attention based on complexity and commit history.

The article explores the tension between control and complexity in systems design, particularly in the context of LLM adoption in software development, contrasting analytical decomposition with complex systems approaches.

The article questions whether the proliferation of AI agents is leading to increased complexity in management and maintenance, potentially outweighing their benefits.

SSOG-Attention introduces a novel attention mechanism using Sum of Separable Gaussians to achieve sub-quadratic complexity, outperforming or matching SDPA with faster convergence and improved efficiency.

A quote from Florian Herrengt about how AI-generated code can lead to convoluted projects where no one understands the system, highlighting risks in software engineering.

A commentary questioning why simple automations still require technical expertise, highlighting usability gaps in automation tools.

This theoretical paper studies atomic concept learning through the geometry of hypercubes and hyperplanes, showing that complexity collapses uniformly on most hyperplanes except the full diagonal, where it grows without bound. It provides a taxonomy of hyperplane behavior and a worked binary and ternary case analysis.

Carl Kolon, an engineering leader, shares his curated 'required reading' list of software engineering articles, covering coding practices, platform design, and frontend development, with personal notes on why each matters.

This paper studies the complexity of answering two-way (conjunctive) regular path queries over knowledge bases with linear and guarded existential rules, establishing completeness results for data and combined complexity.

The paper argues that general intelligence necessitates non-reducible constraints across multiple levels of description, presenting theoretical implications for AI and cognitive science.

An opinion piece arguing that superintelligence may be incremental rather than explosive, as the limits of AI lie not in reasoning but in the slow pace of real-world verification and complex emergent systems.

IBM Research explains why model routing in agentic systems is more complex than a simple classification problem, highlighting how caching and hidden factors like actual workload cost and task difficulty estimation make routing a systems optimization challenge.

This paper introduces ExplAIner, a declarative query language for explaining classification models, addressing limitations of the FOIL language by supporting abductive, contrastive, and optimality-based explanations with tractable evaluation over Boolean circuits.