Am I Seeing the Silver Bullet for the Software Crisis?

Two classics: The software crisis and No Silver Bullet, LLMs may be changing the assumptions behind both.

Two Pressures, One Bottleneck For most of software engineering's history, two old ideas have pushed on us from opposite directions. The first is the software crisis. As computers become more capable, we try to build larger and more ambitious systems. Better hardware expands the frontier of what we attempt, so the demand for complex software keeps outrunnin…

Authoritarian Management in Software Organizations

EVLN, proxy agents, and why non-technical authority weakens corrective voice in knowledge-intensive teams.

Why do skilled engineers leave organizations that seem stable? In software, the reasons are usually structural, not personal. Disengagement is more than a morale issue. It often happens when organizations silence technical feedback but cannot judge it themselves. Two ideas help explain this. First, EVLN describes how people respond to decline: exit, voice,…

RFC: 기록 유산을 "연산 가능한 인프라"로

한반도 역사 인구를 대상으로 족보·유전체·문헌을 결합하는 확률적 계보 DAG 인프라 제안

본 RFC는 한반도 역사 인구의 계보 자료를 PII 없이 통합해, 족보·문헌·유전체를 교차검증 가능한 확률 그래프로 다룰 수 있는지 검토하는 인프라 설계 제안입니다. 요약 한국은 다음 조건을 거의 동시에 갖고 있습니다. - 왕조를 관통하는 기록 문화(실록·승정원일기·향안·호적·족보 등)와 현대 행정 DB의 연속성 - 성씨·본관·항렬 같은 자연스러운 계보 인덱스 키 - 대규모 디지털 전환(OCR/아카이브/전자족보) - 한국인 유전체 레퍼런스 및 샘플 증가 - 수백 GB수 TB RAM, GPU, 그래프 연산 스택의 성숙 이 RFC의 목표는 이름·주민번호·주소 등 PII(개인식별정보, Personally Identifiable Info…

The Ghost in the Mirror: Five Lessons from Building Software with ISO 42010 and AI Agents

Five operational lessons from constraining AI agents with architectural descriptions rather than ad-hoc prompts.

This note summarizes part of an ongoing research project on architectural description languages for software engineering and machine learning systems. For one project, I set a clear rule: I would use ISO 42010-style architectural descriptions to specify the system, but all implementation would be done by AI agents. My goal was not to test "no-code" marketin…

Creativity, Cost, and Context: Competing with AI as a Software Engineer

Why human engineers remain valuable when knowledge is cheap but judgment, economics, and context still matter.

From a talk to senior students at CAU, Seoul, 2024-11-19. Introduction AI has changed how software work is valued. Tasks like boilerplate code, API connections, simple unit tests, and basic debugging can now be done quickly with coding assistants. This does not mean engineering is less valuable. Instead, the value has shifted from routine tasks to good jud…

AI4SE and the V-Model: The Case for Shoot-and-Forget BDD

How LLM-assisted black-box testing changes the economics of verification and reduces knowledge asymmetry in software teams.

Introduction AI for software engineering does not make the V-model obsolete. It changes the cost structure around the artifacts that the V-model depends on. The core idea of the model, pairing specification on the left side with verification on the right, still matters. What changes is that large language models reduce the time required to turn natural-lang…