A disastrous software deployment at Knight Capital left an outdated code path active on one server, causing millions of erroneous orders that wiped out $460 million in seconds and highlighted critical failures in testing, monitoring, and incident response processes.
The paper examines how machine‑learning systems inherit unique forms of technical debt—such as boundary erosion, entanglement, hidden feedback loops, and data‑dependency issues—and discusses mitigation strategies, measurement techniques, and cultural changes needed to maintain sustainable, reliable ML deployments.
This talk shares the experience of a small machine‑learning team at Dangdang, describing how they built a recommendation system from scratch, the tools and processes they used, the challenges of limited personnel, and the many pitfalls they encountered while iterating toward a production‑ready solution.
The article argues that rewriting already‑deployed code rarely adds real product value, incurs high costs and risks, and should only be considered when the existing system is so outdated that a fresh start is the only viable solution.
In this talk, senior architect Neeke shares practical experiences and lessons on software architecture and refactoring, using bridge and building analogies to illustrate how business‑agnostic design, careful analysis of existing support points, and incremental restructuring can transform fragile systems into robust, maintainable solutions.
The article examines why software architectures inevitably degrade in long‑running projects, explains the stages of decay, and presents practical strategies—such as upgrading environments, phased and distributed builds, component isolation, and platform‑plus‑app designs—to keep systems maintainable and performant.
The article presents Uber engineering leader Raffi Krikorian’s twelve practical rules for planning, executing, and managing software architecture refactoring, emphasizing clear purpose, measurable goals, incremental delivery, data‑driven decisions, technical debt control, realistic technology choices, and stakeholder communication.
This article offers goal‑oriented, data‑driven advice and checklists for architecture refactoring, emphasizing realistic technology selection, pressure management, business understanding, handling non‑technical influences, code‑quality practices, and team readiness to ensure effective, sustainable redesigns.
The article presents Uber engineering manager Raffi Krikorian’s twelve practical rules for planning, executing, and evaluating software architecture refactoring, emphasizing clear purpose, measurable goals, incremental delivery, current state assessment, data‑driven decisions, and diligent management of technical debt.
The article analyzes how long‑running software projects inevitably suffer architectural decay due to growing codebases, team size, and build complexity, and proposes practical strategies such as modularization, phased and distributed builds, component isolation, and disciplined technology adoption to keep systems sustainable.
The article recounts how a decades‑old, unused code path was unintentionally re‑activated during a rushed deployment of the Retail Liquidity Program, leading Knight Capital to send erroneous orders that caused a $460 million loss and the firm’s bankruptcy.
The article argues that striving for flawless code is futile because most software constantly evolves, and it advocates focusing on correctness, readability, and pragmatic refactoring rather than chasing unattainable perfection.
