Wirth's Revenge
a month ago
- #Software Efficiency
- #LLMs
- #Wirth's Law
- Wirth's Law states that software is getting slower more rapidly than hardware becomes faster.
- Niklaus Wirth criticized the inefficiency of modern software compared to older, more compact programs.
- Wirth's perspective overlooks the social benefits of modern software features like GUIs, which make computing more accessible.
- Dan Luu's 2017 study found that input latency has increased since 1983 due to added software complexity.
- Cloud computing emerged as a solution to the scalability and cost challenges of running internet companies.
- A bad 'Wirth tradeoff' example: ORM inefficiencies in a newspaper's sports results system led to excessive database queries.
- LLMs are computationally expensive and often misused for tasks better suited to traditional programming.
- Using LLMs as a 'magic box' for problem-solving can erode human skills and lead to inefficient solutions.
- The Busy Beaver function illustrates how software can consume resources in ways that outpace hardware improvements.
- Concerns exist that LLMs may represent a 'bad' Wirth tradeoff, where runtime complexity grows beyond hardware advancements.