We see something that works, and then we understand it
This thought-provoking piece challenges the conventional wisdom that understanding precedes progress, positing instead that 'we see something that works, and then we understand it.' It critiques the 'linear theory of innovation' and 'thinkism,' arguing that real-world breakthroughs in R&D often arise from practical experimentation and observation rather than pure abstract thought. The article's contrarian yet practical view resonates with the Hacker News crowd, particularly its implications for AI development and the messy reality of engineering.
The Lowdown
The post delves into the often-misunderstood process of innovation, asserting that practical success frequently precedes theoretical comprehension, a notion articulated by Thomas Dullien. It challenges the deeply ingrained belief, often fostered in educational settings, that understanding must always come before progress, labeling this approach as 'thinkism'. The author argues that this linear model of innovation, akin to the waterfall model in software, is fundamentally flawed when applied to complex, real-world problems.
- The article debunks the 'linear theory of innovation' by citing historical examples, such as the invention of the pendulum clock occurring before Newton's foundational laws of mechanics.
- It introduces 'thinkism' as the belief that given a problem, sufficient thought will always yield a solution, a concept that works well in structured environments like school or bureaucracy but falters in dynamic fields.
- In research and development, progress often begins with incomplete understanding; discoveries are made by trying sensible things, observing what works, and then formalizing that practical knowledge.
- This practical approach is necessitated by the inherent complexity of the world, where 'nobody knows anything' definitively, and even experts possess partial or slightly inaccurate knowledge.
- The author concludes by offering two key takeaways: prioritize observation and experimentation over abstract thinking for breakthroughs, and temper expectations for AI to solve all problems, as even its vast 'knowledge' will remain insufficient in the face of true complexity.
Ultimately, the article advocates for a pragmatic, iterative approach to problem-solving and innovation, emphasizing the value of hands-on experience and empirical observation over rigid theoretical frameworks.