Gwern.net Discusses On Really Trying (2009)

A recent post on Gwern Branwen’s personal blog, titled “On Really Trying,” has attracted attention in the tech community, earning 66 up‑votes and 41 comments on Hacker News. The article examines the difference between merely attempting a task and engaging in sustained, intentional effort—what the author terms “really trying.” It argues that this deeper level of engagement is crucial for achieving mastery, whether in human skill acquisition or in training machine‑learning models.

Branwen outlines the key components of “really trying,” including deliberate practice, systematic feedback, and incremental refinement. He draws parallels to reinforcement‑learning systems, noting that an agent’s performance improves most when it is consistently exposed to varied challenges and rewarded for progress rather than for surface‑level attempts. The post also discusses how this mindset can be cultivated in educational settings and workplace productivity, suggesting that metrics of genuine effort may better predict long‑term success than simple completion rates.

The piece has sparked discussion among readers about the applicability of the concept to both human and artificial agents. Commenters have debated the practicality of measuring “really trying,” the role of intrinsic motivation, and the potential for such an approach to inform the design of more robust AI systems. The conversation underscores a growing interest in aligning human effort metrics with algorithmic training objectives.