A recent Chinese blog post has sparked debate by accusing local media of distorting the narrative around AI scaling laws. The author argues that the so-called 'discovery' of test-time compute as a scaling alternative is not a 2026 revelation but a retrospective analysis of a technical deviation that occurred between 2020 and 2022. This post provides a valuable technical breakdown of the evolution from Kaplan scaling to test-time compute, emphasizing that the shift was a gradual correction, not a single 'bug' driven catastrophe. For global developers and researchers, this serves as a reminder to critically evaluate media narratives and understand the historical context of AI research. The piece underscores the need for accurate technical reporting in the fast-moving AI field.
A critical analysis of how Chinese media misrepresented the scaling law debate, revealing it as a technical correction rather than a dramatic new finding.