Nobody Argued For Your Stack

Last week, it was discovered that Cursor had mostly completed their transition from SolidJS to React, a process that had begun about seven months prior. However, the migration gained significant attention after the Solid 2.0 RC release. Yesterday, the Anthropic docs example command for their large-scale migration feature caught my attention, prompting reflection on the situation.

My initial reaction was negative, as the example seemed to highlight the very technology they aimed to migrate away from, coinciding with the release of Cursor's biggest update yet. I initially dismissed this as part of a recent news cycle, but further investigation revealed that this example had been present since at least April 2026, four months before Cursor's story broke.

The Anthropic example was mentioned in passing within a January blog post on Cursor, without any accompanying reasoning, benchmarks, or arguments. This raises questions about the origins of such examples and the importance of public discourse in informing migration decisions. In today's industry, conclusions are often broadcasted without proper arguments, and these verdicts can be produced rapidly with minimal reasoning.

This shift has been accelerated by the collapse in execution costs, which has made migrations cheaper and easier in various directions. While this may seem like a positive development, the lack of public arguments and reasoning has led to an oversensitivity towards circulated verdicts. The industry seems to be moving away from the days when arguments were the source of truth, instead embracing a culture that values direction over debate.

This trend is evident in the Anthropic example, where no public reasoning or benchmarks were provided, and the Cursor blog post only criticized performance issues without presenting any data to support its claims. The most convincing arguments are those that can be engaged, checked, narrowed, refuted, or strengthened, whereas verdicts can only be reiterated.

In the case of Cursor, the post characterized performance issues as "perf footguns" and "accidental fan out" without offering any compelling arguments or evidence, ultimately making room for the React Compiler as the solution. This absence of a public debate leaves us with fewer opportunities to evaluate and improve our technology choices.