Not so locked in any more

Origin: A Casual Conversation About "Changing Your Mind" Tech blogger Simon Willison shared an interesting anecdote: a mid-sized tech company had just used AI coding agents to drive a rewrite of their native iPhone and Android apps into React Native. When asked why they didn't just stick with native apps (since agents lower maintenance costs), the reply was nonchalant: React Native is good enough now, and "if it turns out to be the wrong decision in the future, we can just port back to native." This conversation echoes a comment by renowned developer Mitchell Hashimoto on the Bun project's migration from Zig to Rust: "Programming languages used to be LOCK IN, and they're increasingly not so." Breakdown: How AI Makes "Lock-in" Obsolete Here, "lock-in" refers to being trapped by a chosen technology stack (like a programming language, framework, or platform). Switching costs are prohibitively high—requiring massive code rewrites, team retraining, and bearing business disruption risks. Thus, technology selection has often been a careful, even painful, "lifetime commitment." AI coding agents are changing this core equation. They can understand the context of a codebase and automatically execute large-scale, pattern-based code transformations. Tasks that once took months and large teams—like rewriting Swift to JavaScript or migrating React Native components back to SwiftUI—might now have a draft completed by an agent in days. The marginal cost of migration is dropping dramatically, making "changing your mind" a viable, low-cost option. Trend Insight: From "Architectural Decisions" to "Reversible Experiments" This reveals a deeper trend: the fundamental nature of technology decisions is shifting. In the past, selection was an "architectural decision," focused on long-term commitment and stability. In the future, it may resemble a "reversible experiment." Companies can be bolder in trying new technologies because the "worst-case scenario" is no longer being permanently bogged down by an outdated stack, but merely "spending a few more days having AI rewrite it." This could accelerate the adoption cycle for new technologies and allow startups to adjust their technical direction more flexibly. The competitive dimensions for frameworks and languages are also changing—beyond performance and ecosystem, "AI understandability and migratability" could become a new core competency. Practical Value and Counter-Intuitive Angles For developers and team leaders, this means: 1. Re-evaluating Technical Debt: Some technical debt previously deemed "unacceptable" may now have a lower "interest rate" due to reduced AI migration costs. 2. Focusing on "Migratability": When designing architecture and writing code, adhering to clear patterns, good documentation, and modularity benefits not only human maintenance but also makes it easier for future AI agents to understand and operate your codebase. 3. Being Wary of New Dependencies: While freeing you from technology stack lock-in, you might become newly dependent on the AI programming toolchain itself. Who controls your AI rewriting capabilities? A counter-intuitive point is: this may not necessarily lead to a fragmented百花齐放 in the tech world. On the contrary, when migration costs are extremely low, the industry might converge faster on a few truly excellent, universal standards because "trial and error" and "correction" become too easy. If React Native isn't good, people will leave it faster; if it is good, it will become a de facto standard more quickly. In summary, AI coding agents are not just helping us "write code"; they are quietly reshaping the power dynamic between us and our technology infrastructure. The dissolution of lock-in brings a new kind of freedom, accompanied by new strategic considerations.