快一年了 我现在觉得 做软件是非常愚蠢的 因为ai可以快速逆向复制任何软件 包括各种细节 如果你是个做软件的高手 最终的价值就是成为别人的嫁衣 在中国 任何软件…

On July 12, 2026, an X user claiming to be a software practitioner, 'Yang Yi,' posted a discussion-provoking thread. He asserted: 'Making software is very stupid because AI can quickly reverse-engineer any software, including all details.' In his view, even if you are a software expert, the final value becomes 'another's wedding dress'—especially in China's distribution environment, any software can be reverse-distributed.

This judgment is based on his observations over the past year and is embedded in his 'Five Levers' analytical framework: leasing, labor, content, software, and distribution. Yang Yi believes that with the arrival of AI, the software leverage effect has significantly weakened, while the content and distribution levers have intensified. Therefore, he chose to abandon software development and pivot to marketing, building a distribution network through content, thereby 'accumulating productive assets' to leverage the first two levers.

Yang Yi did not provide technical details on how AI reverse-engineers software. From current technology, AI-assisted reverse engineering (such as decompilation and code generation) is indeed progressing rapidly, but fully replicating complex software (especially those with hardware binding, encryption algorithms, real-time server logic) still faces challenges. Software is not just code; it also includes data, community, brand, and service ecosystems.

Moreover, although reverse distribution exists in a gray market in China, legal risks (such as the Copyright Law and Computer Software Protection Regulations) have not disappeared due to AI. Yang Yi's argument may oversimplify the composition of software value—for example, contracts, support, and compliance in enterprise software cannot be 'copied.'

In Yang Yi's framework, the content and distribution levers are intensified by AI. The content lever intensifies in the short term (AI-assisted generation) but may weaken in the long term due to homogenization; the distribution lever continues to strengthen due to AI matching and recommendation capabilities. He himself has pivoted to marketing, attempting to build a distribution network through content, considering this 'an opportunity left for the masses.'

This partially aligns with current industry observations: for individual developers, the barrier to independent software has lowered due to AI, but competition has intensified and customer acquisition costs have risen. Conversely, content creators (such as knowledge bloggers and AI tool reviewers) leverage AI for efficient production and quickly reach users through platform distribution. However, Yang Yi's conclusion remains a case study lacking large-scale data support.

Yang Yi's assertion has an implicit premise: software value lies only in the scarcity of its 'replicas.' But historically, open-source software has proven that free code can also create enormous business value (e.g., Red Hat, Android). AI reverse engineering may accelerate functional replication, but soft barriers such as brand trust, continuous innovation, and user experience are difficult for AI to fully replicate.

Furthermore, Yang Yi did not distinguish between different software types: tools, platforms, embedded systems, SaaS, etc., each with its own moats. His choice (pivoting to marketing) reflects a rational response of an individual in a specific environment (rampant reverse distribution in China), but cannot represent the general trend of the global software industry. In terms of time dimension, 'close to a year' may not be sufficient to capture the full picture of AI's impact.