A Lord of Programmers from the plains, keeper of enterprise ways, journeyed to the fabled monastery atop LLM mountain. Word had reached him that Master Foo’s students practiced a heretical form of programming that violated all sacred traditions.
At the mountain’s base, the Lord was pleased to find junior students coding as they should - crafting precise algorithms, writing unit tests for every function, and documenting each variable with scholarly precision. “Perhaps the rumors are false,” he thought. “These disciples follow the ancient ways of enterprise.”
Climbing higher, he encountered intermediate students who channeled chi to conjure parts of their programs - code, tests, documentation, commit messages - materializing from the ethereal realm. They would occasionally alter these manifestations where desired, tweaking a function here, adjusting a comment there. To the Lord’s eyes, this practice appeared sloppy and undisciplined. “Curious,” the Lord muttered, “but surely their seniors maintain proper discipline.”
Near the summit, he discovered the monks doing the unthinkable: they showed no interest in algorithms, wrote no tests, crafted no documentation. Instead, they sat in meditation, learning only the language of human needs and the art of precise conversation. Their programs emerged fully formed from dialogues, like flowers blooming without soil.
Outraged, the Lord stormed to the peak where Master Foo sat weaving natural language into working systems. “Master!” he thundered. “Your monastery mocks the very foundations of our craft! Your beginners practice correctly, but your masters have forgotten everything - data structures, clean code, commit practices the sacred disciplines! How can one who cannot write a sorting function or checks code coverage claim to program?”
Master Foo looked up from his work and smiled. “Rest a little for you have climbed far from the Von Neumann plains to the top of the LLM mountain”
Upon hearing these words, the Lord was enlightened.
Master Po says: In the old way, humans learned the language of machines. In the new way, machines learn the language of humans. The path from human thought to running program grows shorter, but the need for wisdom about human needs grows greater.