Taste Is All That's Left

In an era where creating software is becoming increasingly effortless, the most crucial aspect of the craft seems to be fading away. The author reflects on the days when building web applications required countless hours of typing, reading manuals, and struggling with APIs. Those challenging times separated the software creators who could deliver functional programs from those who could only talk about it, and the wall that defined the craft of software development has started to disappear.

The distance between an idea and a working program has dramatically shrunk, thanks to advancements in technology. It now takes significantly less time to describe an idea and receive a plausible version of it compared to the painstaking process of manual typing. However, an important consideration arises: the value gained from mastering the original process of building software is not lost. Instead, it undergoes a transformation.

The author questions the relevance of the question "Are the machines any good?" and argues that the focus should shift to the output, which is generally good enough. Good enough has become a solvent that dilutes the drive to improve, as the expense of creating software no longer functions as a filter to prevent the creation of mediocre variations of features. The economics of software development have changed, removing the cost-based barrier that once filtered the output.

This change in the landscape has far-reaching consequences. With the friction removed, developers can generate software fluently from the start, resulting in faster output but at the cost of not learning the importance of discernment. The absence of a rigorous filter means that developers may ship subpar versions without realizing it, as they no longer experience the pain of creating a bad version and having to live with its failure.

The author emphasizes that taste—the ability to recognize the quality of a piece of software—remains an essential aspect of the craft. It is a skill that develops through the painful process of making mistakes, facing their consequences, and learning from them. Taste is not a soft, subjective preference; rather, it is a sharp, wordless verdict that sets apart the right from the wrong. It is honed through repeated exposure to failure, which shapes the palate and hones the judgment.

While removing friction may accelerate the software creation process, the author warns that it comes at the expense of developing taste and judgment. Developers who acquire fluency without experiencing the rigorous trial and error of climbing the initial walls may be more productive in the short term but lack the ability to discern the right from the wrong. This quiet cruelty of removing the painful learning process raises essential questions about the future of software development.