AI's Glut: Why Puzzle Makers Resist the Slop

Overview

The rise of generative artificial intelligence has created a paradoxical glut in the digital content marketplace, prompting creators in niche sectors like puzzle design to resist the integration of modern AI tools. Unlike authors of prose fiction, puzzle makers argue that their craft relies on deterministic logic and algorithmic precision rather than probabilistic language modeling. This distinction has led to a growing divide between "frontier-class" AI systems and the traditional software engineering methods that have long underpinned creative industries.

For many independent creators, the definition of AI being enforced by major distribution platforms feels reductive and misaligned with their actual workflows. The tension is most visible in how platforms like Amazon KDP are now demanding disclosure of AI usage, a requirement that fails to account for the nuanced reality of software-assisted creation.

The Logic of Puzzle Making

The foundation of modern puzzle design, particularly for formats like Sudoku, rests on computer-generated algorithms rather than human hand-construction or generative text models. Creators such as Jim Reynolds, who has published puzzles online for over two decades, emphasize that their work is produced through custom-written software in languages like C++, Perl, and Python. These programs are designed to ensure unique solutions, accurate difficulty measurement, and logical consistency—qualities that are fundamentally different from the pattern-matching output of large language models.

"My puzzle making software... would have been considered a type of AI, back in the 1970s, back when we thought of chess software as a kind of AI."

Reynolds notes that the majority of newspaper Sudoku puzzles published since the 2004 craze have been computer-generated, a practice solvers generally accept without issue. The complexity of modern puzzles, such as kaleidoscope tessellations or cross-figures, often exceeds what is feasible through manual construction alone. Consequently, these creators view their code not as "AI" in the contemporary sense, but as sophisticated power tools that enable volume and precision.

The distinction is critical. Contemporary AI, often referred to as "neural networks" or models composed of vectors mimicking biological neurons, operates on probability. In contrast, puzzle generation requires deterministic logic. A puzzle with multiple solutions or logical gaps is considered defective, whereas an AI-generated text might be factually incorrect yet still linguistically coherent. This structural difference means that the current wave of AI disruption does not neatly apply to all forms of digital content creation.

The Glut and the Response

The proliferation of AI-generated content has flooded marketplaces with low-effort material, leading to what some creators describe as a "slop" phenomenon. In response, a segment of the community is deliberately rejecting these tools to maintain quality and authenticity. Reynolds observes that his audience size has remained remarkably constant over five years, a stability he attributes to a deliberate "keep it steady" approach rather than a pursuit of growth.

This counter-cultural stance extends to technical preferences as well. While interactive digital puzzles were once seen as the future, there is a recent reverse trend where users are "unplugging" and returning to printable formats. Reynolds welcomes this shift, noting that interactive puzzles are significantly harder to produce and that the tangible nature of print puzzles is often perceived as having greater value by consumers.

The influx of AI content has also changed the economic landscape for small businesses. Reynolds argues that steady growth can be detrimental, likening it to "cancer" that ruins the enjoyment of the work. By maintaining a fixed-size audience, creators can preserve the quality of their output and their personal satisfaction, avoiding the headaches that accompany rapid expansion. This philosophy stands in stark contrast to the growth-at-all-costs mentality prevalent in most small business models.

Platform Policies and Ethical Ambiguity

The friction between creators and platforms has intensified with new disclosure requirements. When Reynolds prepares to self-publish a new book of Killer Sudoku via Amazon KDP, he is asked if "AI" was used to produce the book. This is a new question that did not exist for his previous works, reflecting a platform-wide push to label AI-generated content.

However, Reynolds intends to answer "no," arguing that the question was designed for prose fiction authors writing AI-generated novels, not for software engineers using code to generate logical structures. He acknowledges that the answer is complicated, as he increasingly uses coding models and harnesses, such as Claude Code, to aid in the production of his software. Yet, he maintains that using an AI assistant to write code is fundamentally different from using an AI to generate the final creative product.

The lack of nuance in these policy questions creates ethical ambiguity for creators. Reynolds warns against polarized opinions, urging the public to keep an open mind. He believes there are valid arguments on both sides of the AI debate, but the current discourse often fails to distinguish between AI as a tool for code generation and AI as a generator of content itself. For now, many puzzle makers choose to define their work by its logical integrity rather than the tools used to facilitate its creation.