Are Some Natural Languages Easier for Algorithms to Understand Than Others?

When it comes to describing algorithms, natural languages pose significant challenges. The intricacies and nuances inherent in human languages often impede clear and unambiguous communication of computational procedures. This has been highlighted by numerous researchers and practitioners in the field, including Prof. Rutishauser, who emphatically stated that 'an algorithm is an Algol program.'

Algorithms vs. Natural Languages: Clear vs. Ambiguous Expressions

Prof. Rutishauser's assertion points to a fundamental divergence between the structured and precise nature of algorithms and the ambiguous and flexible qualities of natural languages. When we talk about algorithms, we are referring to a set of well-defined steps that a machine can follow to perform a task or solve a problem. These steps must be clear, unambiguous, and devoid of any interpretative leeway. An algorithm, such as 'Then you iterate until epsilon is small enough,' requires a specific definition of 'small enough.' Without it, the algorithm cannot function as intended.

Prof. Rutishauser's Pseudo Algol: A Step Towards Clarity

Prof. Rutishauser addressed this issue through his work with On Jacobi Rotation Patterns. In this paper, he used a pseudo-Algol language that incorporated the structured syntax of a programming language like Algol but retained the natural language descriptions for steps and loops. The key difference is that while loops are clearly defined in Algol, the rotations are described informally, such as 'rotate about... so that... is zero.' This hybrid approach helps to bridge the gap between the structured requirements of an algorithm and the natural language descriptions.

Translation and Algorithmic Identicality

The importance of clear and precise descriptions was further emphasized when the author found that two seemingly distinct algorithmic papers were actually identical when translated into a pseudo-Algol format. These papers used a natural language approach with a style resembling assembly language, described with steps like '1. 2. then repeat steps ... to ... etc.' Although the natural language description might seem clear to humans, the lack of strict definitions and precise terms made it challenging for the author to discern differences.

Lessons Learned and Future Implications

The translation exercise underscores the critical need for precise and unambiguous language when describing algorithms. The lesson learned is that natural languages, despite their richness and flexibility, are utterly inappropriate for describing algorithms. This has significant implications for the field of AI interpretation and algorithm design. Future research and practice should focus on developing clearer and more precise communication methods to ensure the effective implementation of algorithms by AI systems.