IBM's recent patent on using derivatives to find convergents of generalized continued fractions has raised eyebrows. The core of the issue is that this technique, rooted in the work of mathematicians like Gauss, Euler, and Ramanujan, has been around for over 200 years. Essentially, IBM implemented this well-established number theory method in PyTorch and sought patent protection. The move allows them to potentially charge fees for a mathematical concept that should remain open for all to use. The paper "CoFrNets: Interpretable Neural Architecture Inspired by Continued Fractions" by Puri et al. argues that continued fractions can enhance neural network design. The authors rebrand the concept as "ladders" and introduce new terminology, framing it as if they are pioneering a novel approach. However, critics point out that their methods closely mirror existing knowledge in number theory, and their results—an accuracy of just 61% on a non-linear dataset—fail to match state-of-the-art performance. The implications of IBM’s patent extend beyond academia. Mechanical engineers and robotics experts, who use continued fractions in practical applications like gear design, could face legal repercussions if they employ similar techniques. Even educators and students learning these concepts may find themselves at risk. The article highlights a troubling trend where companies can stake claims over mathematical principles, potentially stifling innovation and collaboration in fields that rely on these long-established methods.