A new study reveals that language model behavior can be effectively controlled through geometric methods, specifically by separating angular and radial components in activation steering. This approach suggests that while concepts are primarily represented in angular structure, the norm also plays a crucial role in ensuring stability and effectiveness of interventions.