In the ever-evolving landscape of social media platforms, understanding their algorithmic behaviour is crucial to ensure transparency and accountability. Adversarial algorithmic auditing has emerged as a method to scrutinize platform algorithms, often utilizing emulated accounts to gather data in the absence of official APIs. However, this approach presents challenges due to the need to bypass platform safeguards. Notably, this auditing is typically easier to conduct on desktop browsers rather than mobile applications, which are more secure.

While popular social media platforms like TikTok are predominantly accessed through mobile devices, the majority of algorithmic research is conducted on browsers. This disparity between research methodology and actual platform usage raises concerns about the validity of findings, as mobile behavior might not align with browser behavior.

Recognizing this gap, a pioneering project aimed to bridge this discrepancy by analyzing algorithmic behavior variations between TikTok's web and mobile platforms. The project involved developing a prototype to automate actions on the TikTok mobile app and closely monitor its behavior. To enhance the research, collaboration was established with Junkipedia, an organization with a similar tool designed for browser-based audits.

The research team conducted parallel experiments, simultaneously performing identical actions on both the TikTok mobile app and the Junkipedia infrastructure, using identical IP locations. The objective was to unveil potential differences in algorithmic behavior when comparing web and mobile experiences.