Imagine you're writing a private letter in a room with thin walls, and someone outside can hear the scratch of your pen well enough to reconstruct every word. That's TEMPEST — every screen refreshes pixels by pushing electrical signals through traces that act as antennas, and those electromagnetic emanations carry a ghost of what's displayed. This paper asks whether Brazil's electronic voting machine, the UEB, leaks ballot content to a nearby radio receiver, and the answer is: yes, if nobody's watching the room. The committed claim is that electromagnetic side-channel attacks pose a practical threat to ballot secrecy in Brazilian polling stations, and that the threat's severity scales inversely with physical oversight — not with cryptographic sophistication. This is not a novel discovery in the TEMPEST literature (van Eck demonstrated CRT eavesdropping in 1985), but it is the first systematic application to Brazil's specific electoral infrastructure, motivated by an actual court case where someone attempted exactly this attack. The architecture is classical signals intelligence: software-defined radio (SDR) hardware captures RF emissions in the frequency bands associated with the UEB's video signal, then signal processing reconstructs the screen image. The method leans on the physical properties of the target — display refresh rate, cable shielding (or lack thereof), and the electromagnetic environment of the polling station. No machine learning, no exotic hardware. The attack surface is the physics of the device, not its software. The experimental regime is where you should squint. The authors propose attack scenarios based on publicly available information about polling station layout and UEB hardware, then run SDR experiments — but the paper's abstract doesn't specify whether experiments used actual UEB hardware or a surrogate display, whether testing occurred in realistic polling-station RF environments, or at what distances reconstruction succeeded. The court-case motivation is compelling as a real-world prompt, but the validation appears to be author-controlled simulation of attack conditions, not an independent red-team exercise against a live election. The ladder here is unusual because the paper isn't competing against other TEMPEST research on reconstruction fidelity — it's arguing for a specific policy response. The baseline is public awareness: the paper's central finding is that oversight by voters, poll workers, and party representatives is the cheapest and most effective countermand. This makes the contribution more defensive doctrine than offensive capability research. Prior TEMPEST work (van Eck 1985, Kuhn 2003, various NATO SDIP standards) established the physics; this paper maps the physics onto a specific democratic institution. The milestone question is about shielding standards. The UEB could adopt TEMPEST countermeasures — shielded cables, filtered power supplies, Faraday-cage booths — but the paper argues awareness is the immediate lever. The next concrete threshold is whether Brazil's Superior Electoral Court (TSE) mandates electromagnetic shielding standards for future UEB generations, or whether awareness guidelines become part of official poll-worker training. Neither has a timeline attached. The obvious experiment not run: a blind red-team test during an actual election, with independent evaluators measuring reconstruction success at various distances and with various levels of polling-station countermeasures in place. The honest read on why: legal and institutional barriers. You cannot conduct offensive SIGINT against a live election in a democracy without extraordinary authorization, and the 2024 court case suggests Brazilian authorities are aware but cautious. The authors are likely constrained by access, not by compute or interest.
