Salt and sand

The standard intrusion-detection options are cameras, motion sensors, door contacts, and the network and battery infrastructure each depends on. Against a domestic adversary with consumer tools they are adequate. Against a state-class adversary with budget and time they are not — the camera firmware can be patched, the sensor's wireless signal can be jammed during the entry window, the network gateway can be controlled, the alarm log can be rewritten on the way out. Anything with a chip is in some sense negotiable. The defender who relies entirely on the technology stack is relying on a stack the adversary may know better than the defender does.

Analog methods do not have this problem. They cannot be hacked over the wire because there is no wire. The defender's only requirement is that the adversary cannot put back what they have disturbed, and a small set of physical-world phenomena make that requirement easy to satisfy.

The mechanism is informational entropy. A handful of salt poured loosely onto a flat surface forms a random distribution — each grain's position is approximately independent of every other grain's. The configuration space of even a modest scatter is astronomical: tens of thousands of grains, each at one of a near-continuous range of positions and orientations, each with its own crystal facets catching light differently. There is no algorithm that lets a person — or a person with a phone full of reference photos — reproduce the exact configuration after disturbing it. Even an attempt to re-pour salt over the disturbed area will produce a visibly different pattern, because the new pour follows the new entropy and the photo of the original pour is the only record of the original entropy.

This is what makes the technique structural rather than skill-dependent. The adversary's countermove is not learn a technique or bring a device; it is reproduce a random pattern from memory, and the laws of information theory put a hard floor under how much time and effort that takes. The floor is high enough that no operational window allows it.

The materials.

• Table salt — white, high-contrast against dark surfaces (wood floors, dark countertops, dark luggage), readily available, sticks slightly to damp surfaces (a tiny advantage; can be a tiny disadvantage if humidity rises during your absence). The grain size is small enough that even a quarter-teaspoon produces a usefully complex pattern.
• Fine sand — better against light surfaces; the natural color variation in beach sand adds another entropy dimension; behaves more like a fluid than salt, which means it shifts more readily under airflow.
• Other options for contrast control — coarse-ground pepper, glitter, fine kinetic sand, decorative aquarium gravel for outdoor settings. The criterion is visual contrast against the substrate and granular enough to produce high-entropy distributions.

A small spice jar of the appropriate medium kept in a discreet location is the entire infrastructure required.

Where to apply.

• Entry thresholds. A scatter just inside a doorway, in the area shoes would necessarily cross. The first step into the room disturbs it.
• On top of boxes and luggage. A diagonal sprinkle across the lid of a closed box, suitcase, or computer case. Anyone who opens the lid disturbs the distribution.
• Across drawers. A line of grains across the inside edge of a drawer just behind the rim. Cannot be reached without sliding the drawer.
• On top of document stacks. A scatter on the top page of an undisturbed pile. Reading the documents requires lifting the top page.
• Across keyboards or device surfaces. A few grains across keys or trackpad. Use disturbs them.
• Hotel rooms. Inside the safe, around the perimeter of the luggage when stored, on the floor of the closet. The principle scales to any space where the defender will be absent for hours.

Capture and comparison.

The photograph is the record. Use the highest resolution available, with the camera held as close to perpendicular to the surface as practical, and take more than one — slightly different angles, both with and without flash, with consistent framing so that a feature-aligned image diff can be run later. Note the time and the camera position. If the same surface is being checked repeatedly, fixing the camera's position with a small floor mark or by aligning to a reference object (the corner of the box, the edge of the drawer) makes after-comparison much faster.

For the comparison itself:

• Direct visual comparison is sufficient for any disturbance involving footstep-scale or hand-scale motion. The eye is excellent at noticing displaced clusters and broken patterns against a familiar reference.
• AI-assisted image diff is the tool for subtle disturbance — a single brushed grain, a slight perimeter shift, the difference left by careful adversaries who tried to minimize their footprint. Any of the modern multimodal image models can be given a pair of photographs and asked what differs between these two scatter patterns, with reasonable results when the two photos are taken from approximately the same camera position. Open-source feature-alignment plus a structural-similarity (SSIM) diff is the deterministic alternative.

Why it beats single-point tells.

The classic spy-novel tricks — a hair stuck across a doorframe, a piece of tape under a drawer, a paperclip inside a folded page — are single-position bets. They produce one bit of information per check (was it disturbed or not) and they fail in two distinct ways: an alert adversary may notice and restore them (because the restoration target is one element in one position), and they may be disturbed by environmental factors and produce false positives.

A granular scatter produces effectively infinite information per check. Restoration is not within the adversary's operational budget even if they notice the technique. Environmental disturbance produces visible patterns of disturbance (a uniform shift suggests airflow; a small displaced cluster suggests an insect; a cleanly disturbed area in the shape of a shoe suggests a person) that the defender can interpret rather than misread as a binary.

Failure modes and limits.

• Self-disturbance. Do not deploy where pets, children, or your own normal movement will disturb the scatter. The technique only works where the defender is sure they will not produce a false positive themselves.
• Humidity. Salt becomes sticky in high humidity and can clump differently between deployment and check. Sand is more stable across humidity ranges.
• Vibration. Heavy nearby construction or extreme airflow events can produce uniform shifts. The pattern of shift usually distinguishes these from a human intrusion.
• Conspicuousness. A scatter visible on a polished floor is a sign the defender is doing something unusual. Place where natural ambient texture provides some cover, or where the scatter reads as ordinary mess. In high-stakes contexts the scatter itself becomes a signal to the adversary that this defender is alert; consider whether that is or is not a useful signal to send.
• Single-photo reliance. If the defender's only reference photo is on a device that the adversary can access, the technique is weaker. Keep the reference somewhere off-device — a cloud account separate from the workstation, an external drive, a printed copy.

The deeper principle.

The technique works because it forces the adversary into the physical world, where the laws of thermodynamics constrain what they can do in a given time window. The adversary's advantages — patched firmware, network access, distributed computation, institutional resources — are all symbolic-domain advantages. They translate poorly when the question is can you reproduce this specific arrangement of ten thousand independent grains in the four minutes you have before the resident returns.

Analog defenses that exploit physical entropy are one of the few places where a single individual with no budget can produce something a state actor cannot defeat without unreasonable effort. They are worth using.