Skip to content

Decoy Binding or Sink

Artifact — instantiates Selective Pathway Suppression

Plants a sacrificial look-alike that soaks up a pathway's activator before it can reach the real mechanism, starving that one pathway while others keep their supply.

A Decoy Binding or Sink is a built object — a sacrificial, inert target — placed where a pathway's activator travels, engineered to look enough like the real destination that the activator engages the decoy instead. Bound to the decoy, the activator is spent: it never reaches the mechanism it would have driven. Unlike its siblings, this brake acts on neither the control point nor the driving signal; it acts on the supply of activator, depleting the pool that feeds one chosen pathway. Because a decoy only captures what it resembles, the pathways that don't run on that activator keep their supply and carry on — the suppression is selective by affinity, and its strength is a matter of capacity: a sink works only while it has room to absorb.

Example

A security team is watching automated credential-stuffing hammer their login service — a botnet trying stolen username/password pairs at scale, hunting for account takeovers. Rather than block traffic wholesale (which also catches real users on shared IPs), they seed the environment with honeytokens: fake credentials and dummy accounts salted into the places attackers scrape, indistinguishable from real ones to an automated tool. The bot harvests them and pours its attempts into accounts that lead nowhere. Each attempt spent on a decoy is an attempt not spent on a real account — the takeover pathway is starved of the very guesses that drive it, while legitimate logins, which never touch the honeytokens, proceed untouched.

The design turns on two numbers. The decoys must be attractive enough that the automated attacker engages them at a high rate (affinity), and there must be enough of them relative to the attack volume that the sink doesn't saturate and let real accounts see the overflow (capacity headroom). Touching a honeytoken also trips an alarm — a bonus of the mechanism — but the suppression itself is simply the activator being spent on the wrong target.

How it works

The artifact intercepts by absorption, upstream of the mechanism:

  • Model the activator and its real target. Know what the pathway's activator engages with, so the decoy can be shaped to be mistaken for it.
  • Build a convincing, inert look-alike. Fabricate a target with high apparent affinity for the activator but no downstream effect once engaged — it captures and holds, it does not act.
  • Place it in the activator's path. Position the decoy where the activator travels before reaching the mechanism, so binding happens early.
  • Size the sink for the load. Provide enough decoy capacity to absorb the expected activator flux with headroom, since a saturated sink stops suppressing.

Tuning parameters

  • Decoy affinity — how strongly the decoy attracts the activator versus the real target. Higher affinity captures more but is harder to make convincingly inert.
  • Sink capacity — total activator the decoy pool can absorb before saturating. The dial that sets how much load the brake can hold; too small and the pathway breaks through on a surge.
  • Selectivity — how narrowly the decoy captures the target activator and not molecules or traffic the system needs. Sloppy decoys drain useful supply too.
  • Placement / proximity — how far upstream the sink sits. Earlier interception is more thorough but requires anticipating the activator's route.
  • Replenishment — whether spent decoys are consumed once or refreshed. A regenerating sink holds against a sustained load; a one-shot sink is a finite budget.

When it helps, and when it misleads

Its strength is stopping a pathway without ever touching its machinery or its source — the mechanism keeps working perfectly, it just goes hungry — and doing so with real selectivity, since only what resembles the decoy is captured. It shines when the activator is diffusible and interceptable and when a tripwire on decoy engagement is a welcome side benefit.

Its hard limit is saturation: a sink absorbs only up to its capacity, so a large enough activator surge fills the decoys and spills straight through to the real target, and unlike a metered brake this failure is abrupt.[1] Decoys can also be too good — draining an activator that other, legitimate pathways also needed — or, over time, distinguished from the real target by an adapting adversary or process, at which point they stop being taken. The classic misuse is sizing the sink to the average load and being overrun by the peak, or leaving a once-effective decoy in place long after it has been learned around. The discipline is to size capacity against the worst expected flux with headroom, confirm the decoy stays indistinguishable, and pair it with a monitor that warns before saturation.

How it implements the components

Decoy Binding or Sink fills the fabricate-an-absorber, deplete-the-supply subset:

  • inhibitory_agent_definition — the decoy artifact is the inhibitory element this mechanism defines: a sacrificial, inert target built to be engaged in the mechanism's place.
  • target_non_target_selectivity_map — the decoy must resemble the real target closely enough to capture its activator while sparing molecules or traffic the system needs; that resemblance boundary is the selectivity map.
  • safety_margin — sizing the sink's capacity above the expected activator load, so a surge doesn't saturate the decoys and break through, is the margin this artifact is engineered around.

It depletes the activator upstream but does not occupy the pathway's own control point (that's Competitive Occupancy Inhibition) or halt the activity on a live metric (that's Circuit Breaker Pause); watching for an adversary that learns to avoid the decoy is Washout and Rechallenge and the resistance monitors.

References

[1] The pattern appears across domains as a decoy receptor or molecular sponge in biology and a honeypot / honeytoken in security: a high-affinity, non-functional target that sequesters an agent away from the real one. All share the same hard edge — a sink suppresses only until it saturates. ↩