skip to content
bijan

systems engineer

← all books

working draft

Platform Integrity

Engineering detection, decisions, and enforcement against bots, fraud, and abuse.

Platforms often begin with the wrong question: “Is this a bot?” A detector can be accurate while the policy is vague, the attribution is weak, the intervention is disproportionate, and the appeal cannot repair the damage. Platform integrity is the larger engineering discipline that connects those decisions.

What it does

Build integrity systems that can explain what they know, why they act, and what happens when they are wrong.

  • Define the ecosystem harm before selecting a detector.
  • Separate direct observations from features, identity links, and inferences.
  • Choose rules, behavioral models, machine learning, graphs, and client controls for the evidence they can actually produce.
  • Match confidence and urgency to a proportionate, reversible intervention.
  • Measure prevented harm, legitimate-user burden, recourse, and attacker adaptation—not accuracy alone.

The argument

  1. 01 / define

    What are we protecting?

    Map actors, incentives, scarce resources, and measurable harm before writing policy.

  2. 02 / observe

    What can we know?

    Build a governed evidence supply chain and keep request, session, device, account, person, and organization distinct.

  3. 03 / decide

    What action is justified?

    Set evidence standards according to severity, breadth, duration, reversibility, and decision latency.

  4. 04 / operate

    How does defense adapt?

    Connect appeals, investigations, metrics, and attacker response into a production learning loop.

Who it is for

  • Staff and principal engineers
  • Security, fraud, and risk leaders
  • Trust and safety teams
  • Data scientists and investigators
  • Product, policy, and platform leaders

How it is being built

The public research trail includes NIST risk and identity frameworks, IETF and W3C standards, peer-reviewed abuse-market research, legislation and enforcement actions, platform contracts, and trust-and-safety scholarship. Each structured note preserves claims, assumptions, limitations, and implications.