# What Xerberus Measures
Xerberus measures the mechanisms that make a crypto object work, then scores the risk factors that are unique to those mechanisms.
An object can be an [[Object Classes|Asset, Protocol, Pool, or Organisation]]. Each object class has its own mechanism map because each one fails in different ways.
---
## The Core Idea
Every crypto object is built from mechanisms:
| Object | Example Mechanisms |
|--------|--------------------|
| **Asset** | Mint/burn authority, emissions, vesting, collateral backing, peg design |
| **Protocol** | Smart contracts, oracle dependencies, liquidation logic, governance controls |
| **Pool** | Strategy logic, share mechanics, exposure parameters, fees |
| **Organisation** | Leadership structure, operating capacity, legal entity, licensing |
Each mechanism has a different risk surface. A timelock can fail because the delay is too short. A share token can fail because its price-per-share is manipulable. A pool strategy can fail because it concentrates exposure in a fragile dependency.
The dendrogram keeps those questions attached to the mechanism where they belong.
```
Object
-> Mechanism
-> Unique risk factor
-> Evidence-backed safeguard score
```
The rating is the aggregate view of this map. The map itself is the important part.
---
## What Gets Scored
Each activated mechanism contains **subscores**. A subscore asks one narrow question about a safeguard:
> Is this safeguard present and effective?
The answer is a containment score, called **K**, between 0 and 1:
| K | Meaning |
|---|---------|
| **1.0** | The safeguard is present and effective |
| **0.0** | The safeguard is absent or ineffective |
| **Between 0 and 1** | The safeguard is partially effective |
Every subscore must be tied to a real historical incident. If a failure mode has never happened, it does not become a Xerberus risk factor.
---
## What Does Not Get Scored
Xerberus does not score:
- **Price direction.** A rating is not a prediction that an asset will go up or down.
- **Market sentiment.** Social attention and price action are not mechanism safeguards.
- **Hypothetical risks.** Subscores require a real incident or a structurally analogous incident.
- **Mechanisms that do not exist.** If a pool has no withdrawal queue, queue-specific branches are pruned.
- **Systemic contagion inside the dendrogram.** Dependency propagation is a separate graph layer.
---
## Why This Is Useful
Single scores hide where risk actually lives. Two pools can both score 0.74 for completely different reasons: one may have strong contracts but weak withdrawal controls; another may have clean pool mechanics but fragile underlying assets.
The Xerberus methodology keeps the rating decomposable:
```
Pool score
├── Strategy mechanisms
├── Share-mechanics mechanisms
├── Exposure mechanisms
├── Fee mechanisms
└── Connected asset / protocol / organisation context
```
This lets users see whether the risk comes from the pool itself, the protocol it uses, the asset it holds, or the organisation behind it.
---
## Containment Quality
The scored property is **containment quality**: how well a mechanism is protected against known failure modes.
The older U-I-K model separated Uncertainty, Impact, and Containment. Xerberus now scores only K because it is observable and actionable. Low `K_timelock_duration` means "the timelock delay is not strong enough." Low `K_admin_scope` means "admin privileges are too broad."
For the reasoning behind this simplification, see [[design/Why K-Only]].
---
## Intrinsic And Systemic Risk
The dendrogram measures **intrinsic risk**: the object's own mechanisms and safeguards.
**Systemic risk** asks how failure travels between objects. A pool can depend on a protocol, a protocol can issue an asset, and an organisation can operate a protocol. Those relationships are modeled through a dependency graph, not inside the dendrogram.
This separation keeps the basic method simple: first understand each object's own mechanism map, then understand how objects connect.
---
## Related
- [[Object Classes]] — What gets rated
- [[methodology/The Dendrogram]] — How containment is measured in practice
- [[methodology/Subscores]] — How containment is measured at the subscore level
- [[Rating Scale]] — How scores should be interpreted
- [[Independence]] — Why the rater's independence matters for these measurements