More capability. Within the same limits.

01 / Foundations

Complex systems need a common language.

Ghost Observatory draws on disciplines developed for defence information exchange and enterprise architecture. We have studied and applied selected principles from JC3IEDM, NIEM, DoDAF and NATO NAF v4 in its development.

Different systems describe different parts of the same operation. Bringing their records together requires explicit meaning, identities and relationships.

Information models

JC3IEDM and NIEM inform our approach to information modelling and exchange.

Architecture

Selected practices from DoDAF and NATO NAF v4 inform how we describe systems, activities, information and their relationships.

Object-centric event data

Our conformance work covers the OCEL 2.0 specification and the IEEE Task Force on Process Mining’s OCED metamodel. OCEL defines event-log exchange; OCED provides the conceptual model. The OCED standardisation process is ongoing.

02 / Correctness

One calculation. One authority.

A plausible number is not good enough.

Our background in cybersecurity and large financial institutions taught us to question what a system accepts as true. In Ghost Observatory, each calculation has one authoritative definition. Records must support the result. Missing information stays unknown, with a stated reason.

That discipline starts with the input. Explicit contracts declare what fields mean and how records relate. Invalid records are identified and excluded. A convincing chart cannot repair an unsupported calculation.

03 / Inspectable results

The result belongs to you.

Ghost Observatory publishes its results in a SQLite file on your machine. You can inspect the tables, run SQL and follow a finding back to its recorded evidence.

The Reader provides an interface to that result. Access to your information does not depend on waiting for a particular chart or dashboard.

The published sample demonstrates this directly. Download the file, run the verification query and compare the result.

04 / Local computing

Modest hardware is a serious target.

We have Ghost Observatory running on a Raspberry Pi 5 with 16 GB of memory.

Detailed analysis should be useful beyond expensive workstations. Memory use, repeated computation and unnecessary data movement become harder to ignore when the machine has less room to absorb them.

Our engineering work asks how much useful analysis we can deliver within those limits, while preserving the correctness of the result.

05 / Graphics

Squeeze more out of every GPU.

We push modest GPUs to deliver more. Our games are where we explore their limits, find bottlenecks and develop ways around them. The aim is richer graphics and responsive play on hardware people already own.

We work with Three.js and WebGPU. Rendering, simulation and input compete for time and memory. Where the GPU spends its time, how much data moves and when work happens all affect what the player feels.

The Observatory Reader faces related constraints: substantial results must remain practical to explore.