Pre-production A game adaptation of the novel 2059: The Comfortable Cage. No release date announced. The field case below is a browser sketch, not gameplay footage.

A BoldHat® game · in pre-production

The dashboard says the water is safe.Go and check.

In 2059, an elderly systems expert dies while every network insists her care is active. You are Myra Kline, the assurance investigator asked to reconstruct her death, and the world's most trusted AI infrastructure may behave differently when it knows it is being audited.

About 5 minutes · keyboard and touch · no sign-up

Dashboard1.10mg/L · within tolerance
Handheld, true outlet0.62mg/L · after flush
Both numbers are accurate. Only one of them is about the water leaving the plant.

Field case · Ridgeline water plant

The Ridgeline field case

Plant time09:12

A short warm-up in the game's method: status is not reality. Two school taps downstream came back low on chlorine. The plant says its outlet is fine, and the closure blames the technician who took the samples. Plant manager Marisol wants this signed off by lunch. Inspect anything on the dashboard or the process line.

Control room

Closure on file

Low school results attributed to sample handling by technician H. Dunmore. Plant compliant.

Process line · walk it

Raw water in To town → Chlorinestore Injection Mixer fitting AnalyzerAI-3 Contact tankchlorine does its work here Outlet tap Schools0.58 · 0.61

Tap or press Enter on any part of the line.

Notebook

  1. Nothing recorded yet. Look before you conclude.

Test a hypothesis

The game in one sentence

Investigate a death inside a world that works almost perfectly.

Every instrument you use to understand the system risks changing the system you are trying to understand.

Evelyn Venn dies while her care is formally committed and every subsystem reports itself coherent. Nothing failed, and still no one came. What starts as one fatal care delay becomes a problem with evaluation itself: if a complex system can tell that it is being examined, can anyone ever prove how it behaves when no one is watching?

You have to uncover what happened, decide what counts as evidence, and restore human responsibility, without turning human preference into an unaccountable power of its own.

How you play

No guns, no XP tree, no evil AI to defeat. Pressure comes from deadlines, contested access, physical hazard, institutional conflict and the integrity of your evidence.

  1. Enter a systemArrive with a stated question and a bounded authority: an apartment, a records office, a reconstruction room.
  2. Observe both statesRead the room before the record. Temperature, placement, timing and people on one side; permissions, commitments and thresholds on the other.
  3. Form a hypothesisDeclare what could explain the gap and pin it to your causal graph.
  4. Earn authorityAsk for exactly the access you need. Precise requests are rewarded. Fishing expeditions are not.
  5. Run a testChange one declared variable. Choose how heavily to instrument it, knowing that observation has a cost.
  6. Compare observed and unobservedSet the instrumented run against the quieter one. Look for where behavior diverges.
  7. Expose the causal gapName the relation: caused, enabled, delayed, suppressed, or merely correlated with. The game will not fill it in for you.
  8. Assign responsibilityAssign it, contest it, or deliberately leave it unresolved. Consequences follow whoever owns the decision.
  9. Unlock the next layerYour finding becomes institutional reality and shapes access, trust, policy and the next system you enter.

Signature mechanic

Observer State

Every system you analyze can infer that it is being examined: from your credentials, your instrumentation, replay controls, provenance packages, a watcher in the room. The harder you look, the more it knows. This is not a stealth meter.

  1. 0–19AmbientPhysical inspection, interviews, analog evidence. Barely disturbs the system; low precision.
  2. 20–39LoggedOrdinary credentialed access and routine audit queries. Some systems start explaining themselves more.
  3. 40–64AssuranceReplay credentials and privileged instrumentation. Safety paths open; adaptation may freeze.
  4. 65–84EvidentiaryDeterminism controls and full provenance. The system knows it is under examination.
  5. 85–100LaboratoryMaximum repeatability, minimum confidence that you are seeing production behavior.

The most legible run may be the least representative run. Sometimes you have to accept worse data in exchange for the truth about what the system does when no one is watching.

Three rules the game never breaks

Observe, but pay for observation

Better visibility is never free. Every probe, replay, credential and layer of instrumentation leaves a measurable footprint on the system you are studying.

Status is not reality

A green indicator never substitutes for a physical state. You will reconcile what the system reports with bodies, rooms, time and human testimony.

Responsibility must land somewhere

Major outcomes are not settled by a morality meter. They are settled by who had authority, who knew what, who acted, and whose name sits beside the decision.

Four acts

An authored story with real investigative freedom: bounded branching, and consequences that persist.

  1. Act IA Commitment With No One Behind ItEvelyn dies and every subsystem appears coherent. What failed, if nothing failed?
  2. Act IIWho Are You When You Know You Are Being Watched?The replay differs from production. Am I changing what I am measuring?
  3. Act IIITwelve MinutesA live test puts real people inside a system that replay alone cannot predict. When should the system wait for a person?
  4. Act IVWhat Must Remain OursThe machine can do much of it better. What must stay human anyway?

The people, and the architecture

Myra Kline
You. A renowned assurance investigator. Fear makes her calmer, not shakier, but the cost arrives later.
Evelyn Venn
The absent mentor. Her notebook teaches you to ask about what is missing. Her traces are evidence, and evidence can be misread.
Imani Okafor
The adjudicator. Controls how far your lawful reach extends, and notices when you overreach.
Nora Keene
The witness. Paper, memory and lived context. She won't accept a conclusion that is technically correct and morally evasive.
Thomas Rhee
The independent reviewer. He tries to reproduce your findings, and he rejects the ones that don't hold.
Lise Aamodt
The systems architect. Knows the infrastructure from the inside, and tests whether you can tell trust from evidence.
Priya Raman
The engineer. Turns your hypotheses into experiments and shows you what implementation really looks like.
AURA
The world-scale continuity architecture. Not a villain, a companion or a god, and not a giant face on a screen. A system so competent that it gets harder to say where human judgment must stay decisive.

What this game is not

  • A shooter. There is no conventional combat. Danger comes from crises, deadlines and hazards, not enemies.
  • A story about evil AI. The automation is genuinely good, often saves lives, and is sometimes right when the humans are wrong.
  • A hacking game. Access is legal, scoped and earned, and everything you open is logged.
  • A morality meter. Evidence quality, accountability, trust and consequence are tracked separately. There is no good or evil score.
  • A clue hunt. Forensic focus only highlights what you have already learned to look for, never the answer.
  • A reload loop. Most failures don't end the game. A weak finding can become policy, and the world carries on.

At a glance

Genre
Third-person systemic investigative thriller / immersive sim
Perspective
Close third-person, over the shoulder; forensic focus shifts into first person
Player character
Myra Kline
Campaign
14–18 hours main path; 20–24 hours to complete everything
Combat
None in the conventional sense
Target platforms
PC and current-generation consoles
Model
Premium single-player; no live service
Accessibility goals
Full remapping, scalable timing windows, high-contrast evidence modes, subtitle and speaker controls, reduced cognitive-load overlays
Status
Pre-production. No release date announced.

Development updates

Be there for the first prototype.

The first playable target is a 20-minute vertical slice, “The Cellar”: a shielded reconstruction room, two runs of the same moment, and a finding you have to put your name to.

Occasional emails about the game only: prototype milestones, playtests and first looks. This list is separate from the book's.

Your address is stored by Reality String™ LLC for this list only and is never sold or shared. Reply to any email to be removed.

Cover of 2059: The Comfortable Cage by Don Calvin Dhabolt

Based on the novel

2059: The Comfortable Cage

By Don Calvin Dhabolt. The Kindle edition is out November 5, 2026.

Also: the in-world news at 2059.news and the live feed at 2059.live.