Project Summary

The 4:47 Decision is an original browser-based simulation in which an Evening Operations Lead must interpret an abnormal dissolved-oxygen reading in a Giant Pacific octopus habitat while the rest of the operational picture still appears normal.

The fully playable prototype was developed from authoritative husbandry, water-quality, instrumentation, and recirculating-system sources. I built it in custom HTML, CSS, and JavaScript without an authoring tool.

The Challenge

Some performance problems are not about whether someone is willing to act. They are about determining what the available evidence means.

A low dissolved-oxygen reading appears, but the main pump is running, total flow remains within its displayed band, temperature and pH are stable, and the animal shows no obvious distress. The reading is valid, but no single person, instrument, or procedure explains the situation.

The learner must construct a judgment from evidence distributed across the habitat, monitoring system, maintenance history, animal observations, and staff communications. As new information arrives, the learner must revise the working diagnosis and coordinate Animal Care, Life Support, and Guest Services before certainty is available.

How This Differs from SHIFT 17

SHIFT 17 examines whether responsible action survives operational pressure. The 4:47 Decision begins after the decision to investigate has already been made.

It assesses how a knowledgeable professional organizes an investigation, determines which evidence matters, revises a working diagnosis as conditions change, and decides when an incomplete but emerging pattern justifies intervention.

My Role

I served as the sole researcher, learning experience designer, and developer.

I conducted the technical research, created the Source Record, wrote the Learning Experience Storyboard, designed the scenario and branching architecture, developed the operations console, and built and tested the simulation.

The Solution

At 4:47 p.m., near closing, dissolved oxygen falls below the facility warning threshold at a probe positioned low in the habitat. The reading is real. What is not yet visible is why it is happening.

Three decisions follow:

• The first asks how to frame and organize the investigation.

• The second asks what the emerging pattern justifies doing while the underlying cause remains unconfirmed.

• The third asks what counts as recovery: a displayed value returning to range or the habitat itself recovering.

Both acting and waiting carry consequences. Four outcomes are possible, including one in which the console clears while the condition producing the alert remains unresolved.

What Makes the Experience Distinct

The ambiguity comes from system behavior, not withheld information

A service bypass left outside its intended operating position reduces useful circulation to the exhibit while the main pump and total-loop flow continue to appear normal. Oxygen falls first in the least-mixed portion of the habitat.

Every reassuring signal on the console is telling the truth. The learner must determine why those truthful signals do not describe the complete condition.

Two valid readings disagree by location

The fixed probe is clean, and a handheld instrument corroborates the low reading beside it. A second reading near the return remains higher.

The disagreement is spatial, not instrumental. Recognizing that distinction changes the investigation from checking whether the number is real to determining what is happening inside the habitat.

Both early strategies are professionally defensible

The first decision is not a disguised correct choice paired with a negligent one. Both approaches initiate a responsible investigation.

What changes is which evidence arrives first, when the spatial pattern becomes visible, how quickly the life-support history enters the picture, and how much operational margin remains when the cause is identified.

Consequences replace conventional feedback

The experience does not award points, badges, or Correct/Incorrect messages.

Feedback emerges through oxygen trends, readings from different habitat locations, animal observations, staff availability, guest access, escalation requirements, and the work inherited by the night team.

The aquarium continues operating, and the learner sees what the decisions produced.

Research Foundation

Technical and environmental decisions were constrained by AZA husbandry guidance, public-aquarium and university materials, dissolved-oxygen instrumentation documentation, water-quality protocols, and recirculating-system literature.

A Source Record preserves:

• what each source established;

• which design question it informed;

• how the evidence influenced the experience;

• what remained a design inference;

• what alternatives were considered or rejected;

• and why those decisions were made.

For example, a generic aquaculture threshold was considered but rejected because it could not be transferred directly to this habitat without accounting for the species, temperature, monitoring method, and facility context.

Key Contributions

• Technical research and source documentation

• Performance-problem definition

• Learning Experience Storyboard

• Scenario and branching architecture

• Decision and consequence design

• Operations-console UX

• Custom HTML, CSS, and JavaScript development

• Accessibility implementation

• State and branch testing

• Visual asset direction and integration

AI in the Project

AI assisted with source location and comparison, evidence organization, alternative generation, consistency checking, visualization, prototyping, coding, and iteration.

The design decisions remained mine: defining the performance problem, selecting the situation, determining what the learner needed to experience, choosing the failure mechanism, deciding what information arrived and when, evaluating authenticity, and determining what to accept, reject, or change.

When the first decision was initially drafted as verify-versus-defer, I recognized that it duplicated the central tension of another project. I changed the competency and rebuilt the decision so both options represented responsible investigation.

AI helped test and prototype the revision. The decision to redefine what the experience should assess remained mine.

Value

The 4:47 Decision demonstrates research-led experience design: transforming technical source material into an authentic performance situation where judgment—not recall—is the assessed capability.

The approach transfers to any environment where the data is valid, the operational picture is incomplete, and someone must act before certainty arrives.