We Built a Spectra Game: Trace the Anomaly Before the Comms Window Closes

Reading a dashboard tutorial is fine. Hunting a live telemetry anomaly while a comms window closes — clicking one chart and watching every other chart snap to your selection — that sticks.

So we built it. Observation Window is a free, five-level browser game that teaches cross-filtering by handing you a problem you actually want to solve: the MARS satellite constellation started throwing errors right after a firmware push, you have one comms window to find the cause, and the only way through is to drill the charts.

Play Observation Window → No sign-up, no install, runs in your browser.

The Setup

A new firmware build just deployed to the constellation, and the observation-deck dashboard lit up red. Greg from QA — last seen saving a rocket launch in Launch Window — is back, now in mission ops, and he is not calm about it. Four satellites are up there. One is faulting. The clock does not stop.

In front of you is a live dashboard: events by satellite, events over time, a subsystem ring, a severity breakdown. Every chart is clickable, and clicking one refilters all the others — the same linked brushing that powers Spectra. Your job is to narrow thirty-six telemetry events down to the single root-cause signal before the window closes.

What You Learn (in Five Levels)

1. Which craft is faulting? The per-satellite bar chart has one tower. Click it, and every other chart cross-filters to that craft — your first taste of linked brushing.
2. When did it start? The timeline has a dashed marker where the firmware deployed. Click the spike that lands right after it — correlating a change with its fallout.
3. Which subsystem? A donut splits that day's spike by subsystem. One slice owns most of it. Click it to drill again.
4. Signal or noise? Info and warnings are chatter. Isolate the critical band — the events that actually threaten the mission.
5. Find the root anomaly. Three critical signals remain, and by event count they are identical — flat, useless triplets. Native dashboards stop here. You toggle the readout to time-in-status, and one signal breaks the pack: it has been degraded for eight straight days, since before the deploy. The push didn't cause it; it exposed it. Signal locked.

Why Charts?

Cross-filtering is one of those things you can't feel from a screenshot. Reading “click any chart and the whole dashboard refilters” is not the same as clicking the Ares bar and watching the timeline, the subsystem ring, and the severity chart all snap to Ares in the same instant. The game makes you do it, five times, until it's muscle memory.

And the last level is the point of the whole thing. The first four narrow the field with counts. The fifth one can't be solved by counting — the culprit is invisible until you group by a derived metric Jira doesn't even store. That gap is exactly why Spectra exists: the questions that matter (“how long has this actually been stuck?”, “where is work really aging?”) live in cycle time, time-in-status, and WIP age — metrics that aren't Jira fields. The game hands you a problem only a derived metric can crack.

Spectra is live on the Atlassian Marketplace — the real thing does this to your actual Jira data, no query language required.

For Teams

If you onboard delivery leads, scrum masters, or anyone who lives in Jira reports, the game is a five-minute exercise that gets them past “dashboards are static pictures” and into “I can interrogate this.” Drop the link in your onboarding doc. It runs in any browser, on any device, with no setup.

And if you already know your way around a control chart — play it anyway. The last level is a good reminder that the number you're counting is rarely the number that matters.

Play Observation Window →