I'm not a researcher. I'm just a developer who got a bit too interested one Saturday with this case and wanted to understand why the plane still hasn't been found after three major search operations.
So I built a tool to explore the public data myself.
The problem with the official analysis
The search areas were defined by models with hardcoded assumptions — a specific cruise speed, fuel burn rate, interpretation of the satellite data. Change any of those and the probable crash location shifts. No public tool existed that let you actually see that sensitivity.
This tool does. Every assumption is a slider.
What it shows
MH370 pinged an Inmarsat satellite seven times after disappearing. Those pings define distance rings — arcs of possible positions. The tool renders those arcs, samples thousands of candidate flight paths through them, and produces a probability distribution along the final arc.
It also shows the actual sonar coverage from Geoscience Australia — what was searched, what wasn't, and where high-probability zones overlap with coverage gaps.
The thing that stuck with me
The 6th and 7th arcs — recorded 8.5 minutes apart — have identical BTO values. At cruise speed the plane should have moved 67 nautical miles. The distance from the satellite shouldn't have stayed the same.
This anomaly means the plane either slowed dramatically, or changed heading to fly nearly tangential to the satellite. Each scenario puts the impact in a different place. The tool shows both families side by side.
Try it
Web version: notlimey.github.io/mh370-analysis-tool
Full desktop app with adjustable parameters: github.com/notlimey/mh370-analysis-tool
Built with significant AI assistance. Physics from ATSB reports and the Independent Group researchers. If something is wrong it's an implementation error on my part — open an issue.