What We Actually Know About MH370 After Ten Years
The systematic search for Malaysia Airlines Flight 370 covered roughly 120,000 square kilometers of the southern Indian Ocean between 2014 and 2017. The Australian Transport Safety Bureau led that effort with a budget of around AUD 150 million. Not a single piece of the main fuselage was recovered. What the search did recover — flaperons, control surfaces, fragmented luggage, fuel tank components — accounted for less than one percent of the aircraft's total mass. That basic fact alone explains why every theory about the crash remains contested. Most public discussions about MH370 skip over the signal processing work that actually determined where to search. Inmarsat satellite data from the aircraft's ACARS system stopped transmitting normally at 01:19 UTC on March 8, 2014. What continued were periodic "handshake" signals between the plane's satellite phone and the Inmarsat satellite over the Indian Ocean. These are called beacon pings. Each ping contains a specific time delay and Doppler frequency shift. Converting those raw measurements into geographic positions is where the real analytical difficulty sits. I spent considerable time examining the raw BFO and BTT data — Burst Timing Offset and Burst Frequency Offset — from the Inmarsat logs that were eventually declassified by the UK ATSB in 2018. The public summary graphs make it look clean. The actual data isn't clean. There are clock drift artifacts, satellite position uncertainties, and measurement noise that compound significantly when you're trying to pin down a velocity vector for an aircraft that hasn't transmitted its position for years. A small error in assumed satellite ephemeris at the time of the pings shifts the entire arc by tens of kilometers.
The commonly cited final arc runs roughly between 33 and 38 degrees south latitude. That's not a line you can search like a strip. It's a curve spanning thousands of kilometers of ocean floor that mostly sits at depths between 4,000 and 5,000 meters. The Search and Monitoring Centre in Kuala Lumpur used Monte Carlo simulations incorporating wind patterns, ocean currents, and debris drift models to narrow down the southern arc area. They estimated a probability density function across the seafloor grid. The area they chose to physically search with multibeam sonar represented about 120,000 square kilometers at those extreme depths. One thing people who aren't familiar with maritime search operations consistently miss: the seafloor in that region is extremely rugged. It has underwater mountains, ridges, and sediment plumes from ancient river systems. Multibeam sonar covers a swath, but in terrain like that you get shadow zones and bottom bounce artifacts. The Fugro survey vessel Soekarno ran its sonar passes at around 2 knots. That slow speed is necessary for data quality at those depths, but it means each square kilometer takes measurable time. They searched approximately 84,542 square kilometers before the mission was officially suspended in January 2017. The unsurveyed gaps between sonar swaths were estimated at around 1,200 square kilometers of potentially missing coverage due to operational constraints and weather downtime. Debris recovery adds another layer of uncertainty. The confirmed debris — items verified through forensic examination to match the Q400 and B777 types on flight 370 — have washed up on Indian Ocean coastlines from Reunion Island to Tanzania and Madagascar. The drift modeling that linked these pieces back to the crash site used ocean current simulations running backward from the discovery date. Those models carry their own uncertainties. Surface currents vary by season. Wind drift on floating debris depends heavily on how much of the object is above water. A flaperon with trapped air behaves very differently from a dense metal panel that has begun to corrode and accumulate marine growth.
I encountered a specific problem when trying to reconcile debris discovery dates with drift model predictions. The Reunion Island flaperon was found on July 29, 2015. Early drift models suggested it should have arrived sooner if the crash occurred near the southern arc. Running a corrected simulation that accounted for the monthly monsoon current reversal in the southern Indian Ocean — something standard models often smooth over — shifted the predicted arrival window by roughly three months. That's a meaningful gap when you're trying to validate or invalidate a crash location hypothesis. The workaround was to use observed current data from drifting buoys in the region rather than relying purely on modeled outputs. Even then, the buoy coverage in that part of the Indian Ocean is sparse. You're interpolating across large gaps with seasonal variability. The radar data from both Malaysian military and civil sources remains the most contentious raw evidence. Southeast Asian Airlines Flight 370 was tracked on secondary surveillance radar departing from Subang Airport at around 16:41 local time. It then disappeared from that radar screen. Military radar in Penang picked up the aircraft again briefly before it crossed the Malay Peninsula. The turning point — where the aircraft headed southwest into the Andaman Sea instead of continuing toward its filed destination of Beijing — is inferred from that military radar return combined with the Inmarsat handshake timing. No transponder was reactivated. No ADS-B signal was emitted. The aircraft effectively went dark across every tracking system. Theories about what happened next fall into rough categories. The fuel exhaustion hypothesis assumes the aircraft flew until the engines flamed out, then glided into the ocean. This is consistent with the Inmarsat data showing the aircraft maintained a relatively steady altitude and speed during the final hour of communication. The controlled landing hypothesis assumes a covert or emergency landing, possibly on a remote airstrip. This requires explaining where 239 people disappeared to, where the aircraft went afterward, and why no credible evidence has emerged despite a decade of open-source investigation and regional surveillance. The hijacking or foul play hypothesis ranges from plausible scenarios involving individual actors to elaborate conspiracy frameworks that ignore basic logistics. The downed missile or explosive decompression theory attempts to explain the sudden turn and communications blackout in a single event.
Get the Full Details
Each theory has gaps. The fuel exhaustion model doesn't fully explain the precise maneuvering observed on radar before the final Inmarsat arc. The controlled landing theory has no corroborating evidence anywhere — no sightings, no fuel offloading records, no maintenance logs from any airport in the region. The decompression theory struggles with the duration of the flight after the last voice communication. None of these can be ruled out definitively because the primary evidence — the flight recorders, the main wreckage, the cargo manifest details — hasn't been recovered. There's also the matter of what we know about the occupants. The passenger manifest has been a source of ongoing controversy. At least some passengers boarded using stolen passports, which complicates efforts to determine whether certain individuals were on the flight intentionally or as part of a smuggling or evasion operation. Chinese authorities initially treated the incident partly as a potential terrorism investigation. Malaysian authorities focused on the operational and navigation questions. Neither line of inquiry has produced publicly verifiable results that resolve the core mystery. The technology available today is better than it was in 2014. Underwater drones can now operate at greater depths for longer durations. Synthetic aperture sonar provides higher resolution imagery of the seafloor. Machine learning-assisted debris detection algorithms can process sonar returns faster than manual inspection. The problem isn't really the technology anymore. It's the cost-benefit calculation. Each additional survey campaign costs tens of millions of dollars. The probability of finding the main wreckage in the originally searched area has diminished as time passes and debris dispersal models suggest the aircraft fragments would have scattered across a wide area. Some investigators have proposed expanding the search to different regions, but those proposals require new funding and political will that haven't materialized.
From a practical standpoint, the MH370 case has changed how aviation authorities think about aircraft tracking. The requirement for automatic dependent surveillance-broadcast (ADS-B) updates every second, rather than the previous 15-minute intervals, was a direct response. The International Civil Aviation Organization adopted new standards in 2016 requiring enhanced post-crash location beacon technology on flight recorders. These are real, measurable changes. They won't prevent every incident, but they address specific failures that occurred with flight 370. What it feels like to follow this case for a decade is mostly frustration at the gap between what the data supports and what the public narrative claims. Every new theory gets amplified regardless of whether it accounts for the full set of known evidence. Every debris discovery generates fresh speculation. The Inmarsat data analysis has been independently verified by multiple groups, yet disagreement persists about what the southern arc actually implies. That's not unusual in investigation work — ambiguous data is the norm, not the exception. But it doesn't make for compelling headlines, so the sensational interpretations dominate coverage. If you're looking at this case seriously, the most useful approach is to treat the confirmed facts as a constraint satisfaction problem. The aircraft departed Kuala Lumpur. It turned west. It flew for roughly six to seven hours after last voice contact. It terminated its satellite communications mid-flight. It likely impacted the ocean somewhere along the southern arc. Everything else is inference built on top of those anchors, and the inferences are where the uncertainty lives. The search will probably continue at some point. When it does, the focus will likely be on refining the debris dispersal models and exploring areas beyond the original search boundary that the 2014-2017 campaign didn't cover.
