Having seen how he tracked a couple of my own flights I can see some value.
In both cases he was given a departure point (outside of Australia, outside of both radar and ADSB), an initial heading, but not an arrival port. In both cases he was able to accurately to show not just where we landed (two very different ports) but also our arrival time with an accuracy of a couple of minutes. His enroute ETAs were all within a couple of minutes of ours also.
It’s better than throwing darts in a map and seems to be much better than a ouija board.
Tracked? If you're referring to his "blind test" using QF6036 back in June, you might mean "guessed at". In his published paper on the first two hours of that exercise he nominated nine "position indicators" that were meant to be accurate determinations of the target aircraft's position using his "technology".
There is limited ADS-B data for that flight that shows that of the seven position indicators that there's good data for none of his fixes came within 30 km of the target aircraft's position with the error growing to nearly 70 km toward the end of the first two hours. The smallest error recorded exceeded his sampling rate of 2 minutes; that's not good.
He was basically making educated guesses of the aircraft's location based on the departure time, known airways, known aircraft performance (eg actual fuel load, climb and cruise speeds) and weather courtesy of Nullschool.
If you look at the track that he has plotted for MH370 he has the aircraft doing a coast hugging tour of north-western Sumatra before entering a 22 minute one lap racetrack south of Indonesia all flown at cruise speed. Does that sound even vaguely likely given what we already know?
