UPDATED 10 FEB 08: "Investigators believe the crash-landing ... may have been caused by ice clogging its fuel supply, according to a leaked memo to US regulators." 
LOOKING AT THE WRECKAGE of a Boeing 777 at Heathrow, I see the photos through the eyes of an flight safety officer. I held that job title several times in the 1980s and along with cleaning up several fatalities, I also saw some things happen to my €20m four-engined jet aircraft that made me flush with momentary panic. First, fuel levels are never what they seem. This should not happen in a glass cockpit but I have experienced issues with indicators and with the inner workings of computers that exist to optimise fuel usage. Our indicators would not show accurate results when within 10% of the gauge's rated capacity. On top of that, if we encountered minor electrical problems, we could end up with fuel that we could not remove from the largest tanks. At one island destination, I quickly shut down the inboard engines because I knew we were running on fumes from the main tanks. One of the extended range tanks had not pumped around 5,000 gallons of fuel into the main tanks. We fixed the problem after landing but it was eye-opening to know we did not have enough fuel to go-around and reattempt our landing approach. When training highly experienced aircrews on a new Fuel Savings Advisory System, I watched when two of the systems simply failed and then started reducing the fuel flow to zero. This is not a good thing when airborne. When I heard about BA 38's sudden and uncommanded drop in power, I vividly recalled what I saw aboard a highly complex aircraft more than 20 years ago. Certainly, the systems have evolved but most certainly, digital gremlins can run around inside any kind of on-board avionics technology.
The second event occurred while permitting the aircraft to land itself in order to maintain proficiency. This was a normal procedure for pilots who needed to renew their landing currency with an on-board all-weather landing system. Both the aircrew and the aircraft can accomplish an automatic approach in clear weather without creating a hazard to flight. When doing the approach, a lot depends on the accuracy of the signals coming from the airfield. So when a large truck stops in front of the main instrument landing system antenna array, the results inside the cockpit can be confusing. In one instance, the aircraft thought it was closer to touchdown than it actually was. Just like in the BA 038 incident, our throttles reduced themselves and the aircraft started settling for landing. It is too early to tell if either of these scenarios relate to the short landing at Heathrow but most accident investigators would try to discount both of them as part of their initial analysis of the accident.
What seems very intriguing is the fact that the autopilot wanted thrust but the engines started winding down on their own. From the initial report of the accident investigation board:
Initial indications from the interviews and flight recorder analyses show the flight and approach to have progressed normally until the aircraft was established on late finals for Runway 27L.
At approximately 600ft and two miles from touch down, the autothrottle demanded an increase in thrust from the two engines, but the engines did not respond.
Following further demands for increased thrust from the autothrottle, and subsequently the flight crew moving the throttle levers, the engines similarly failed to respond.
The aircraft speed reduced and the aircraft descended onto the grass short of the paved runway surface.
The investigation is now focussed on more detailed analysis of the flight recorder information, collecting further recorded information from various system modules and examining the range of aircraft systems that could influence engine operation.
Graphic and accident investigation quotation from BBC News.
1. Steven Swinford -- "Ice in fuel theory in Heathrow jet crash" in The Sunday Times, 10 February 2008.