An influential US report says that defective software could contribute to, or cause, a major air accident but leave no physical evidence.
It also says that investigators could have their efforts undermined by manufacturers or suppliers who may seek to "disrupt or bias the investigation" rather than face the blame for a major aviation accident.
The report questions whether air crash investigators have the skill, facilities or the independence from suppliers to establish the truth in accidents that involve high levels of complexity and "system-level inter-relationships".
Researched at the request of Jim Hall, chairman of the USNational Transportation Safety Board, the report seeks to provide a critical examination of the organisation's ability to investigate major commercial aviation accidents.
The board is the largest accident investigation organisation in the world, with nearly 500 staff and a budget last year of $56m (£35m). Many of the report's findings will apply to the UK and other air accident organisations around the world.
"As complexity grows hidden design or equipment defects are problems of increasing concern," says the report. With complex systems, "failure states can be 'reactive' leaving no permanent record to discover in the wreckage".
To establish the truth, investigators have to rely on the crashed aircraft's software suppliers and manufacturers who understand the complex system inter-dependencies. But this can lead to companies seeking to exonerate their systems from blame.
The report says it found evidence that representatives of companies involved with the investigation "may be acting to further various interests beyond prevention of a similar accident".
This, adds the report, raises "numerous questions about the extent to which representatives are motivated to influence the outcome of the safety-related investigation in anticipation of litigation". It says that investigators trying to assert their independence may be "no match" for a large corporation.
The report's findings are supported by other studies which show that, although software is suspected as a contributory cause in a number of air disasters, it has never been proved to have been a main cause of any major fatal air accident.
Some specialists are concerned that if manufacturers or their contractors cannot be held liable for negligence after a major accident owing to the difficulty of establishing conclusively whether software was to blame, then firm could, with impunity, cut corners on design, programming and testing of safety-critical software.
This was first published in February 2000