Features - Technical
JULY 9, 2001
The anoraks/geeks shall inherit the earth
BY PETER WRIGHT
Many, but not all the cars that have failed to leave either the dummy grid or the start line in the GP's since Spain have been the victims of software failures, and nothing contributes as much to losing a race as not starting it. McLaren, for some fascinating but unknown reason, has been a particular victim of this malady, and there may be some clues as to the reasons in the comments of its senior personnel after such events.
Software writing and development is culturally different from the other engineering disciplines applicable to motor racing. Tracy Kidder wrote a cult book titled The Soul of a New Machine, back in the 1960s, about the dynamics of and the relationships between the various individuals creating a new computer for Digital (this was back in the days when computers did not fit on desktops, let alone in briefcases). This surprisingly gripping book provides a fascinating, and still relevant insight into the personalities of people who are good at this type of work. They do not work in the same way as mechanical engineers and tend to use a vocabulary that is incomprehensible to outsiders. As a result they are difficult to supervise and manage, and a great deal of faith has to be placed on the integrity of the code they produce. Even with specialist software integrity checks, as developed by the aerospace and nuclear industries and applied by the FIA to Formula 1 software checking, error paths still survive undetected - using Windows illustrates the point every so often. Only that which can be imagined can be checked for, and in motor racing there is little time to imagine.
On top of these potential failures, problems occur when humans are brought into contact with automation, such as traction and launch control. Drivers develop a mental model of the car's handling, and expect it to behave in a predictable manner in order to be able to drive it at the limit. If the car's characteristics change e.g. when software takes over some function, or it rains, they have to re-learn and generate a new mental model before being able to operate it again at its peak performance. Some drivers are better at it than others, and can easily be identified by the speed at which they adapt to changing track conditions - Ayrton Senna and Michael Schumacher spring immediately to mind. When Richard Noble set out to select a driver to attempt the Land Speed Record in Thrust SS, he involved the aviation-medical people at DERA to devise a test for adaptability to changing control characteristics. Andy Green (a fighter tactics development test pilot with the RAF) excelled in this test and his performance was a major influence in his selection to drive the car. His ability to adapt to driving Thrust, with its rear-wheel steer and nasty transonic characteristics, contributed enormously to the success of the project.
A significant number of aircraft accidents have been caused by the pilots becoming confused as to which mode of automation the aircraft is operating in as it transitions from one flight regime to another. Adrian Newey's kind comments about Mika Hakkinen stalling his McLaren at the start in Austria, blaming it on the launch control software, hint at Mika being confused by "mode transition".
Perhaps the biggest problem of all is that many of the senior members of the teams - Team Principles, Technical Directors and Chief Designers - do not fully comprehend the technology. Most have at some time designed, built, maintained or driven racing cars themselves, but hardly any have written computer code. Indeed I know of more than one who hardly uses computers, relying instead on the expensive experts they have hired. While they can look at a suspension system or gearbox internals and judge the integrity of the design, they are almost blind to what is going on inside the microprocessors that control so much of a modern GP car's performance and reliability. They cannot check and approve all the failure modes personally, and are often at a loss to comment when something goes wrong in the full glare of TV publicity.
I wonder if Ferrari's stated explanation for vetoing the introduction of traction and launch control at the beginning of the season, vis. that they would not be ready in time, holds the key to their dominance at the moment. It may be that they truly understood the enormity of the task of introducing these systems and training their drivers to use them, and were not prepared to be pushed into using them too quickly. There has certainly been less "brain fade" at Ferrari than elsewhere.