Thursday, March 18, 2010

New Evidence for Faulty Toyota Logic

Toyota today sent a four-page letter to ABC News today asking that the network retract its previous story that theorized that faulty electronics were at fault in the unintended acceleration of Toyota and Lexus vehicles. Toyota is also asking for an apology from ABC News. For Toyota, the timing of its request could not be worse. The letter was sent on the same day that new evidence strongly suggests that there is something amiss in Toyota control logic after all.

A new report today from the National Highway Traffic Safety Administration offered this finding from a Toyota Prius that crashed into a stone wall in Harrison, New Jersey: “Information retrieved from the vehicle’s onboard computer systems indicated there was no application of the brakes and the throttle was fully open.”

To Toyota executives, this is proof that the crash was the result of driver error. But this is faulty logic on Toyota’s part. A simpler explanation for what happened in Harrison, an explanation that is also more consistent with other recent experience with Toyota controls, is that the automobile spontaneously stopped reading the controls of the car moments after it was turned on.

From everything we know, the driver in Harrison applied the accelerator, then the brake. The car’s data recorder says that only the accelerator was applied. This is most easily explained if the car stopped receiving valid data from the controls during the time when the accelerator was applied.

It is as if a computer document is suddenly filled with thousands of letter l’s. Did the user decide to press the L key for several minutes, or did the computer keyboard break in the middle of a word? Either is possible, but the hardware failure is more likely. Computer keyboards, if heavily used, tend to fail after just a few years. Most of us have experienced this. The pedals in a modern automobile are digital controls that are not so different from a computer keyboard or mouse, and if there is a design problem with them, they could fail in the same way. After today’s reports, I now believe this is what is happening with the Toyota pedals.

The logic involved in this question is familiar to any computer programmer or recording engineer (or a modern philosopher, for that matter). If box B is receiving bad data from box A, is it because A is sending data incorrectly, or because B is receiving data incorrectly? The logical answer is that it cannot be determined as long as A and B are considered together. They have to be examined separately to learn anything about the nature of the failure (and the two hypotheses I suggested are by no means the only possible explanations). Based on its statements today, this appears to be logic that Toyota has not yet considered.