If they don't have anything to hide, why not allow access to the logs? I highly doubt the 0-100 claim. Raw sensor data of the accelerator is likely to be sample

If they don't have anything to hide, why not allow access to the logs? I highly doubt the 0-100 claim. Raw sensor data of the accelerator is likely to be sampled at a high enough rate that should make something like that impossible. In fact, the car should have at minimum a low pass filter and reject sudden accelerations like that, which would indicate sensor or signal failure. If it's a digital sensor, and sampled high enough, even if someone did "floor it", you'd still be able to see a gradual increase, just at a faster rate. When a sensor all of a sudden reports a huge, non-linear jump, it generally means there is some sort of failure. And if that's within normal operating conditions, the sample rate of the sensor isn't enough to tell the difference between failures and intentional (real) readings, and reject a failed sensor. And that's a problem too and 101 in any sensors and signals implementation. If it's a digital sensor on something as important as braking and acceleration, there had better be redundant sensors, a mechanism to fuse the data together, and reject a bad sensor from the state estimate of the pedal. Either way, just based on the statements they made anyone who has worked with anything robotics or autonomous can tell you that first instinct would be sensor failure. Of course, all of this is impossible to tell without the raw log data. Companies who are unwilling to provide such data or make it difficult to do so generally have something to hide in either their implementation, data quality, or the actual findings from the logs themselves.