3 ms·

> Eg: One team I was an auditor for, didn't realize that the instrument they were using had effectively two (2x) ON/OFF switches - and the detector was turned o

> Eg: One team I was an auditor for, didn't realize that the instrument they were using had effectively two (2x) ON/OFF switches - and the detector was turned off. 2 years of data had to be rejected. How is this kind of negligence legal?

wirrbel 7y ago
One code base I inherited at a job had a unit test suit with no assert statements. No one verified that the tests would fail with a faulty implementation. It's the same kind of error. In the lab no one verified that detector would be able to detect. As to forgetting switches, on lab equipment, especially custom built equipment it's sometimes not obvious, which switch is which. Also as a student I worked in a lab where many experiments where built by one or two generations of students and then continued by others so those producing data. Working with an existing setup is like inheriting a code base. You never know what surprises await you.
- flowersjeff 7y ago
  The reason why the code failed to realize there was even an issue?...(in the case I mentioned earlier) The software looked at the calibration data using linear best fit (standard stuff) and provided the r^2 was within the specified range - that instrument and the data was good to go. Well, if you run your Standards{1,2,3,4,5} and all output "0.0". That is a straight (perfectly so) line. They had no idea how to read the computer monitor and the real time data coming off the instrument. They had zero real training / understanding of even how the instrument worked. I see this a lot still - people using an instrument without a great understanding of how the thing even works. Why? There's just no money/resources being spent to ensure correctness - self regulation is (and has been) all the rage for decades here in the US. Honestly, I believe in open software - and this type of software NEEDS to be opened up. I can only imagine the horrors that will be found.
- flowersjeff 7y ago
  PS - regarding the actual switch. Agreed. This instrument was of a poor UI. One switch was in the front, the other in the back. Still, this was a regulated contract laboratory with cert's. Upon discovery - they did perform corrective actions and such. As for "in school" - oh yeah. It's a wild wild west sort of situation. Custom hardware (with who knows what thought being given) and software that someone wrote years ago (and having zero documentation and zero standard coding practices observed).

wirrbel 7y ago

One code base I inherited at a job had a unit test suit with no assert statements. No one verified that the tests would fail with a faulty implementation. It's the same kind of error. In the lab no one verified that detector would be able to detect. As to forgetting switches, on lab equipment, especially custom built equipment it's sometimes not obvious, which switch is which. Also as a student I worked in a lab where many experiments where built by one or two generations of students and then continued by others so those producing data. Working with an existing setup is like inheriting a code base. You never know what surprises await you.

flowersjeff 7y ago
The reason why the code failed to realize there was even an issue?...(in the case I mentioned earlier) The software looked at the calibration data using linear best fit (standard stuff) and provided the r^2 was within the specified range - that instrument and the data was good to go. Well, if you run your Standards{1,2,3,4,5} and all output "0.0". That is a straight (perfectly so) line. They had no idea how to read the computer monitor and the real time data coming off the instrument. They had zero real training / understanding of even how the instrument worked. I see this a lot still - people using an instrument without a great understanding of how the thing even works. Why? There's just no money/resources being spent to ensure correctness - self regulation is (and has been) all the rage for decades here in the US. Honestly, I believe in open software - and this type of software NEEDS to be opened up. I can only imagine the horrors that will be found.
flowersjeff 7y ago
PS - regarding the actual switch. Agreed. This instrument was of a poor UI. One switch was in the front, the other in the back. Still, this was a regulated contract laboratory with cert's. Upon discovery - they did perform corrective actions and such. As for "in school" - oh yeah. It's a wild wild west sort of situation. Custom hardware (with who knows what thought being given) and software that someone wrote years ago (and having zero documentation and zero standard coding practices observed).