HKUST PhD Chronicle, Week 9, Out of Control

After days and days of hard working, including overtime this weekend, I finally managed to reproduce our lab’s project in the cloud! It was a painful process, especially migrating from a machine-specific codebase to a reproducible cloud environment.

Although it was painful, I managed to gain a full understanding of all the system dependencies. As an engineer, I love having this granular control over my system. I appreciate knowing every specific detail about the dependencies and being full aware of how it all fits together.

One great result should ideally follow another, right? That was my objective, at least. I managed to get the cloud setup working just before an important demo. There was a big man coming to our lab, and I needed to demo this setup in front of a group of people. Even a photographer showed up for the event.

Guess what? The system inevitably broke precisely when its was supposed to work perfectly. I’ll always remember how awkward I felt when the system didn’t work as expected. I have to admit, I panicked. At that moment, my fast, intuitive reaction( what Kahneman calls System 1 thinking ) took over from my slower, more analytical side(System 2). Because of that, I missed a very important piece of information in the terminal output.

Remember the photographer? Well, when he walked into my workspace, he accidentally stepped on the cable connecting to my Realsense camera. I didn’t even notice the cable had become disconnected. The terminal did show some red error messages, but in my panic, I was already thinking about what else might be wrong and rushed to re-run the process, completely missing the crucial information.

Oh lord! So much for one success leading to another. This was definitely a painful lesson learned this week.

Lesson learned: Always investigate the errors thoroughly first. Rushing to ‘re-run’ or reboot relies on unexplained luck, which is no substitute for understanding the actual problem.