Talking about impostor syndrome on the GR-OSS OUT Podcast

When my good friend Jay Faulkner asked me to hop on the GR-OSS OUT Podcast to talk about imposter syndrome, I jumped at the chance. I haven’t talked much about it since 2015 when I spoke at Fedora Flock in Rochester, but it’s been on my mind lately, especially with the advances in AI and automation.

Thanks for the opportunity, Jay! 👏

If you’re in a hurry, watch the full episode below or subscribe to the audio version. Otherwise, keep reading for a brief overview of what we discussed.

The OODA Loop#

Fighter pilots use the OODA loop (observe, orient, decide, act) to react quickly in combat. I borrowed this as a mental model for navigating those challenging moments that impostor syndrome creates. This really helps when your assessment of yourself isn’t matching up with the assessment from the people around you.

• Start by observing how other people react to your ideas, suggestions and actions.
• Get feedback often and orient yourself to the reactions of the people around you.
• Decide on how you’re going to take that new knowledge and apply it to your next move.
• Act on that decision and go through the loop again afterwards.

Impostor syndrome often feels like a whirlwind of uncertainty and doubt. Using the OODA loop keeps me focused and cleans up some of the confusion.

Signs of growth#

One point I keep coming back to: imposter syndrome tends to resurface every time you join a new team or stretch into a bigger role. That’s not a signal that you’re in over your head. It’s a signal that you’re growing.

If you don’t feel this unsettling feeling from time to time, you might not be in a role that isn’t challenging enough. That’s a great starting point for a conversation with your manager or someone more senior to you on the team. Seize the opportunity to take on a broader role with your team where you expand your thinking and take on larger, more complex issues.

“Fake it ’til you make it” isn’t a bad thing. 😉

AI brings change#

LLMs and AI coding tools create a whirlwind of feelings in multiple directions.

These tools empower novices to explore codebases and better understand the code and the reasoning behind it. In the past, someone would need to sit down with a more senior developer and ask lots of questions. However, no developer knows everything about a codebase that is maintained by a large team. Now you can open your favorite AI coding tool and ask “Is this variable always a number?” or “Why is there so much complexity in this SQL query?”

While a human might avoid asking something they consider to be a “dumb question” to another human, this apprehension disappears when you can ask an LLM instead. I’ll admit that LLMs might give an incorrect or incomplete answer. However, novices will come away with the right knowledge and understanding to ask better questions of their peers.

On the other hand, these tools can lead into dangerous Dunning-Kruger territory. This happens when people with a relatively low ability in a specific area begin thinking that their actual abilities are much more advanced.¹ There’s no hallucination-free AI coding tool. This easily leads to performance issues, security vulnerabilities, and other bugs without a thorough review.