Human Skills · Discussion

Level of Expertise Dialogue

Definition

Writing can be polished, rehearsed, even generated. An unscripted, in-person dialogue about your own work is much harder to fake — which makes it one of the most honest tests of whether you actually understand what you claim to. This isn’t an interview or a quiz; it’s a two-person conversation, without notes, where a small set of open questions guide but don’t constrain the discussion.

Learning Outcome

After using this technique, you should be able to speak about your own research or project — its substance, its methods, and the people whose work informs it — clearly and specifically, without relying on notes or a prepared script.

Core Structure

Five questions guide the dialogue, meant as prompts rather than a rigid script:

  1. The one-minute pitch. If you had a minute to describe what your work is and why someone should care about it, what would you say?
  2. Algorithms. What algorithms were developed and used as part of your work?
  3. Data structures. Which data structures were developed and used?
  4. Architecture and design. What is the architecture and design of the software you developed, used, or both?
  5. Key players. Who are the key people in your area — individuals, teams, or both? Which articles, lectures, or videos mattered most to your own learning, and who would you point someone else to?

The format matters as much as the questions: it’s done in person, as a genuine two-person dialogue, without extensive notes. That’s what makes it a test of understanding rather than a test of recall or preparation — a rehearsed answer to question 1 falls apart the moment a follow-up question pushes past the surface.

Worked Example

A dialogue on a project about caching strategies might open with the one-minute pitch: “I looked at how write-through caching with tiered eviction affects tail latency under different workloads.” A good follow-up push might be “why tiered rather than a single eviction policy?” — and a real answer draws on the architecture question (4) to explain the actual design decision, rather than repeating the pitch in different words. The key-players question (5) might surface that a specific conference paper on adaptive replacement caching shaped the whole approach — the kind of detail that’s easy to lose in a written report but comes up naturally in conversation when someone asks “what got you started on this?”

Common Pitfalls

  • Preparing and memorizing an answer to question 1 as if it were a script — a genuine dialogue will move past a rehearsed pitch within the first follow-up question.
  • Being able to describe what was built (architecture, question 4) but not why choices were made — the dialogue format is meant to surface reasoning, not just facts.
  • Treating question 5 as a formality — being unable to name any actual person or source that shaped your understanding is a sign the work wasn’t grounded in the field’s actual conversation.
  • Reaching for notes when a question gets hard — the entire value of the format is testing what you can say without them.

Rubric / Checklist

  • Can give a clear, specific one-minute pitch without a memorized script
  • Can name and explain the algorithms actually used, not just gesture at them
  • Can name and explain the data structures actually used
  • Can explain the architecture and design, including why choices were made, not just what was built
  • Can name real people, teams, or specific sources that shaped the work — and could point someone else toward the best of them
  • Held the conversation without relying on notes