See Through Their Eyes.

Sometimes, you will notice that it is with such ease to follow someone’s thinking process while they are talking, whereas you may encounter a tremendous number of clarification questions when you are leading the project meeting, about things you thought you had clearly articulated. The divergence is even greater if you have had interdisciplinary collaborations. Some domain experts can always make it easy for you to understand what they are trying to convey and clearly present their opinions and rationales to you, even if you know nothing about their territory. Some others, on the other hand, trigger confusion with every sentence they speak: “Wait, what does that term mean?” “Am I understanding the concept the same way they are using it?” “Am I the only person not understanding their words? Should I ask the (stupid) question?”

If I had to name the single skill that most junior researchers underestimate, it would not be the glamorous kinds, like writing an award-winning paper or giving an insightful keynote, but the everyday kind: communication. I am not just talking about verbal communication, but about a broader landscape, including explaining an idea to your advisor, describing a problem to collaborators (and sometimes fighting with them), or writing an introduction that a reviewer can follow without guessing or challenging your key motivation.

Most students believe they communicate well, but, unfortunately, most are wrong. The gap between what they think they conveyed and what the other person actually understood is where the majority of research collaboration problems begin.

Why the Gap Exists

In everyday conversation, an enormous amount of context is carried implicitly. When you talk to a close friend, you share years of common experience, and your friend can fill in the gaps you leave because they already know your background, your preferences, and what you have been dealing with lately. You can say something vaguely, or even just a couple of words, and your friend understands exactly what you mean. Nonverbal cues (facial expressions, tone, pauses) also handle a surprising amount of the meaning. Many times when you and your best friend make eye contact, you both know what the other is thinking. These norms are so pervasive that they largely, if not entirely, shape how you interact socially.

When you are working with someone else on a research project, things are fundamentally different. The topic you are all exploring could be new to each one of you. A paper you read last week may not have been read by anyone else on your team. The iterations of a research idea in your mind are entirely inaccessible to your collaborator. You are unlikely to work with the exact same group of people on two projects, so the shared understanding that makes everyday conversation effortless rarely exists in research contexts.

The problem is that most students are not aware they are relying on implicit shared knowledge until someone misunderstands them, and even then, the instinct is to blame the listener (“I had explained it clearly!”), which, as it often turns out, they had not.

The solution to this challenge is easier to say than to do. But the challenging part is that you need to be aware of the following things every single time you talk and write.

Skill 1: Provide Sufficient Context

We work on CSCW research; thus, you should be very familiar with the fact that the most important prerequisite for effective remote collaboration is grounding the context. You should always be aware that when you are articulating a specific viewpoint, your rationales and justifications are not coming from nowhere. They are built on top of the knowledge you have acquired through readings, through conversations, through the critical thinking skills you have built through iterative discussions and critiques, and through what you have learned from your own prior experiences. Many of your rationales also come from specific papers or evidence that you have read. All of these are exclusive to your own experiences, and it is not reasonable to assume that your collaborators share the same context with you.

In fact, your audience grew up in a totally different background and had different experiences. They were trained differently, had gone through different research projects, and were advised by different mentors, all of which shape every person differently even when they read and think about the same topic. We do appreciate that diversity in thinking, but when we want to communicate effectively, we need to make sure the diversity in our backgrounds and professional training is addressed by careful and sufficient grounding in the context needed to ensure everyone is on the same page.

What you should do, subconsciously, before you speak in a meeting, write an email, draft a paragraph, or respond to a question is to ask yourself several questions: what does the person know based on my understanding of them and our prior conversations? What information is being conveyed in the next sentence I want to say, both explicitly through the words and implicitly through the rationales behind them? And does the person listening have enough information to understand my point of view as well as my rationales accurately?

If the answer to the above questions is no, then it is the indicator telling you that you need to add more context before you deliver the point.

Skill 2: Think from the Audience’s Perspective

The first skill is already very challenging to practice and make a habit in your day-to-day life, because you have never needed this type of thinking before when talking to your friends, family members, and people you are familiar with. The fundamental issue is that you never needed to think from the audience’s perspective and to evaluate whether you were articulating your thoughts carefully enough.

I believe this scenario will not be unfamiliar to you: you say “I think we should try approach X,” and your collaborator asks “why?” and you are surprised by the question, because in your own mind the reasoning was obvious, or you just think it is natural to derive the approach you proposed. It was obvious because you spent the whole week reading papers that led you to this specific approach, but your collaborator has not read those papers and knows nothing about your rationale. You will see how often this kind of situation happens when you try to propose something to your collaborators. They could ask a question about every single decision you make, and commonly, you do not have the answer ready, because you have never thought from the audience’s perspective when you were framing the rationales.

Even if you provide sufficient context and believe you did so, you will still encounter clarification questions or “why” questions from your audience indicating they do not really understand what you are expressing. And often, after a very long discussion or debate, it turns out your audience does not understand your words as you intended.

What happened? Even when you can accurately express the knowledge in your mind, it does not guarantee the understanding you expect, because the words you choose often carry meanings or associations that you did not account for and are not aware of. In your daily casual chat, people are fine with having vague interpretations, even slightly derailed ones, as long as the vibes are aligned. You may use adjectives unconventionally and even make mistakes when referring to a particular concept, but that is all fine because your audience is enjoying the vibe while listening to you, unlike in serious business, like research.

In research, a term that is unambiguous in your reading context may have entirely different meanings in your audience’s area of expertise. The word you choose without careful consideration could carry meanings or associations that you did not account for, you are not aware of, and that are completely different from what you intended to convey.

Imagine you are trying to sell the idea of an “LLM agent,” and you are standing in front of a medical expert who knows nothing about AI technologies, or a finance professional, or a real estate consultant. It is easy to predict that each of your audiences possesses a completely different understanding of the word “agent.” If you do not provide a precise description and definition that avoids any confusing or ambiguous terms, your audience will misinterpret your message.

Reflection on Your Writings

Like it or not, paper writing eventually becomes the testbed of assessing how well or poorly you have mastered this skill. If you remember Ch. 2 Ep. 5 (The Craft of Prose), I define the ultimate goal of academic writing as:

Academic paper writing requires that the paper be uniformly interpreted by diverse readers (i.e., readers with different backgrounds and years of experience) in the way the authors intend.

If you read it again today, I would assume you now have a more thorough understanding of why I said so and, without further ado, you should revisit each piece of advice I provided in that episode and try to, from the readers’ perspective, understand what benefits the advice provides to the readers or what problems it addresses.

Frequently, I notice students can verbally walk me through the core rationale reasonably and logically: “sentence A establishes the problem, sentence B identifies the gap, and sentence C proposes our approach.” However, when they present the Introduction draft to me, the logical relationships between sentences are not written down and become invisible; in particular, the conjunctions are missing. When I read the same paragraph as someone encountering the argument for the first time, I can honestly perceive the sentences in a completely different direction from what the student intended, and the students are always surprised to realize that their narratives are riddled with a thousand gaping wounds. The missing conjunctions are exactly why I emphasize explicit connectives so strongly: “because,” “nevertheless,” “as a result,” “which means.” These words feel redundant to the writer who already knows the logic, but they are essential to the reader who does not.

If one day you start to master this skill and can identify the issues in others’ narratives instinctively, you will be surprised to see how many people cannot do so even after they finish their PhD program and seem successful in their career.

The Asymmetry You Should Never Forget

One thing I want to make explicit, because it is the strongest practical argument for everything in this episode: the effort to provide context is tiny compared to the cost of not providing it.

Thirty seconds of thought before you speak, checking whether your audience has what they need to follow you, is all it takes. Not providing those thirty seconds costs the entire room twenty minutes of confusion, clarification questions, “let me re-explain,” and “go back and think about it more.” I have watched this happen in meetings so many times that I can predict the moment a student is about to trigger a twenty-minute detour: it is the moment they jump straight to their conclusion without laying down the path that got them there. The math always favors preparation, and the ratio only gets worse as your audience gets larger (a conference talk, a paper with hundreds of readers, a grant proposal read by a panel).

The Same Skill, Amplified by AI

Everything in this episode applies to communicating with humans. In the age of AI, the same principles become even more critical, because an AI system has zero implicit shared context with you.

When you talk to a friend, they can infer what you mean from years of shared experience. When you talk to a collaborator, they can at least draw on their own expertise to fill some gaps. When you prompt an AI system, it knows exactly what you typed and nothing else. Every piece of context you leave out is context the AI does not have. Every ambiguous term you use will be interpreted in whatever way the model’s training suggests, which may not be your intended meaning. Every assumption you make about what the AI “should know” is a potential source of a response that misses your point entirely.

The students who communicate well with humans are the same ones who get the most out of AI tools, because the skill is identical: provide sufficient context, choose precise words, and think about whether your input contains everything the recipient needs to understand your intent accurately.

The Difference Between the Two

Go back to the opening of this episode. Some people are effortlessly clear: you follow their thinking without strain, you understand their terms without asking, and you leave the conversation knowing exactly what they meant. Some others trigger confusion with every sentence.

The difference between those two people is not intelligence, not vocabulary, and not charisma. The difference is that the first person has learned to provide sufficient context and to think from the audience’s perspective before they speak, and the second person has not. The good news is that the first person was not born that way. They practiced, and so can you.