The adolescent scientist

Imagine our scientific career is mapped to a normal human lifespan. Then, in the first PhD year, we are all babies. We don’t know anything, and we are eager to learn. We receive ample guidance to make baby steps. And we grow up fast. By the end of the second year, which is where I am now, we have become adolescents.

Adolescence is a tricky stage. The adolescent scientist is not a child, neither is he an adult. He has gained certain skills, and there are certain expectations of him. Yet, he is still looked upon as a student. He needs to act confident but not arrogant. He needs to gain independence while yearning for guidance. It is hard to strike a balance.

The difficulties not only stem from the outside, but also from within the adolescent scientist himself. The more he learns, the more he needs to learn. With that comes self-doubt: am I good enough? Will I ever be? While asking where he is now, the adolescent scientist needs to think about where he is going. More questions. Part of growing up.

Growing up is the journey where one discovers his identity; it is a process of self-awareness and self-adjustment, baffling and tedious. Learning is a journey where one discovers his passion; it is a process of searching for a question and working for an answer, perplexing and laborious. But with every discovery comes enormous joy; it pushes him forward, ready to ask another question. In a sense, he embraces his adolescence. This stage is, without a doubt, a crucial and memorable part of his scientific career.

Advertisements

Look at the whole audience, and live in the story

Look at the entire audience, not just the most important person in the room. Student speakers often do the latter. Their eyes are drawn to the figure of authority—in the classroom, it’s the instructor, and in a seminar, it’s the department head or a senior professor. While this is still better than no eye contact at all, it is certainly not ideal. Putting aside the fact that the rest of the audience is ignored, this act weakens your position as a speaker. When you speak to the entire audience, you are a scientist sharing your results with the group. When you speak to only the instructor or the chair, you are a student presenting to a professor (and waiting to be assessed). The two positions are different!

The next piece of advice came from S, my unofficial mentor. He said that when he presents, he lives in the story. Now that just brings the “presenting science is telling a story” idea to a whole new level. If one is able to imagine the entire talk as a story in which he is a character, one can actually act out the story. The message will be loud and vivid.

Now, bring the two pieces together: look at the whole audience, and live in the story. The whole talk has just come alive.

Some notes on academic talks

Over the past two years, I have listened to many academic talks, and I have been paying attention to the speakers. Through these observations, I have drawn some important lessons for my own talks. Here are a few things to look out for.

The title carries the most important ideas of a talk, yet many speakers often fail to highlight it. Most speakers, especially younger ones, just quickly read through it from the slide and jump to the next slide. Experienced speakers, however, often pause to tell a bit more; sometimes they tell a story about the origin of the work, and sometimes they explain one or two key terms in the title if the audience are not experts in the field. Personally, I like to use the title as an outline for the talk. I emphasize several keywords, which appear one at a time on the title slide, and tell the audience that these are the things that I will talk about. I then show the remaining words, those that link the keywords, and along with them I give the overall story of how everything comes together.

The “Table of content” slide is a dreaded one, especially if you show 10 items and read through them one by one. Some avoid this by showing 10 items, saying “This is the outline” then jumping right to the next slide. That does not work either. The title slide, if present, should contain only three or four big ideas that strongly resonate with the title (which, by the way, is still fresh in the audience’s mind). Avoid boring content slide that says “Introduction, Methodology, Results and Discussion, Conclusions”. In an academic talk, these go without saying.

To make people interested in your talk, you must motivate it well. The way to do that is to answer why. “Why this? Why now? Why this way? Why should the readers care?” are the four crucial questions in the introduction of a paper, according to Jean-Luc LeBrun¹; certainly they must be addressed in the introduction of a talk as well. Tell the audience why your work is important. Tell them why you set out doing what you did. Tell them why you did it in this way and not that. Tell them what the results mean to science and to the world. Don’t just tell them “This is the problem, this is what I did, this is what I found, and in summary, we have ABC”.

This one I learned from Joshua Schimel²: end your talk with the Conclusion slide, and leave it hanging there during Q&A until you have to go back to another slide to answer a question. The Conclusion is the juice of your entire work, the “smile” of your paper (Jean-Luc LeBrun again), and the take home message you want the audience to remember. So stay on the Conclusion slide. There’s no need for a “Thank you” slide or worse, a “Q&A” slide; just say it out loud, e.g. “Thank you for your attention, and I look forward to your questions.”, and leave the audience time to read and assimilate your conclusions.

Be aware of filling words. One speaker starts every sentence with “OK”, another with “So, OK, so…”, and another with “Basically,…”. One speaker fills every sentence with “like”, and another ends every sentence with “alright”. These filling words are naturally present in speech, but when they appear in every sentence, the speech becomes heavy and distracting. Avoiding them takes a lot of self-conscious effort, but it is doable.

Always leave ample time for Q&A. Questions are very important. Firstly, they help you gauge how well you presented—having many questions indicates that you have engaged the audience and triggered their curiosity. Secondly, questions offer you critics, feedback and new ideas, so make sure they happen.

These are my personal lessons I learned from observing others. This blog post is my self-reminder, but maybe it can help you along the way. Don’t take my word for it though, draw your own lessons from your own observations, and you’ll remember those much better. Have fun doing science and sharing science.

Footnotes:

1. Scientific Writing 2.0—A Reader and Writer’s Guide

2. Why do people blow the punchline in scientific talks? The destructive effect of acknowledgements slides

 

A question as a paper’s title

Yesterday, I came across a paper titled “What are the best covariates for predicting Y?”¹ Having a question as a title is tricky; sometimes it’s good, sometimes it isn’t. Joshua Schimel wrote an excellent guideline² on this topic. Based on that guideline, I would like to analyse this title in this post.

As I read this title, I imagined myself as a professor (my aspiration, after all). A student came up to me and asked the exact same question. I would think to myself, “Best in terms of what?” The question is not specific enough. It is certainly not possible to have a set of covariates that is the best in all situations. So, I would have to ask the students to be more specific; I might have to list several situations and say which set of covariates is generally considered best for each of these situations and why; and I must then conclude that this is the state-of-the-art knowledge at the moment, but when a new situation comes up, more research is needed.

When a question is asked in a paper’s title, the reader expects that the answer will be provided in the body of the paper. But in this case, most readers may react somewhat similarly to how I did. They may think that the paper had better qualify the claims specifically (as above). This means the the reader is prompted to be careful and skeptic. In other words, he knows that he will be disappointed because the original question will not be answered, but only a more specific one will be.

So, why not ask a more specific question from the beginning? From the abstract, it seemed that a more suitable question should be in which circumstance is which covariate good to use. That is what they did in the paper. They considered 6 covariates in 3 classes and tried different combinations and assessed goodness of fit. They concluded that one class of covariates was the best for one type of model, another the class for another type of model, and the third class was never the best. They did not give a best overall combination.


Footnotes

  1. Out of respect for the authors, I did not show the actual title, but a paraphase. Schimel also did this trick in his Writing Science book.
  2. https://schimelwritingscience.wordpress.com/

Misconstrued actions

I have to begin my writing challenge with a rather sad story—one about a moment of social awkwardness I encountered today.

There was a series of presentations today by several PhD students. Before it started, I was asked by the programme coordinator to be the timekeeper for the talks. When it was about to start, the department head was absent, and one faculty member, I’ll call him A, agreed to be the chair. As the first talk went over its time limit and took up most of its Q&A time, A said that the speaker had a couple of minutes left. As I was keeping the time, I said that there was 5 minutes left for both presentation and Q&A. More than a minute later, when the presentation ended, I said there were three and a half minutes left, so we could have two questions. Faculty member B, sitting next to A, said “Now he wants to be the chair”, and A said something similar. I clarified that “I am the timekeeper.” When the first talk is over, A thanked the speaker. Another faculty member, C, said “Why are you thanking him, he [pointed towards me] should be thanking him.” The audience laughed.

Now, in retrospect, I think A was not very happy when I first clarified the time, and I think neither A, B nor C knew that I was the timekeeper. They thought I went over the line. Having had some time for recollection, I think I did. But that was not my intention. I was just clarifying things. I overdid it (I am always serious about what I do). My actions annoyed these people, and the actions were misconstrued as “trying to be the chair”, which was never my intention.

A commitment to writing

I set up this blog as a place to practice writing. Ironically, as I’m now writing my first paper, I have been neglecting this blog for a while. I thought I was already writing, so there was no need to practice writing. I now realize that such reasoning is flawed. Writing a paper is like running a marathon, it is a long and enduring process. When we train for a marathon, we don’t just run long distances. That is the core of the training, but we need to do more. We need to do interval training to beef up the cardio, we need to train different muscles to strengthen them individually, and so on. Similarly, practicing academic writing doesn’t mean just writing papers. I need to do a variety of other writings to flex up the writing muscles. That I am already writing my paper should not be an excuse to stop practicing.

Yesterday, I read about a business consultant who maintained a strict habit of writing every day. The result was that despite his busy schedules, he managed to publish two books. What’s more, the core content of the books came from his daily writings. I was inspired.

So, at this very moment, on a long bus ride home in a rainy evening, I commit myself to writing something every day. It could be this blog, my baby’s blog, my research journal or event in my little notebook if I don’t have any access to a computer. For now, I will keep this writing time flexible while trying to find a good routine. I’ve frequently heard and read that it is best to fix a time slot for writing, but it’s just not possible right now, and I don’t want to overwork myself.

In the spirit of deliberate practice, I will focus on one aspect of writing in each practice piece. The focus can be on conciseness, fluidity, story structure, etc. as I recently learned or am learning at the moment of writing.

This commitment is for a very long term. But as a baby-step start, let me challenge myself to do it for one whole week.

And the challenge starts now.

A nice duality problem

In the recent Qualifying Exam, there was a very nice (I mean, tough) problem in the Linear Optimization paper that nobody could solve. I was able to crack it two days later. Here it is:

Problem

A matrix A \; (n \times n) is positive semidefinite if x^\intercal A x \geq 0 \; \forall x \in \mathbb{R}^n. Prove that if A is positive semidefinte then the following system has a solution: \{x^\intercal A \geq 0; \ x \geq 0; x \neq 0 \}.

Solution

\begin{cases} x^\intercal A \geq 0 \\ x \geq 0 \\ x \neq 0 \end{cases} \Leftrightarrow \begin{cases} x^\intercal A \geq 0 \\ x \geq 0 \\ \mathbf{1}\cdot x > 0 \end{cases} \Leftrightarrow \begin{cases} x^\intercal A \geq 0 \\ x \geq 0 \\ -\mathbf{1}\cdot x < 0 \end{cases}

where \mathbf{1} is a vector in \mathbb{R}^n whose entries are all 1.

Consider the following linear program

\min \ -\mathbf{1}\cdot x

subject to

x^\intercal A \geq 0

x \geq 0

Its dual is

\max \ \mathbf{0}\cdot y

subject to

Ay \leq -1

y \geq 0

Observe that the dual is infeasible, because if there exists a solution y such that Ay \leq -1 < 0 and y \geq 0 then this implies y^\intercal Ay < 0, contradicting the fact that A is positive semidefinite.

Since the dual is infeasible, the primal is either infeasible or unbounded. But x = 0 is a solution to the primal, so it is unbounded. Thus there exists a solution x such that \{-\mathbf{1}\cdot x < 0; x^\intercal A \geq 0; x \geq 0\} Q.E.D.

Now that the math is done, it’s time for some story. All of us were taught that the Farkas lemma is extremely important, and thus all of us tried to use it to solve this problem, to no avail. I was particularly bugged by the condition x \neq 0, which we had never seen in all our exercises. The exam was on a Wednesday. Two days later, on a Friday evening, on a bus ride from school home, I was determined to solve it. Perhaps it was the motion and the rumbling sound of the bus, perhaps it was my subconscious mind having spent enough time on the problem, or perhaps it was just my shining moment, but I finally realized that x \neq 0 means \mathbf{1}\cdot x > 0, given that x \geq 0. Now the next question is how do I use that in an LP? Well, I can’t use it in the constraints, so it must be in the objective function. Let’s try to \max \; \mathbf{1}\cdot x. But wait, I have a \geq constraint, so I must \min \; -\mathbf{1}\cdot x. And the rest is easy. That was probably one of my proudest moments in the PhD program so far.