Ok, now I've finished the qualifying exam (and passed, yay), this article is ready to release.
Qualifying exam at Berkeley is basically a presentation (two hours at maximum). In the exam, you talk about your past project and your thesis plan. While you may have made some progress towards a promising research topic, failing the exam is still possible if your proposal isn't delivered clearly. Failing an exam due to presention rather than any technical merit is annoying, but it is what the system is. The best strategy is to make full preparation for the exam.
I've been through four iterations revising my slides.
The first iteration is basically a translation from my written paper to the slide format. I copied all the figures and used the titles of each section for slide sections. The overall translation was simple and straightforward, so I was happy. Until I made the first practice talk. The slides contain too much background and I didn't finish the presentation in two and a half hours. Because I have learned a lot of interesting knowledge along the research work, I am eager to share (or maybe it's a desire to impress the audience of my knowledge). Having too much background and explaining them leads to little time for discussing my proposal.
In the second pass, I tried hard to cut the total number of slides from 60 to 40. While I personally felt good the presetnation's status--the logic seems clear and content is succinct. When I did another practice with another set of fellow graduate students, I did not finish again. There were a couple of lessons learned in this practice and many of them are summarized below. But the core lesson is that "I am assuming too much for the audience." Because I moved away all the background and I didn't try to explain them, they kept asking questions that causes a back delay. Although the solution is quite simple---"offer figures and enough examples," my second iteration was too abstract.
My third pass incorporated some technology tools from the presentation software. Specifically, Keynote allows you to add links to another slide, jump back and forth, and press "Z" to go back to the previous slide, etc. These are useful to simplify the presentation outline: the main content is kept lean and all background stuffs are moved to backup slides with links to them. This also helps save space and time for examples and figures. After this iteration, my slides went from 40 to 30. The only problem with this iteration was that I got too exhausted in the end and the logic flow between slides broke. My friend pointed out a number of places that I could make improvements and I am genuinely grateful for those suggestions.
The last pass was actually a fun one. I almost rebuilt the slides from scractch except reusing figures and animations from the third pass. The main slides are kept minimum but each slide has a sentence about the points I need to make. This version has a ton of backup slides. To give some concrete numbers, I had 30 slides for presentation and 32 backup slides. In this way, during the exam, whenever a question was raised, I was able to jump to a slide that I've prepared. Even on that slide, the argument was not strong; the fact you've thought about the question and you can show evidence of your thought is a great way to convince the committee for your preparation. At least I thought this technique effectively impressed the committee, and give them a reason to reward my hard work.
With the last pass, I passed my qual. If you are curious what the final version looks like, here is the PDF version (27.5 MB). Note that there is no animation with the PDF file.
Ok, next, I will talk about a few presentation guidelines learned in this process. I'd like to emphasize the properties of these guidelines: you can apply in your presentation. By "apply", I mean you can almost run it like an algorithm to check if your presentation satisfy them.
Synonyms removes a feel of repetitive in written materials. Synonyms can show off you knowledge and vocabularies. Synonyms, when used properly, may even trigger interesting discussions about identity and concepts. But synnonyms are bad in presentations. The audience will be confused by what you meant in different contexts because it's very hard for them to go back and compare.
In early iterations of my slides, I used "accuracy" at one place and used the word "utility" at a different place. Although it takes one sentence to explain they represent the same concept and can be used interchangeably, this still takes time. And when this situations happen for multiple words, the audience will lose track of what you are talking about.
Use the most meaning word and keep synonyms away.
As the slides maker, you perfectly know what all the symbols mean. But for an audience who just see your slides for the first time, they don't. What are all the points in the figure? What does the line mean? A figure without legend can lead to a completely different interpretation.
You may be thinking: I will talk about what they mean with verbal explanations. This will work if your audience pays full attention; and you know that's a big "if."
Check every rectangle, circle, curve, line or point on your figure. If things are not labelled, label them.
To deliver a message with your presentation, you don't have to cover everything upfront. Some people prefers to structure the presentation in a story-line mode: story opening, expanding, further investigation and concluding. While it's a good structure, it does require the audience to pay enough attention along the way.
The alternative way is to talk about a high-level view first and then dive into details.
Here the only time when you need your audience's attention is when you are talking about the glance view.
This may be hard if you are trying to stuff a lot of details to show off. From my experience, too many details will only confuse your audience because they don't know what you are talking about. Don't try too hard on proving something; simply focus on delivering a message on the topic.
Check if you have "what" (the glance) first and then "why and how" (the details).
If people are allowed to stop you in the middle and ask questions, you may want to reduce the attack surface so that at least you can finish the presentation.
Avoiding synonyms is one technique to minimize attack surface. Be conservative with the details you give out is another one. Don't get me wrong here: you do need details to explain the problem, but avoid any unnecessary points that doesn't contribute to your arguments.
One way to check this rule is to count the number of slides. Having too many is a bad sign!
You may want to have some fancy slides where each slide has only a few words. That style could work if the few words capture what you want to express in a precise and concise way. However, for most times, you want evidence to back up your claim. Putting the evidence in a separate slides may make you feel cleaner and cooler; it will be difficult for the audience because "each slide transition is a context switching." It's easier for someone to follow when the content is smooth. Sudden transition definitely requires mental efforts to keep up.
Make sure slide transitions happen only when necessary; and during the transition, move the focus from familiar materials to unfamiliar materials.
When you preppare your presentation, you have a certain logic order of organizing the materials. You have decided to postpone discussing some points in the begining. When the audience asks, you will get excused by saying: I will talk about it later. But if the same term appears for the second time, and you are still planning to talk about it later, it will make the audience unsatisfied. Because they are now not sure what you mean by "later."
Check your slides, if you have a term that is only explained in the third time, try to restructure a bit.
Some presentations have the goal to impress the audience with glorious math and details; when the audience leave, they will say "wow, that's impressive." But more effective presentations should leave the audience with some new gained knowledge. So focus on clarity and don't feel bad if you include materials that seem obvious.
On the other hand, some materials are only obvious to you. And without communicating with your audience first, you don't know if any particular points are self-explanatory or not. So add enough explanation, examples and figures so that the points are self-contained.
The audience will often have a pleasant experience if they understand 80% of your talk and for the other 20%, they feel they've learned something.
To check this rule, walk over your slides with a fresh perspective. You may ask a friend, your mom or anyone else to see if they can get 80% of the materials without your too much help.