Transcript dry-run은 위에께 준비된 상태로 3번정도?

How To Give a Good Talk
Last Revised on 2008.10.27.
Sue Moon
Associate Professor
Computer Science Department
Why Is It Important?

Because …
 A good talk is a highly effective means of one-to-many

communication
Prospective employers/readers are in the audience
Internships, post-docs, academic/industry positions

Vicious Cycle
 Good speaker
More invitations, more talks, better speeches
 Bad speaker
If you’re a student : no job interviews
If you have a job: lose popularity, get fewer invitations, disappear
into oblivion
Know Your Audience
Who are they?
 What do they want from your talk?

 Their technical background determines:
Academic info vs industry overview
Technical details vs opinions
At the Podium
Always face the audience
 Have eye contact with audience

 Don’t show the back of your head to audience
 Have your computer monitor right in front of you

Look relaxed
 Check your idiosyncratic gestures
Swinging, hands in pockets, on waist, or in the back

Use moderate amount of gestures
 Keep audience alert
Your Title Slide

It should be informative
 Talk title
 Location and Time
 Your work or someone else’s?
 Collaborators?

Any title page should be as informative
Your Slides

Be succinct and descriptive
 Avoid full sentences
 Do not list only nouns; use action verbs to be descriptive

Use a small # of colors
 Too many colors distract audience from main focus

Use big fonts
 Readable without restraining
 Limit # of lines per slide
Graphs, Tables, and Equations

Use as few tables and equations as possible
 Tables are hard to read
 Equations are hard to follow

Use as many graphs as possible
 Graphs are easy to read and remember

Graphs
 Make legends and axis labels big enough

Use animation and figures when possible
 In RGB colors; pastel colors don’t always work due to
lighting
Time Your Talk

Allocate 1 ~ 3 minutes per slide
 Every slide counts and takes up time
 15 slides for 20 min talk
 30~35 slides for 40 min talk
 100+ slides for 1hr-long talk => horrible

Prepare transitional comments between slides
 Keep audience involved

Plan time for intro & motivation
 For talks shorter than 30 minutes, make sure you spend
1/3 of time on intro & motivation
Prepare Answers to Likely Questions

Ask yourself 3~5 most likely questions
 Prepare backup slides for those questions

If asked an unexpected question
 And if you don’t have an answer
Acknowledge you haven’t thought about it and thank the
person
Appendix A:
Guideline for Your 1st Public Talk
For First-Time Non-Native Speakers
[Dry Run #0]

Go over the storyline/storyboard with co-authors
[Dry Run #1]



Have the complete set of slides ready
Expect lots of structural changes
Write down a script for the first 5 pages
** Most pointed-out weaknesses **



“You don’t explain why you’re showing me the slide”
“You don’t explain what lesson to take from the slide”
“Why” @beginning and “So What?” @end
For First-Time Non-Native Speakers
[Dry Run #2]


Incorporate all the comments
Record your talk and see it for yourself
Physical peculiarities: body swinging, showing the back of your head to
the audience, hands in pockets, hands on your waist, …
Others: frequent coughing
[Dry Run #3]



See if you can replace tables with animations
See if you explain any part better with animations
Write down a script for the complete talk
[Dry Run #4]



See if you can escape from the typical “monotonous” speech
Final check on all the points above
Do you deliver your enthusiasm about your work?
You Shall Not Get Onboard
Before You Have Not Done
Four Dry Runs
“You SHALL NOT register before a
decent dry run” – Sue Moon
At the Conference
[Dry Run #5]

Upon arrival in the hotel room by yourself
[Dry Run #6]


The day before the real talk
By yourself or in front of whoever you can entice
You’re not the only one





Stefan Savage practiced his 1st SOSP talk 5 times
Zhi-Li Zhang did more than 7 dry runs of his job talk
Stefan and Zhi-Li both recorded and watched their
talks
Jeff Mogul still practices his talk whenever possible
XXX had to give the SIGCOMM talk instead of the
student
Appendix B:
Non-Native Speaker’s
Disadvantage
How Harder Do You Have to Work?

IMHO, at least 30%
 In paper writing and presentation

If you have to work harder than 30%
 Either you’re not ready for PhD
 Or study English intensively for 6 months
Take a leave of absence!!!

How to bridge the 30% gap?
 So much an advisor can do
 Start now and invest time for your future
Appendix C:
Bad Talks
Opinions about Bad Talk

Too many bad talks in local workshops/confs
 Slides full of diagrams and words
 Graphs w/o proper accreditation
 No distinction of original contributions from related work
 No transition between slides
 No “why” and “so what”
 No respect for time limit
 More of a propaganda than a research talk
More “We should” than “we have done”

Don’t turn yours into yet another one of them
Appendix D:
Tips from Fellow Students
장 건의 경험담

0) slide에 알아야 할 내용 다 적고, 다양한 animation을 통해 혹시 발음을 못알아
듣더라도 따라갈 수 있도록.
1) full script를 준비
2) 첫 10페이지 정도 완벽하게 외우기(실험 결과들 전까지)
- 사실 영어가 잘되면 이야기할 내용들만 정확하게 다 외워도 되겠지만,
non-native speaker입장에서 한번 당황하기 시작하면 겉잡을 수 없으므로 거의
다 외우다시피 하는게 좋은거 같아요. 결과들은 그래도 설명하기가 쉬운거 같은
데, 그래프 설명하는거는 생각보다 어렵습니다.--;
그래프도 어떻게 말할찌 꼼꼼하게 준비하고 axis설명 다 하고 해야 합니다.
3) 파워포인트에 녹음 기능 사용해서 들어보기(들어보면 엄청난 konglish에 압박
이.) (시간도 재줘서 좋습니다.)
4) dry-run은 위에께 준비된 상태로 3번정도?
5) 만약을 대비한 각 페이지별 얘기할 내용들에 대한 cheat sheet
6) 강조할 부분(강조해서 말할 부분) 미리 찾아서 연습!
7) 예상 질문과 대답
0,5,6,7은 dry-run을 하면서 많이 comment를 받을 수 있으리라고 보입니다.
그 외에 어려운 단어를 되도록 발음하기 좋은 단어로 바꾸는것도 한가지 방법인
거 같습니다.