[PDF] - Ho w to Presen t a P ap er in Theoretical Computer PDF Document

SLIDE 1 Ho w to Presen t a P ap er in Theoretical Computer Science: A Sp eak er's Guide for Studen ts

Ian

P arb erry y Departmen t

f

Computer Sciences Univ ersit y

f

North T exas July 29, 1993 There are man y p

in

ts in y

ur

career at whic h y

u

will b e called up

n

to presen t a pap er in Theoretical Computer Science, p erhaps a pap er written b y someb

dy

else in a graduate seminar,

r

y

ur
wn

researc h at a conference, departmen tal collo quium,

r

job in terview. This skill is particularly imp

rtan

t if y

u

in tend to pursue a career in academia. While researc h excellence is the main criterion for success as a theoretical computer scien tist, y

ur

career will b e assisted if y

u

gain a reputation as a comp eten t sp eak er. A comp eten t sp eak er will more lik ely b e in vited to giv e collo quia at leading univ ersities and in vited talks at imp

rtan

t conferences than a medio cre

ne,

pro vided their researc h is

f

similar qualit y . The exp ected qualit y

f

the presen tation can b e the deciding factor in the selection

r

rejection

f

a con tro v ersial conference pap er in cases where there is no clear consensus from the program committee . The author do es not claim to b e an exp ert in

ral

presen tation, but has attended (and giv en) enough bad talks to b e able to dra w up a few rules

f

th um b. This note is written primarily for studen ts in Theoretical Computer Science, but most

f

it is relev an t to

ther

sub-disciplines

f

Computer Science, and some to an y scien tic discipline. Dieren t st yles

f

presen tation w

rk

for dieren t p eople. These guidelines are in tended to assist y

u

in dev eloping a w

rk

able st yle

f

y

ur
wn.

The remainder

f

this note is divided in to four sections. The rst, \What to Sa y and Ho w to Sa y It", concen trates

n

c ho

sing

and

rganizing

the material to b e presen ted. The second, \Getting Through to the Audience", fo cusses

n

presen tation. The third, \Visual and Aural Aids", discusses the prop er use

f
v

erhead pro jector transparencies and the microphone. The fourth, \Question Time", is dev

ted

to the p erio d after the talk during whic h the audience asks questions

f

the sp eak er. An earlier v ersion

f

this note has app eared in [2, 3].

Cop

yrigh t c

Ian

P arb erry . This guide ma y b e repro duced without fee pro vided the copies are not made for prot. y Author's mailing address: Departmen t

f

Computer Sciences, Univ ersit y

f

North T exas, P .O. Bo x 13886, Den ton, TX 76203{3886, U.S.A. Electronic mail should b e directed to ian@ponder.csci.u nt.e du. 1

SLIDE 2 1 What to Sa y and Ho w to Sa y It The selection

f

material to include in a talk, and ho w it is arranged and presen ted is crucial to success. Communic ate the Key Ide as Most new results are

btained

b y using a few k ey ideas

r

tric ks, plus the application

f

standard to

ls

and tec hniques. Mak e sure that y

ur

talk emphasizes the k ey ideas and skips

v

er what is standard,

b

vious,

r

merely complicated. Don 't get Bo gge d Down in Details This rule cannot b e

v

er-emphasized. Man y sp eak ers launc h righ t in to the tec hnical details from the start, confusing and alienating most, if not all

f

the audience. Remem b er that y

u

will ha v e b een thinking deeply ab

ut

the material for mon ths

r

ev en y ears, whereas most

f

the audience will either hear ab

ut

the sub ject matter for the rst time during y

ur

talk,

r

at b est will b e v ery rust y . Things that y

u

tak e for gran ted will ha v e to b e explained carefully . Details are

ut
f

place in an

ral

presen tation. After all, an yb

dy

with mo dest tec hnical skills can ll in the details b y studying the pap er at leisure, and an yb

dy

lac king tec hnical skills will w an t to hear ab

ut

the details ev en less. The audience w

uld

appreciate an

v

erview

f

the pap er so that they can determine whether the pap er is w

rth

reading. A go

d

talk motiv ates the listener in to reading the pap er and mak es the task

f

reading it easier. Structur e Y

ur

T alk Y

ur

presen tation should b e brok en in to sev eral distinct parts, eac h with its

wn
b

jectiv es and st yle. Eac h part should b e clearly delineated. The audience should b e steered gen tly from

ne

part to the next. A w ell-structured talk is easier to understand than a ram bling, unstructured

ne,

and it also mak es more ecien t use

f

time. Use a T

p-down

Appr

ach

The follo wing is a template for pro ducing a talk. It consists

f

four parts; the In tro duction, the Bo dy , T ec hnicalities and the Conclusion. The In tro duction is a general, informal description

f

the pap er. The Bo dy giv es a more formal, but abstract description, and the T ec hnicalities section tak es a detailed lo

k

at a critical part

f

the pap er. The audience is th us in tro duced to the material in a top-do wn fashion. The Conclusion concisely summarizes the k ey results, and wraps up the talk. Naturally , v ery few talks will t this template exactly , since the structure

f

a talk v aries greatly with sub ject matter, length

f

presen tation, author, presen ter, and audience. The In- tro duction and the Conclusion are fairly standard, but there is ample ro

m

for customization in the

ther

t w

parts.

F

r

example, some complicated sub jects ma y need sev eral passes, 2

SLIDE 3 eac h

f

successiv ely greater detail, instead

f

the t w

-pass

b

dy-tec

hnicalities sc heme describ ed b elo w. Y

u

should use this template as a starting p

in

t

nly

. Don't b e afraid to b e inno v ativ e. The In tro duction This is p

ssibly

the most imp

rtan

t part

f

y

ur

presen tation. It sets the tone for the en tire talk. It determines whether the audience will pric k up their ears,

r

remain slump ed in their c hairs. A lot

f

snap decisions ab

ut

y

ur

comp etency are made b efore the In tro duction is

v

er. First impressions are v ery imp

rtan

t.

Dene

the Pr

blem

An amazing n um b er

f

sp eak ers forget this simple p

in

t. No matter ho w dicult and tec hnical the problem, it can usually b e describ ed succinctly and accurately in under v e min utes. This time is w ell in v ested. If the audience do esn't understand the problem b eing attac k ed, then they w

n't

understand the rest

f

y

ur

talk.

Motivate

the A udienc e Explain wh y the problem is so imp

rtan

t. Thro w in a little philosoph y if necessary . Ho w do es the problem t in to the larger picture? If it in v

lv

es a mo del

f

a real-w

rld

phenomenon, then ho w go

d

is the mo del? What are its applications? What mak es the problem non trivial? Y

u

can return to these issues in the Conclusion, when y

u

can re-address them with the b enet

f

hindsigh t.

Intr
duc

e T erminolo gy The use

f

terminology and jargon should b e k ept to a minim um , but is imp

ssible

to a v

id

en tirely . All terms must b e in tro duced early . It is also useful to remind the audience

f

the denitions at critical p

in

ts later in the talk.

Discuss

Earlier Work Researc h is not usually carried

ut

in a v acuum. There will almost alw a ys b e

ther

relev an t

r

related w

rk,

whic h y

u

should describ e. Presen t an

rderly

synopsis

f

these previously-obtained results. A table is

ften

used for this purp

se.

Be sure to men tion the author

f

eac h pap er and its date

f

publication. Compare and con trast them with eac h

ther

and with y

ur

pap er.

Emphasize

the Contributions

f

your Pap er Mak e sure that y

u

explicitly and succinctly state the con tributions made b y y

ur

pap er. The audience w an ts to kno w this. Often it is the

nly

thing that they carry a w a y from the talk. Don't mak e the audience searc h for the information in a morass

f

details | they ma y get it wrong.

Pr
vide

a R

ad-map

Giv e the audience a brief guide to the rest

f

the talk, along the lines

f

the last paragraph

f

the in tro duction to this note. But don't mak e it a dry litan y

f

dull 3

SLIDE 4 generalizations (\rst I will presen t the in tro duction, then summarize earlier w

rk,

then presen t the main b

dy

, and end with the conclusion"). Instead, giv e a short preview

f

what will b e in eac h section. This part

f

the talk can b e eliminated for v ery short conference presen tations. The Bo dy This con tains the meat

f

y

ur

presen tation, and is the p

in

t at whic h the atten tion

f

the audience will start to w a v er if y

u

messed up y

ur

In tro duction.

A

bstr act the Major R esults Describ e the k ey results

f

the pap er. Y

u

ma y presen t the statemen ts

f

the ma jor theorems, but not their pro

fs.

Y

u

will probably ha v e to get a little tec hnical here, but do so gradually and carefully .

Explain

the Signic anc e

f

the R esults P ause, and explain the relationships b et w een the formal theorems that y

u

ha v e just presen ted and the informal description that y

u

ga v e in the In tro duction. Mak e it clear to the audience that the results do liv e up to the adv ance publicit y . If the statemen ts

f

the theorems are v ery tec hnical then this ma y tak e some time. It is time w ell-sp en t.

Sketch

a Pr

f
f

the Crucial R esults The emphasis is

n

the w

rd

\sk etc h". Giv e a v ery high-lev el description

f

the pro

fs,

emphasizing the pro

f

structure and the pro

f

tec hniques used. If the pro

fs

ha v e no structure (in whic h case it ma y b e assumed that y

u

are not the author

f

the pap er), then y

u

m ust imp

se
ne
n

them. Gloss

v

er the tec hnical details. It is a go

d

idea to p

in

t them

ut

but not to explore them. T ec hnicalities If y

u

are doing w ell, then most

f

the audience will ha v e follo w ed y

u

up to this p

in

t. Ho w ev er, the exp erts ma y b e in danger

f

getting b

red.

The more can tank erous among them ma y not b eliev e y

ur

results, since y

u

ha v e presen ted them

nly

in high-lev el terms. No w y

u

can giv e some tec hnical details. This will also help the non-exp ert to see what is really going

n

b ehind the scenes, lest he

r

she is lulled b y y

ur

p

lished

st yle in to thinking that it is all to

easy

.

Pr

esent a Key L emma Cho

se

just

ne

k ey result. It should b e imp

rtan

t, non-trivial, should giv e the a v

ur
f

the rest

f

the tec hnical details and should b e presen table in a relativ ely short p erio d

f

time.

Pr

esent it Car eful ly Y

u

ma y descend in to messy tec hnical details, but try to b e as succinct and clear as p

ssible.

Y

u

migh t w an t to ll in

ne
f

the gaps that ma y app ear in the pap er, to 4

SLIDE 5 giv e the audience something that they could not get b y sta ying home and reading the pap er. (Almost all tec hnical pap ers con tain small gaps. Lac k

f

space usually prev en ts pro

fs

from app earing in all their gory detail.) The Conclusion Y

ur

aim here is to round

the

talk neatly . Y

u

should discuss the results briey in retrosp ect placing emphasis where it is needed.

Hindsight

is Cle ar er than F

r

esight Y

u

can no w mak e

bserv

ations that w

uld

ha v e b een confusing if they w ere in tro duced earlier. Use this

pp
rtunit

y to refer to statemen ts that y

u

ha v e made in the previous three sections and w ea v e them in to a coheren t synopsis. Y

u

will regain the atten tion

f

the non-exp erts, who probably didn't follo w all

f

the T ec hnicalities section. Lea v e them feeling that they ha v e learned something nonetheless.

Give

Op en Pr

blems

It is traditional to end with a list

f
p

en problems that arise from y

ur

pap er. Men tion w eaknesses

f

y

ur

pap er, p

ssible

generalizations, and indications

f

whether they will b e fruitful

r

not. This w a y y

u

ma y defuse an tagonistic questions during question time (see Section 4).

Indic

ate that your T alk is Over An acceptable w a y to do this is to sa y \Thank-y

u.

Are there an y questions?". Know Y

ur

A udienc e Mak e sure that y

ur

talk is prepared at the righ t lev el. There are four t yp es

f

audience that y

u

will most lik ely meet. 1. Scientists Most

f

the audience w

n't

ev en kno w what Computer Science is. Emphasize the In tro duction and the Bo dy . Omit the T ec hnicalities section. Remem b er that y

u

will ha v e to dene man y terms that y

u

ha v e tak en for gran ted for y ears. Pro vide more philosophical bac kground. Ordinarily y

u

will not meet this kind

f

audience un til later in y

ur

career. 2. Computer Scientists Most

f

the audience will not kno w what Theoretical Computer Science is. Y

u

can in tro duce a small T ec hnicalities section, but k eep it brief. Be careful with y

ur

denitions. The emphasis should remain

n

the In tro duction and the Bo dy . This t yp e

f

audience will b e t ypically encoun tered during job in terviews and departmen tal collo- quia. 5

SLIDE 6 3. The

r

etic al Computer Scientists Y

u

will no longer need to b e quite so careful with denitions, but it is p

litic

to pro vide reminders for the rust y , the y

ung

and the less men tally agile in the group. The emphasis should b e

n

the Bo dy

f

the talk, but the T ec hnicalities section need no longer b e m uted. This t yp e

f

audience will t ypically b e encoun tered during departmen tal collo quia and ma jor theory conferences. 4. Exp erts If y

u

address exp erts in y

ur

eld, then y

u

can get righ t do wn to the core

f

the matter. The emphasis should b e

n

the Bo dy and the T ec hnicalities. Keep in mind that the more exp erts there are in the audience, the more argumen tativ e they are lik ely to b e. This t yp e

f

audience will b e t ypically encoun tered during small, sp ecialized conferences and w

rkshops.

2 Getting Through to the Audience Once y

u

ha v e selected and

rganized

y

ur

material, the next ma jor h urdle is when y

u

nd y

urself

actually standing in fron t

f

the audience. F ault y deliv ery can ruin ev en a w ell-prepared talk. Use R ep etition Oral presen tations ma y b e summed up as follo ws: \T ell them what y

u're

going to tell them. T ell them. Then tell them what y

u

told them". In the In tro duction y

u

tell them what y

u

are going to tell them. In the Bo dy and T ec hnicalities y

u

tell them. In the Conclusion y

u

tell them what y

u

told them. Don't b e scared

f

this rep etition. Sometimes rep etition is the

nly

w a y to clarify misconceptions. Naturally , this means that y

u

should rep eat things in dieren t w a ys, and not quote y

urself

v erbatim. If y

u

ha v e studied Co ding Theory , then y

u

will kno w the imp

rtance
f

redundancy when comm unicating

v

er a noisy c hannel. The c hannel b et w een ears and in tellect is extremely noisy . R emind, don 't Assume If y

ur

pap er assumes a \standard" result in y

ur

eld (for example, the Ch urc h-T uring Thesis, Co

k's

Theorem, etc.), it is w

rth

while to pro vide y

ur

audience with a brief reminder

f

exactly what the result is. Phrasing it as a reminder a v

ids

an tagonizing those who for v arious reasons are not familiar with it (for example, those who w

rk

in a dieren t eld, the absen t-minded,

r

studen ts) b y allo wing them the

pp
rtunit

y to learn while hiding their ignorance, and a v

ids

an tagonizing those who are familiar with it b y allo wing them to \tune

ut".

6

SLIDE 7 Don 't Over-run There is v ery little that is more exasp erating than listening to a talk that drags in terminably past its sc heduled nishing time. The author has

bserv

ed that the qualit y

f

a talk is almost alw a ys in v ersely prop

rtional

to the time that it

v

er-runs. Unless it is explicitly stated

therwise,

when y

u

are told that the talk is to last for x min utes, plan to talk for at most x

5

min utes, and lea v e 5 min utes for question-time. If in doubt ab

ut

the prop

rtions,

then enquire. Departmen tal collo quia generally last for ft y min utes, and conference presen tations from fteen to thirt y min utes, including question-time. Don't try to co v er to

m

uc h material. As y

u

gain more exp erience y

u

will learn ho w m uc h y

u

can co v er in an y giv en p erio d. Un til then, y

u

will ha v e to rehearse y

ur

talk un til it is the correct length. A fort y-v e min ute departmen tal collo quium m ust naturally dier greatly in comp

sition

from a fteen min ute conference presen tation. If y

u

are short

f

time, the rst thing to cut is the T ec hnicalities section. Maintain Eye Contact Main tain ey e con tact with y

ur

audience. Spread y

ur

atten tion throughout the audience instead

f

concen trating

n

an y

ne

p erson

r

group (ev en if they are the

nly
nes

who matter). A go

d

strategy for b eginners is to c ho

se

a few p eople at random in dieren t places in the audience, and lo

k

at them successiv ely . If presen ting at a conference, b e sure to glance p erio dically at the session c hair, who will signal y

u

when y

u

are running

ut
f

time. Contr

l

Y

ur

V

ic

e Sp eak clearly and with sucien t v

lume.

Don't sp eak in a monotone. Av

id

information-free utterances (\Um, ah, er", etc.) Av

id

fashionable turns

f

phrase. Av

id

h yp e. Contr

l

Y

ur

Motion Pro ject energy and vitalit y without app earing h yp eractiv e. Use natural gestures. T ry not to remain ro

ted

in

ne

sp

t,

but a v

id

excessiv e roaming. Don't get b et w een the pro jector and the screen. T ry not to fall

the

dais, if there is

ne

(sp eak ers with true sang fr

id

will con tin ue to talk without in terruption after they ha v e fallen from the dais, ev en if they are momen tarily in visible). T ake Car e with Y

ur

App e ar anc e Go

d

gro

ming

and dress helps, but a v

id

app earing

v

erly

sten

tatious. A jac k et and tie for men and professional attire for w

men

is imp

rtan

t at a talk giv en as part

f

a job in terview, but casual dress is de rigeur for collo quia and conference presen tations. 7

SLIDE 8 Minimize L anguage Diculties English app ears to b e the common language for theoretical computer scien tists

f

all nations. If y

u

are called up

n

to deliv er a talk in a language in whic h y

u

are not completely uen t (b e it English

r
therwise),

the eect

f

y

ur

deciencies in language will b e minim iz ed if y

u

pa y careful atten tion to all

f

the

ther

asp ects

f

y

ur

talk describ ed in this note. It is a go

d

idea to get a nativ e sp eak er to lo

k
v

er y

ur

transparencies b efore y

u

deliv er the talk. A t all costs, try to a v

id

sp eaking from a prepared text. In general, theoretical computer scien tists are accustomed to, and toleran t

f

foreign accen ts and mo des

f

sp eec h. T ry Not to Get A nxious It is easy to b ecome anxious and nerv

us

during a talk, and forget all

f

the advice in this note. The b est an tidote for this is exp erience. Anxiet y can b e reduced b y making sure y

u

prepare adequately b eforehand, and practice the talk in fron t

f

fello w studen ts

r

y

ur

facult y adviser. It is a go

d

idea to sp end 10{15 min utes alone b efore the talk calmly lo

king
v

er y

ur

slides and

rganizing

y

ur

though ts. One sure w a y to increase anxiet y is to pa y undue atten tion to the reaction

f

the most imp

rtan

t p erson in the audience. The pained expression

n

his

r

her face ma y b e the result

f

last nigh t's pizza, not y

ur

presen tation. If y

u

nd y

urself

panic king during a talk, it is b est to pause, close y

ur

ey es, tak e a few deep breaths, calm y

urself,

and then con tin ue. If y

u

b

tc

h a few talks, it will not b e the end

f

y

ur

career. What will b e remem b ered ab

ut

y

u

is the qualit y

f

y

ur

researc h, not the qualit y

f

y

ur

rst few talks. 3 Visual and Aural Aids No w that y

u

ha v e a w ell-prepared talk and can deliv er it with st yle, there is the next stum bling blo c k: the

v

erhead pro jector and the microphone. Use Overhe ad Pr

je

ctor T r ansp ar encies The accepted metho d

f

presen tation diers from

ne

academic comm unit y to another. In the Lib eral Arts, a sp eak er generally reads dispassionately from a prepared text with

r

without the assistance

f

35mm slides that con tain

nly

pictures and diagrams. In Mathematics, a sp eak er generally sp eaks

-the-cu

with the aid

f

a blac kb

ard
r

whiteb

ard.

In the Sciences, a sp eak er generally uses 35mm slides

r
v

erhead pro jector transparencies as an adjunct to their talk. The standard in Theoretical Computer Science is to use

v

erhead pro jector transparencies. A talk app ears more p

lished

when y

u

prepare

v

erhead pro jector transparencies in adv ance. When sp eaking a w a y from home, alw a ys mak e sure in adv ance that an

v

erhead pro jector is a v ailable. Y

ur

hosts will b e able to pro vide

ne

under most circumstances. Y

u

ma y prepare y

ur

transparencies b y hand if necessary , but if y

u

will b e tra v elling and giving the talk man y times, the transparencies will surviv e longer if y

u

pro duce them mec hanically . It is

ften

a matter

f

mac hismo for y

ung

theoreticians to prepare their transparencies b y 8

SLIDE 9 hand at 2 a.m. the nigh t b efore a F OCS

r

STOC conference presen tation. Ho w ev er, v ery few

f

these p eople consisten tly giv e ab

v

e a v erage presen tations, and y

u

are ill-advised to em ulate them at least un til y

u

are more exp erienced. Make L e gible T r ansp ar encies If y

ur

normal writing is a c hic k en-scra wl, then y

u

probably should use a computer to prepare y

ur

transparencies. If this is imp

ssible,

y

u

m ust tak e great pains to ensure that y

ur

writing is legible. Mak e sure that the text is large enough to b e read from a distance. Normal-sized handwriting and 12-p

in

t t yp e are unacceptable. Use standard do cumen t preparation soft w are to pro duce large-t yp e cop y; the author uses SliT E X (see Lamp

rt

[1, App endix A]), whic h automatically pro duces text in a large, easy- to-read fon t. This can b e prin ted directly

n

to transparencies using a laser-prin ter,

r

the hardcop y can b e photo copied

n

to the transparencies. Mak e sure in b

th

instances that y

u

are using transparencies that are made for photo cop ying,

therwise

the high temp eratures inside the mac hine will bak e the transparency

n

to the drum, whic h will mak e y

u

highly unp

pular

with ev eryb

dy

except the repair p erson. The preparation

f

slides b y computer has an added b

n

us: y

u

can use sp elling c hec k ers to disco v er man y

b

vious errors. Don 't Overlo ad T r ansp ar encies Don't put to

m

uc h

n

eac h transparency . T ry not to write full sen tences. W rite do wn denitions, imp

rtan

t p

in

ts, k ey-w

rds

and phrases. The transparency is an adjunct to y

ur

talk; it should b e used for emphasis, to resolv e am biguit y , for precision and for the reten tion

f

information

v

er a short p erio d

f

time. It do es not con tain a transcription

f

y

ur

talk. A go

d

transparency underlines the k ey p

in

ts for the audience and sim ultaneously acts as a prompter

r

cue-card for the sp eak er. Don 't Use T

Many

T r ansp ar encies A common mistak e is to prepare to

man

y transparencies. This indicates that y

ur

talk con tains to

m

uc h material. If y

u

c hange transparencies to

ften,

then the audience will not ha v e time to digest eac h

ne

prop erly . There should b e enough time to read eac h transparency sev eral times while the sp eak er is talking. The n um b er

f

transparencies p er talk will v ary from p erson to p erson, and will dep end up

n

the t yp e

f

material b eing presen ted, and the amoun t

f

text

n

eac h transparency . Allo w an a v erage

f

1:5 to 2 min utes for eac h transparency . V ery short conference presen tations are an exception to this rule, in whic h case

ne

min ute p er transparency can b e managed with eort and practice. A void Slide Covering Av

id

situations where y

u

w an t to sho w

nly

part

f

a transparency . Mak e t w

trans-

parencies instead. Man y p eople

b

ject to \slide co v ering", the act

f

co v ering part

f

the transparency with an

paque
b

ject suc h as a sheet

f

pap er to prev en t the audience from seeing it. It is hard to do gracefully . The co v er sheet will in v ariably slide

,

and y

u

will 9

SLIDE 10 lose v aluable time and savoir fair e in retrieving it. The use

f
v

erla ys is equally fro wned up

n

when used for this purp

se.

Use Overlays Pr

p

erly Ov erla ys are acceptable when y

u

need to mak e small and incremen tal c hanges to a complex transparency . They can b e used to great eect in the righ t circumstances. Ov erla ys come in t w

forms.

In the rst form, t w

r

more transparencies are successiv ely placed

n

top

f

eac h

ther

as the talk progresses. This is to b e a v

ided

since transparencies are t ypically hard to align in the rst place, and are easily dislo dged. In the second form, the

v

erla ys are fragmen ts

f

a transparency that are tap ed to the

uter

edges

f

the main transparency , folded bac k at the start

f

the presen tation, and

pp

ed

v

er

n

to the transparency at the appropriate momen ts. This is far preferable to the rst form, but y

u

should a v

id

constructing transparencies that ha v e so man y

v

erla ys that they resem ble a hedgehog. Do not tap e the

v

erla ys to the cen ter

f

the transparency . Adhesiv e tap e is not as transparen t as it seems, and will usually sho w up as a ragged shado w

n

the screen. It is wise to a v

id

the

v

er-use

f
v

erla ys, particularly with p

rtable
v

erhead pro jectors. In the latter the ligh t source is ab

v

e the transparency and hence m ust pass through the transparency t wice, compared to

nce

for the traditional pro jector with the ligh t source b elo w the transparency . T ransparencies ma y lo

k

transparen t, but they are actually sligh tly

paque:

a stac k

f

as few as six

f

these

n

a traditional pro jector,

r

three

v

erla ys

n

a p

rtable

pro jector, ma y app ear almost completely

paque

under some ligh ting conditions. Use Colour Ee ctively The use

f

colour can enhance a presen tation, particularly when used in mo deration in the gures and diagrams. The colour

f

the text can b e c hanged p erio dically , but it is most eectiv e when there is a clear motiv ation (a simple example ma y b e to put the statemen ts

f

theorems in a dieren t colour from the rest

f

the text). A t all costs a v

id

the temptation to pro duce a rain b

w
f

text. Some colours are more visible than

thers.

Y ello w is almost alw a ys in visible. Use Pictur es and T ables Remem b er the

ld

clic he \A picture is w

rth

a thousand w

rds".

Ho w ev er, b e sure to explain the signicance

f

y

ur

diagrams and tables. There is nothing more m ystifying than a sequence

f

esoteric, unlab elled, unexplained pictures. \Our results lo

k

lik e this" is not sucien t explanation. In this con text, the clic he should b e mo died to \A picture plus a h undred w

rds

is w

rth

a thousand w

rds".

Bewar e

f

the Micr

phone

Y

u

will probably need to use a microphone during a conference presen tation. The session c hair should assist y

u

to attac h it to y

ur

clothing. It t ypically consists

f

a small b

x

whic h go es in to a p

c

k et, and a v ery small microphone

n

a clip whic h should b e attac hed 10

SLIDE 11 to y

ur

clothing as close to y

ur

face as p

ssible

(preferably

n

the collar

r

the upp er part

f

the tie, if y

u

are w earing

ne).

The microphone is attac hed to the b

x

with a short wire, and the b

x

is attac hed to the amplier b y a long wire. Instead

f

a clip, the microphone ma y ha v e a lo

p
f

string that go es around y

ur

nec k. If it is long enough, pass it

v

er y

ur

head. If not, it should detac h from

ne

side

f

the microphone so y

u

can pass it around y

ur

nec k. It is usually hard to detac h and reattac h, particularly when it is

ut
f

sigh t under y

ur

c hin and y

u

are nerv

usly

facing sev eral h undred p eople for the rst time. If y

u

are clumsy in attac hing the microphone,

r

unluc ky enough to drop it, there will b e a loud unpleasan t scraping

r

impact noise heard

v

er the sp eak er. T

a

v

id

this, there is

ften

a small switc h

n

the b

x

whic h deactiv ates the microphone during the attac hmen t. Use this switc h, but don't forget to reactiv ate it. The microphone will b e dislo dged b y brushing against it while gesturing, kno c king the b

x
ut
f

y

ur

p

c

k et, snagging the short wire,

r

stepping

n
r

getting tangled in the long wire. This will

f

course result in a loud noise and a loss

f

time and concen tration while replacing the microphone. If y

u

roam while y

u

sp eak, b e a w are

f

the microphone and its wires. 4 Question Time As w e ha v e already seen, it is customary to end a presen tation with a short p erio d for questions. Y

u

can exp ect to receiv e three t yp es

f

question. The rst is the gen uine request for kno wledge, whic h should cause y

u

no diculties if y

u

are adequately prepared. The second is the selsh question, in whic h the questioner merely wishes to dra w atten tion to him

r

herself and elicit w

nder

at his

r

her abilit y to devise suc h an incisiv e and cogen t question. It is p

litic

to tak e a few seconds to comp

se

an erudite reply that directly

r

indirectly complime n ts the questioner. The third and most imp

rtan

t category is the malicious question, in whic h the questioner attempts to exp

se

the sp eak er as a c harlatan and a dissem bler. This ma y come from

ne
f

t w

p
ssible

motiv es. Unfortunately , there are man y insecure p eople in

ur

comm unit y who constan tly seek to build up their reputation b y tearing

ther

p eople do wn. Alternativ ely , particularly if y

u

are in terviewing for a job, the question ma y b e in tended to see ho w y

u

react to criticism under pressure. Either w a y , exp ect to ha v e y

ur

ego bruised

ccasionally

. Equally under attac k will b e y

ur

selection

f

problem (\b

ring,

irrelev an t, con triv ed"), y

ur

pro

fs

(\harder than necessary , w ell-kno wn, a minor t wist to an

ld

tec hnique"), and y

ur

references (\I published/read/disco v ered something more in teresting/general/dicult a few y ears ago"). The b est defense against this t yp e

f

question is to b e prepared, b e p

lite,

and a v

id

getting in v

lv

ed in a length y exc hange. It is sometimes advisable to short-circuit suc h questions b y

ering

to tak e them

-line,

since a

ne-on-one

discussion is lik ely to b e less acrimonious than a public

ne.

Do not b e afraid to answ er \I don't kno w" to some questions. Don't, ho w ev er, confuse \I don't kno w" with \it is not kno wn". Only sa y the latter when y

u

are sure that the question is

p

en. If y

u

ha v e to sa y \I don't kno w", sa y it with assurance rather than meekness, and don't b e afraid to approac h y

ur

questioner after the talk | y

u

migh t learn something. 11

SLIDE 12 5 Ac kno wledgem en ts The author w

uld

lik e to thank Virginia Holt for her commen ts

n

sev eral drafts

f

this man uscript, and a n um b er

f

colleagues and studen ts, including P aul Ho w ard, Pierre Les- canne, Thomas Sc h w en tic k, and Kim Y ates, for their useful suggestions. References [1] L. Lamp

rt.

L a T E X: A Do cument Pr ep ar ation System. Addison-W esley , 1986. [2] I. P arb erry . Ho w to presen t a pap er in theoretical computer science: A sp eak er's guide for studen ts. SIGA CT News, 19(2):42{47, 1988. [3] I. P arb erry . Ho w to presen t a pap er in theoretical computer science: A sp eak er's guide for studen ts. Bul letin

f

the EA TCS, (37):344{349, 1989. 12