presentation tips
DESCRIPTION
A useful guide on how to prepare, design, and deliver effective and communicative presentationsTRANSCRIPT
PresentationtipsMarco D’Ambros
95% of presentations SUCK
“”—Guy Kawasaky
—Guy Kawasaky
OK, maybe I’m exaggerating.“”It is actually 99%
99%
1%
suck
don’t suck
Examples?
Ok, but be prepared for what follows
Death by Powerpoint
Preparation Design Delivery
Preparation Design Delivery
Preparing a 30-slide presentation takes 36-90 hours
“
”—Nancy Duarte
Start with the goal
What is the message?
Know youraudience
Simplify
to the essential(but not more)
—John Maeda
Simplicity is aboutsubtracting the obvious,and adding the meaningful
“
”
—John Maeda
More appears like lesssimply moving it far, far away
“”
Get alone
Multitasking, when it comes to paying attention, is a myth
“
We are biologically incapable of processing attention-rich inputs simultaneously”
—Dr. Medina
People who are interrupted:
50%Make
moreerrors
50%Take
longer tocomplete a task
Being always onlineis being always distracted and
unproductive
You need time off the grid to prepare
Go analog
Use post-it
Use whiteboard
Brainstorming andmind mapping
If you are stuck go for a walk or a run...just move!
If you are stuck go for a walk or a run...just move!
why?
How the brain works in 12 rules
Exercise boost brain power
Rule #1
By Dr. Medina
Examples of anti-brain environmentsaccording to Dr. Medina
Lecture hall
Classroom
Office
Exercise is not just good for general health, it actuallyimproves cognition
Exercise increases oxygen flow into the brain, which reduces brain-bound free radicals [...] an increase in oxygen is always accompanied by an uptick in mental sharpness.
“
Exercise acts directly on the molecular machinery of the brain itself. It increases neurons’ creation, survival, and resistance”
—Dr. Medina
Even more benefits!
• Reduces depression
• Treats dementia
• Improves reasoning
• Improves long-term memory
• Improve fluid intelligence
• Helps you solve problems
• and more...
If you are stuck go for a walk or a run...just move!
Your audience will haveto seat and listen
think from their perspective
Create a scalable structure
5min
presentation
solutionkey point
key point
key point
15min
presentation
solutionkey point
explanation
explanation
explanation
45min
presentation
key pointexplanation
detail
detail
detail
Create thestory
We do not pay attention to boring things
Rule #4
If keeping someone’s attention in a lecture was a business, it would have an 80% failure rate.
“
”— Dr. John Medina
Good stories are:
simple
Concre
te
Credible
Emotional
• story
• exposition
• conflict & climax
• resolution
How should the story be?
exposition
conflict & climax
resolution
• pick another perspective
Take another perspective
In the beginner’s mind there are many possibilities, in the expert’s mind there are few
—Shunryu Suzuki
Adopt a beginner’s mind
“
”
Preparation Design Delivery
Common mistakes
People tend to put every word they are going to say on their PowerPoint slides. Although this eliminates the need to memorize your
talk, ultimately this makes your slides crowded, wordy, and boring. You will loss your
audience’s attention before you even reach the bottom of your ...
1. T E L E P R O M P T I N G
Slide from Don McMillan, “Life After Death by PowerPoint”: http://bit.ly/aYxegN
2. Spelling mistakes
Many people do not run spel cheek before there presentation
BIG MISTAK!!! Nothing makes you lok stupder than speling erors
Slide from Don McMillan, “Life After Death by PowerPoint”: http://bit.ly/aYxegN
3. Bullet pointing
Slide from Don McMillan, “Life After Death by PowerPoint”: http://bit.ly/aYxegN
• Avoid• Excesive• Buller-Pointing• Only• Bullet• Key• Points• Too• Many
• Bullet-Points• And• Your• Key• Messages• Will• NOT• Stand• Out
4. Too many levels
• What is worst
• Too many bullet point levels are shown
• Type size gets smaller and smaller
• Until it is utterly unreadable
• Even for audiences in the 4th row
• So you better have just one bulletpoint level
• Better yet, forget about bullets (bullets, not guns, kill people. Don’t you know?)
• Use them sparingly
• There are many other ways of detailing your ideas!
Slide from Don McMillan, “Life After Death by PowerPoint”: http://bit.ly/aYxegN
5. Color schemes gone wrong
• Distraction• Confusion • Headache• Nausea• Vomiting• Loss of bladder control
schemesbadcolor
canlead to...
Slide from Don McMillan, “Life After Death by PowerPoint”: http://bit.ly/aYxegN
6. Stick to the default template
7. Use things people won’t understand
In the fist time interval t1, we have a total of 4 changes,and the change frequencies of the files (i.e., their probabilityof change) are pA = 2
4 , pB = 14 , pC = 1
4 . The entropy in t1is therefore H = �(0.5 ⇥ log2 0.5 + 0.25 ⇥ log2 0.25 + 0.25 ⇥log2 0.25) = 1. In the second time interval t2 the entropy ishigher: H = �( 2
7 ⇥log227 + 1
7 ⇥log217 + 4
7 ⇥log247 ) = 1.378. In
addition, we take into account the amount of change per fileby measuring the size of change (lines added and/or deleted).Hassan also defined the Adaptive Sizing Entropy as:
H 0 = �nX
k=1
pk ⇥ logn̄ pk (2)
where n is the number of files in the system and n̄ is thenumber of recently modified files. To compute the set of re-cently modified files we use previous periods (e.g., modifiedin the last six time intervals), whereas also time can be used(e.g., modified in the last six months). To use the entropyof code change as bug predictor, Hassan defined the Historyof Complexity Metric (HCM) of a file j as
HCM{a,..,b}(j) =X
i2{a,..,b}
HCPFi(j) (3)
where {a, .., b} is a set of evolution periods and HCPF isdefined as:
HCPFi(j) =
⇢cij ⇥H 0
i, j ⇤ Fi
0, otherwise(4)
where i is a period with entropy H 0i, Fi is the set of files mod-
ified in the period i and j is a file belonging to Fi. Accordingto the definition of cij , there are three types of HCM :
1. (1) cij = 1, every file modified in the considered periodi gets the entropy of the system in the considered timeinterval. This defines approach HCM.
2. (2) cij = pj , each modified file gets the entropy of thesystem weighted with the probability of the file beingmodified, defining WHCM.
3. (3) cij = 1|Fi|
the entropy is evenly distributed to allthe files modified in the i period. We do not use thisdefinition since Hassan showed that it performs lesswell than the other.
Concerning the periods used for computing the History ofComplexity Metric, we use fixed time intervals of two weeks.
Variants. We define three further variants based onHCM, with an additional weight for periods in the past.In EDHCM (Exponentially Decayed HCM) , entropies forearlier periods of time, i.e., earlier modifications, have theircontribution reduced exponentially over time, modelling anexponential decay model. EDHCM was introduced by Has-san. Similarly, LDHCM (Linearly Decayed) and LGDHCM(LoGarithmically decayed), have their contributions reducedover time in a respectively linear and logarithmic fashion.Both are novel. The definition of the variants follow:
EDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
e�1⇥(|{a,..,b}|�i) (5)
LDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
�2⇤(|{a,..,b}|+1�i) (6)
LGDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
�3⇤ln(|{a,..,b}|+1.01�i)(7)
where �1, �2 and �3 are the decay factors.
4.5 Churn of Source Code MetricsUsing churn of source code metrics to predict post release
defects is a novel approach that we propose. The idea isthat higher-level metrics may better model code churn thansimple metrics like addition and deletion of lines of code.The idea is to sample the history of the source code everytwo weeks and to compute the deltas of source code metricsfor each consecutive pair of samples. To do so, we first haveto create “delta matrices”, one per each source code metrics,by performing the following steps:
1. We check out one version of the system from the CVSrepository for each two weeks time interval, and parsethe source code of every version and, for each class,compute the CK+OO source code metrics.
2. For each source code metric, we create a matrix wherethe rows are the classes, the columns are the sampledversions, and each cell is the value of the metric forthe given class at the given version. If a class does notexist in a version, we indicate that by using a defaultvalue of -1. We only consider the classes which existat release x for the prediction.
3. We generate a matrix of deltas, where each cell is theabsolute value of the di�erence between the values ofa metric for a class in two subsequent version. If theclass does not exist in one of the two versions (or bothof them), i.e., at least one value is -1, then we use thedefault value of -1 also for the delta.
10Class Foo
Class Bar 42
Class Bas -1
2 weeks
Release X
50
32
50
22
70
22
48
40
10 15
Version from
1.1.2005
Version from
15.1.2005
Version from
29.1.2005
Time
10
-1
0
10
5
Figure 5: Computing metrics deltas from versionsof a software system sampled every two weeks.
Figure 5 shows an example of deltas matrix computa-tion for three classes. The numbers in the squares arethe values of the metrics in the corresponding versionsof the system, while the numbers in the circles are thedeltas.
Once we have computed the deltas matrices for each sourcecode metric, we compute the churn measures as follows:
CHU(i) =CX
j=1
⇢0, deltas(i, j) = �1PCHU(i, j), otherwise
(8)
PCHU(i, j) = deltas(i, j) (9)
where i is the index of a row in the deltas matrix (corre-sponding to a class), C is the number of columns of the ma-trix (corresponding to the number of samples considered),
In the fist time interval t1, we have a total of 4 changes,and the change frequencies of the files (i.e., their probabilityof change) are pA = 2
4 , pB = 14 , pC = 1
4 . The entropy in t1is therefore H = �(0.5 ⇥ log2 0.5 + 0.25 ⇥ log2 0.25 + 0.25 ⇥log2 0.25) = 1. In the second time interval t2 the entropy ishigher: H = �( 2
7 ⇥log227 + 1
7 ⇥log217 + 4
7 ⇥log247 ) = 1.378. In
addition, we take into account the amount of change per fileby measuring the size of change (lines added and/or deleted).Hassan also defined the Adaptive Sizing Entropy as:
H 0 = �nX
k=1
pk ⇥ logn̄ pk (2)
where n is the number of files in the system and n̄ is thenumber of recently modified files. To compute the set of re-cently modified files we use previous periods (e.g., modifiedin the last six time intervals), whereas also time can be used(e.g., modified in the last six months). To use the entropyof code change as bug predictor, Hassan defined the Historyof Complexity Metric (HCM) of a file j as
HCM{a,..,b}(j) =X
i2{a,..,b}
HCPFi(j) (3)
where {a, .., b} is a set of evolution periods and HCPF isdefined as:
HCPFi(j) =
⇢cij ⇥H 0
i, j ⇤ Fi
0, otherwise(4)
where i is a period with entropy H 0i, Fi is the set of files mod-
ified in the period i and j is a file belonging to Fi. Accordingto the definition of cij , there are three types of HCM :
1. (1) cij = 1, every file modified in the considered periodi gets the entropy of the system in the considered timeinterval. This defines approach HCM.
2. (2) cij = pj , each modified file gets the entropy of thesystem weighted with the probability of the file beingmodified, defining WHCM.
3. (3) cij = 1|Fi|
the entropy is evenly distributed to allthe files modified in the i period. We do not use thisdefinition since Hassan showed that it performs lesswell than the other.
Concerning the periods used for computing the History ofComplexity Metric, we use fixed time intervals of two weeks.
Variants. We define three further variants based onHCM, with an additional weight for periods in the past.In EDHCM (Exponentially Decayed HCM) , entropies forearlier periods of time, i.e., earlier modifications, have theircontribution reduced exponentially over time, modelling anexponential decay model. EDHCM was introduced by Has-san. Similarly, LDHCM (Linearly Decayed) and LGDHCM(LoGarithmically decayed), have their contributions reducedover time in a respectively linear and logarithmic fashion.Both are novel. The definition of the variants follow:
EDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
e�1⇥(|{a,..,b}|�i) (5)
LDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
�2⇤(|{a,..,b}|+1�i) (6)
LGDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
�3⇤ln(|{a,..,b}|+1.01�i)(7)
where �1, �2 and �3 are the decay factors.
4.5 Churn of Source Code MetricsUsing churn of source code metrics to predict post release
defects is a novel approach that we propose. The idea isthat higher-level metrics may better model code churn thansimple metrics like addition and deletion of lines of code.The idea is to sample the history of the source code everytwo weeks and to compute the deltas of source code metricsfor each consecutive pair of samples. To do so, we first haveto create “delta matrices”, one per each source code metrics,by performing the following steps:
1. We check out one version of the system from the CVSrepository for each two weeks time interval, and parsethe source code of every version and, for each class,compute the CK+OO source code metrics.
2. For each source code metric, we create a matrix wherethe rows are the classes, the columns are the sampledversions, and each cell is the value of the metric forthe given class at the given version. If a class does notexist in a version, we indicate that by using a defaultvalue of -1. We only consider the classes which existat release x for the prediction.
3. We generate a matrix of deltas, where each cell is theabsolute value of the di�erence between the values ofa metric for a class in two subsequent version. If theclass does not exist in one of the two versions (or bothof them), i.e., at least one value is -1, then we use thedefault value of -1 also for the delta.
10Class Foo
Class Bar 42
Class Bas -1
2 weeks
Release X
50
32
50
22
70
22
48
40
10 15
Version from
1.1.2005
Version from
15.1.2005
Version from
29.1.2005
Time
10
-1
0
10
5
Figure 5: Computing metrics deltas from versionsof a software system sampled every two weeks.
Figure 5 shows an example of deltas matrix computa-tion for three classes. The numbers in the squares arethe values of the metrics in the corresponding versionsof the system, while the numbers in the circles are thedeltas.
Once we have computed the deltas matrices for each sourcecode metric, we compute the churn measures as follows:
CHU(i) =CX
j=1
⇢0, deltas(i, j) = �1PCHU(i, j), otherwise
(8)
PCHU(i, j) = deltas(i, j) (9)
where i is the index of a row in the deltas matrix (corre-sponding to a class), C is the number of columns of the ma-trix (corresponding to the number of samples considered),
In the fist time interval t1, we have a total of 4 changes,and the change frequencies of the files (i.e., their probabilityof change) are pA = 2
4 , pB = 14 , pC = 1
4 . The entropy in t1is therefore H = �(0.5 ⇥ log2 0.5 + 0.25 ⇥ log2 0.25 + 0.25 ⇥log2 0.25) = 1. In the second time interval t2 the entropy ishigher: H = �( 2
7 ⇥log227 + 1
7 ⇥log217 + 4
7 ⇥log247 ) = 1.378. In
addition, we take into account the amount of change per fileby measuring the size of change (lines added and/or deleted).Hassan also defined the Adaptive Sizing Entropy as:
H 0 = �nX
k=1
pk ⇥ logn̄ pk (2)
where n is the number of files in the system and n̄ is thenumber of recently modified files. To compute the set of re-cently modified files we use previous periods (e.g., modifiedin the last six time intervals), whereas also time can be used(e.g., modified in the last six months). To use the entropyof code change as bug predictor, Hassan defined the Historyof Complexity Metric (HCM) of a file j as
HCM{a,..,b}(j) =X
i2{a,..,b}
HCPFi(j) (3)
where {a, .., b} is a set of evolution periods and HCPF isdefined as:
HCPFi(j) =
⇢cij ⇥H 0
i, j ⇤ Fi
0, otherwise(4)
where i is a period with entropy H 0i, Fi is the set of files mod-
ified in the period i and j is a file belonging to Fi. Accordingto the definition of cij , there are three types of HCM :
1. (1) cij = 1, every file modified in the considered periodi gets the entropy of the system in the considered timeinterval. This defines approach HCM.
2. (2) cij = pj , each modified file gets the entropy of thesystem weighted with the probability of the file beingmodified, defining WHCM.
3. (3) cij = 1|Fi|
the entropy is evenly distributed to allthe files modified in the i period. We do not use thisdefinition since Hassan showed that it performs lesswell than the other.
Concerning the periods used for computing the History ofComplexity Metric, we use fixed time intervals of two weeks.
Variants. We define three further variants based onHCM, with an additional weight for periods in the past.In EDHCM (Exponentially Decayed HCM) , entropies forearlier periods of time, i.e., earlier modifications, have theircontribution reduced exponentially over time, modelling anexponential decay model. EDHCM was introduced by Has-san. Similarly, LDHCM (Linearly Decayed) and LGDHCM(LoGarithmically decayed), have their contributions reducedover time in a respectively linear and logarithmic fashion.Both are novel. The definition of the variants follow:
EDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
e�1⇥(|{a,..,b}|�i) (5)
LDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
�2⇤(|{a,..,b}|+1�i) (6)
LGDHCM{a,..,b}(j) =P
i2{a,..,b}HCPFi(j)
�3⇤ln(|{a,..,b}|+1.01�i)(7)
where �1, �2 and �3 are the decay factors.
4.5 Churn of Source Code MetricsUsing churn of source code metrics to predict post release
defects is a novel approach that we propose. The idea isthat higher-level metrics may better model code churn thansimple metrics like addition and deletion of lines of code.The idea is to sample the history of the source code everytwo weeks and to compute the deltas of source code metricsfor each consecutive pair of samples. To do so, we first haveto create “delta matrices”, one per each source code metrics,by performing the following steps:
1. We check out one version of the system from the CVSrepository for each two weeks time interval, and parsethe source code of every version and, for each class,compute the CK+OO source code metrics.
2. For each source code metric, we create a matrix wherethe rows are the classes, the columns are the sampledversions, and each cell is the value of the metric forthe given class at the given version. If a class does notexist in a version, we indicate that by using a defaultvalue of -1. We only consider the classes which existat release x for the prediction.
3. We generate a matrix of deltas, where each cell is theabsolute value of the di�erence between the values ofa metric for a class in two subsequent version. If theclass does not exist in one of the two versions (or bothof them), i.e., at least one value is -1, then we use thedefault value of -1 also for the delta.
10Class Foo
Class Bar 42
Class Bas -1
2 weeks
Release X
50
32
50
22
70
22
48
40
10 15
Version from
1.1.2005
Version from
15.1.2005
Version from
29.1.2005
Time
10
-1
0
10
5
Figure 5: Computing metrics deltas from versionsof a software system sampled every two weeks.
Figure 5 shows an example of deltas matrix computa-tion for three classes. The numbers in the squares arethe values of the metrics in the corresponding versionsof the system, while the numbers in the circles are thedeltas.
Once we have computed the deltas matrices for each sourcecode metric, we compute the churn measures as follows:
CHU(i) =CX
j=1
⇢0, deltas(i, j) = �1PCHU(i, j), otherwise
(8)
PCHU(i, j) = deltas(i, j) (9)
where i is the index of a row in the deltas matrix (corre-sponding to a class), C is the number of columns of the ma-trix (corresponding to the number of samples considered),
Design the zen way
simplicity
clarity
uncluttered
Design right-brain slides
Design rightbrain slides
Be visual
Vision trumps all other senses
Rule #10
We have a better recall
for visual information
IRSYMCAWTFIBMKGBFBI
we are wired to pattern”—Dr. Medina“
IRSYMCAWTFIBMKGBFBI
we are wired to pattern”—Dr. Medina“
Visual information are easier to remember
Oral
Visual
Oral &Visual
10%
35%
65%
3x6x
Source: Najjar, LJ (1998) Principles of educational multimedia user interface design (via Brain Rules by John Medina, 2008)
decoratedon’t
slides
communicationadd
values
90freshwater
%
of the
in the world is
Slide from Garr Reynolds: http://www.slideshare.net/garr/sample-slides-by-garr-reynolds
90of the
in our
%
planet isfreshwater
Inspired by www.slideshare.net/garr/sample-slides-by-garr-reynolds
ice
Source: SCAR
iceInspired by www.slideshare.net/garr/sample-slides-by-garr-reynolds
90 of the ice in our planet %
Antarcticais in
Inspired by www.slideshare.net/garr/sample-slides-by-garr-reynolds
Antarctic
of the world’s freshwater80
Source: SCAR
%
is ice in the
Slide from Garr Reynolds: http://www.slideshare.net/garr/sample-slides-by-garr-reynolds
80 of our planet’s freshwater%
Antarcticis ice in the
Inspired by www.slideshare.net/garr/sample-slides-by-garr-reynolds
Eat onlyuntil80% full
Bring your message visually
Eat onlyuntil80% full
Bring your message visually
foxy sad happy
angryundecidedsmiley
Use faces
Find beauty
Slide from Christina Quick : http://www.slideshare.net/ChrisQuick/new-rules-for-power-point-presentations
Dramatize
Slide from Christina Quick : http://www.slideshare.net/ChrisQuick/new-rules-for-power-point-presentations
2% of theworld
wealth50%owns
of the
Use metaphorical image
Slide from Christina Quick : http://www.slideshare.net/ChrisQuick/new-rules-for-power-point-presentations
The poorest 50% of the world
owns 1% of the wealth
Slide from Christina Quick : http://www.slideshare.net/ChrisQuick/new-rules-for-power-point-presentations
66% of Americansare obese or overweight.
All adults (66%)
Women 65 million (62%)
Men 69 million (71%)
134 million
OECD Factbook 2007
Be provocative
Slide from Garr Reynolds: http://www.slideshare.net/garr/sample-slides-by-garr-reynolds
Text and fonts
No more than 2-3 types
Text and fonts
use color and size to build contrast
Text and fonts
Blinking, sparkling, or twirling text is just not cool
Rotation can make the slide more
interesting, but don’t overdo it
Text and fonts
Reduce to the maxText is redundant
People cannot listen andread at the same time
Simplicity means the achievement of maximum effect with minimum means
“
—Dr. Koichi Kawana”
Use quotes
Design principles
11 po
int
per
slid
e
Contrast
Repetition
Alignment
Proximity
Design principles
Contrast
R e p e t i t i o n
Repetition examples
Repetition of design elements gives a
cohesive look
Slide from Jesse Desjardins: http://www.slideshare.net/jessedee/steal-this-presentation-5038209
Alignment
Proximity
no templatesor
the following one
Rule of third
Use color properly
Black: is all about elegance
White: is about trust
Apply a contrasting color palette
Choose with tools
http://kuler.adobe.com
www.colorschemer.com
Do not forget that 7-10% ofpeople are color-blind
Use images properly
Do not useUgly clip art
Small images
Watermarked images
Distorted images
Go full quality
Have a neutralbackground
Make images transparent
oruse
Where to get good images?
www.istockphoto.com (pay)
Where to get good images?
google images
Where to get good images?
sxc.hu (free)
Do not omit the credits
Image from irishfireside.com
Be consistentStick to your settings
avoid clutter: make more slides
a slide costs 0$
use empty space
This is different from...
... this
Show the data
A must see:www.ted.com/talks/hans_rosling_shows_the_best_stats_you_ve_ever_seen.html
Make the data memorable
2005 2007 2008
11M
9M
2M
World of Warcraft users worldwide
Not memorable
1.4 times the swiss population
World of Warcraft users worldwide
Facebook users
800Mil l ions
Not memorable
307M
800M
1,170M
1,333M
World most populated countries
800M
iPod capacity
5 GB
vs
1,000 songs
Break the rules, but do it sparingly
Slide from Eduardo S. de la Fuente: http://www.slideshare.net/eduardo.delafuente/the-art-of-presentation-following-the-zen-path-why
Some before/after examplesby presentation guru Garr Reynolds
before
after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
before
after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
before
after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
before
after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
before
after after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
before
after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
before
after
Slide from Garr Reynolds: www.slideshare.net/garr/sample-slides-by-garr-reynolds
Preparation Design Delivery
RehearseRehearseRehearseRehearse
RehearseRehearseRehearseRehearse
RehearseRehearseRehearseRehearse
Spend some time in the light table view
• show your passion
Show your passion
Introduce yourself
Start strong
Be confident
Keep it shortAudience attention steadily drops after 10 minutes
Minutes of class time
10 20 30 40 50
High
Low
Attention
The 10-minute rule
Source: www.brainrules.net/attention
Source: www.brainrules.net/attention
The 10-minutes rule
Attention
Minutes of class time
at each 10-minute markto regain attention
do something
emotionally relevant
End on a high note
Move away from the podium
audience
youlaptop
screen
• use the presenter view
Use a remote
Make
Make
pauses
• do not apologize
Do notapologize
Use Keynote presenter display
actually no
BRem
embe
r th
e B
key
Make good eyecontact
• keep the lights on
Keeplightson
Be prepared
Check the beamer beforehand
55%7% 38%
Body language matters
content voicebody language
Be slightly more elegant than the audience
Preparation Design Delivery
Always recap
Share your work
www.slideshare.net
speakerdeck.com
www.ted.com
Resources
www.garrreynolds.com
Brain RulesTakeaways & Quotes
from Dr. John Medina’s
What all presenters need to know
A presentation (of sorts)
by Garr Reynolds
Sample slidesHere are a few before/after slides
Garr Reynolds
http://slidesha.re/fausgs
http://slidesha.re/8Ykmry
http://slidesha.re/3mMo3c
http://slidesha.re/i8QMa
Credits
SEMINAR (I)
Alberto de Vega Eduardo S. de la Fuente
Zen Rocks by Lane Pierce
Following the ZEN path
http://slidesha.re/17P2Hh
Marco D’AmbrosComputer science researcher
Marco earned a PhD in Informatics from the University of Lugano (Switzerland), and MSc degrees from both Politecnico di Milano (Italy) and the University of Illinois at Chicago (USA).
His research interests lie in software engineering, software evolution, and software visualization. He authored more than 30 technical papers, and is the creator of several software visualization and program comprehension tools.
Marco is passionate about presentations: He distilled his experience, gained by giving more than 30 talks at international conferences, in this presentation.
www.linkedin.com/in/dambros
www.slideshare.net/marcodambros
twitter.com/marquitodambros
www.inf.usi.ch/phd/dambros/ On the Evolution ofSource Code andSoftware Defects
amzn.com/1460953568