E55 - Technology Forecasting

in the Telecoms, Media and Hi-Tech industries

Session 1

Antonio Schuh
September 3rd, 2010
Agenda

 Introductions

 Course overview

 Limits of forecasting

 IBM’s Watson: “I think there is a world market for

about five computers”

1
A. Schuh
A note about me
 Experience  Education
 Director of Partnership  PhD drop-out: Innovation
Development at Telefónica. Management, University of
Previously, Business Development Manchester
for Content at Corporate Center;
Business Development at  Master in Management of
Telefónica Latin America (TV Communication Firms, Annenberg
launch) School - University of Southern
California. Graduate education in
 Former strategy consultant in the Institutional Economics and
“TIME” industries (Telecom, Management at PUCRS and
Media, Information & UFRGS, Brazil
Entertainment) -- Oliver Wyman
 BA, Advertising & Mass
and Arthur D. Little
Communication (UFRGS) and BSc,
 Junior executive; auditor; Computer Science (PUCRS)
computer programmer at RBS
 At the IE since 2007
(Media) and Dana Corp
2
A. Schuh
A note about me
 Forecasting
 RBS: demand planning of newspaper sales
 Non-obvious
 Consulting:
 very 1st assignment, TV sets  Frequent
manufacturing  Important
 Mobile operators, web services,
antennas…
 Telefónica: IPTV, digital music…
 Expectations
 Support professional positioning
 Learn
 Learn  Share
 From students  Grow
 Abstract from day-to-day issues
 Formalize
3
A. Schuh
Guest lecturer: Rafael Martínez Alonso
 Director at Telefónica’s Chairman Office
 Formerly
 Business Intelligence Manager at Strategy and Business
Development division
 Head of Internet and network management planning
 Wholesale Services - Planning and Business Development
manager.
 Six Sigma Champion and EFQM evaluator
 IE MBA; Telecommunications Engineer; IESE PDG
 Tutor, Entrepreneurship at IE
 Board member, Club de Amigos de la Sociedad de la Información
 Blogger: www.estratega.com; twitter @estratega
 Will be in charge of sessions 5 and 6

4
A. Schuh
You

 Introduction
IEIE“orla”
“orla”  Highlights of background
 12
12students
students  Experience in forecasting
 70%
70%engineers
engineers
 Expectations
 10
10nationalities
nationalities
 Why are you interested in
technology forecasting?
 Plans?

5
A. Schuh
Administrative matters
 Grading
 Take-home exam: 30%
 Class participation: 20%
 Student presentations, group assessment by lecturer: 40%
 Student presentations, assessment fellow students: 10%
 3 groups, 3 students each
 Define by no later than Sept 16th

 Attendance: non mandatory

 Contacting lecturer:
 No IE office hours, after class OK
 E-mail ([email protected])
 Phone: 606 111 420 – please use sparingly

6
A. Schuh
Agenda

 Introductions

 Course overview

 Limits of forecasting

 IBM’s Watson: “I think there is a world market for

about five computers”

7
A. Schuh
Course objectives

 to provide students with basic

instrumental knowledge on innovation
 to support students in understanding the
scope of forecasting techniques and
concepts in enhancing decision-making
capabilities in the management of
innovation
Better
 to enable students to use basic tools that
decisions
help analyse and forecast the impact of
the introduction of new products and
new technologies at firm, industry and
society levels.
 to enhance students’ appreciation of the
importance of forecasting

8
A. Schuh
 What type of decision to improve?

Market
Success False negative: Positive hit
missed
opportunity
Results

Negative hit False positive:

Market lost investment
Failure

Failure Success

Forecast
9
A. Schuh
Definition of technology forecasting

•A systematic assessment
•Of technological developments/new products
•Leading to a quantified prediction
•Impact in the market
• Timing
• More than 1 year
• Less than 10 years

10
A. Schuh
 Non-obvious: major differences in adoption
rates
% of US households

Source:W. Michael Cox, Federal Reserve Bank of Dallas (apud Ulrich, Karl, “Reducing Risk in New-Category Innovation”)

11
A. Schuh
Necessary and frequently done

•Required
•New venture creation “Forecasting in business is
like sex in society: we have
•Acquisitions
to have it, everyone is
•New product development doing it in one way or
another, but nobody is sure
•Business development
he is doing it the best way.”
•Time consuming G W Plossl
Manufacturing Control, The
•About 50% of all time Last Frontier for Profits
dedicated to a business plan
•Major source of discussions

12
A. Schuh
With room to be improved

• Kahn, 2002: 3 surveys of marketing and • Accuracy

new product development executives • Average absolute error
• Covering all types of product innovation: at the end of forecast
product/cost improvement; line/market period
extension; new (to company or to the • Average: 58%
world) • More accurate in cost
• Conclusions improvement (71%)
• 43% of companies don’t use market • Lower accuracy in
research as an input for forecasting new-to-the-world:
• Multiple techniques (2-4) 40%
• Most common: Qualitative or basic
quantitative
•Jury of executive opinion (44%)
•Sales force composite (39%)
•Analogies (30%)
•Trend analysis (19%)
Source: Kenneth B. Kahn . An exploratory Investigation of new product forecasting practices, The Journal of Product Innovation Management 19 (2002) 133—143

13
A. Schuh
High impact …

•Example, quick and dirty 700 NPV:

business plan for a new 632
+100
business 600

• Share: from 80% to 40% 500

479 NPV:
• Opex: fixed component, +18
400 400
10; variable, 55% of 363
revenues 300 303
275
208
• Capex: fixed, set-up stage 230 232
200 158
(first 3 years) 60 units 132 174 176
100 133
77
• Hurdle rate: 9% 100 58 101 NPV:
0 14
10
7 - 40
0 1
•3 scenarios: base, plus/minus 0
1 2 3 4 5 6 7 8 9 10
58%
14
A. Schuh
Course approach

• Forecasting • Innovation strategy

• Focus: number • Focus: field-of-play
• Statistics, models • Competitor dynamics
• Accuracy rate • Protection of
competitive advantage
Optimized
decisions

• Historically informed
• Previous experiences
• Do’s and don’ts

Need to go beyond forecasting techniques to improve ability to

estimate adoption of products based on new technologies
15
A. Schuh
Plan
1. What can we forecast? IBM’s forecast: ‘I think there is a world market for about
five computers”.
2. Review of key concepts in Innovation
Management.

3. Diffusion of innovation. digital music: iPod vs Rio.

4. Short-term forecasting. Forecasting penetration of Pay TV.
5. Industry-level and mid-term technology VHS/Betamax versus HD-DVD/Blu-Ray.
forecasting.
6. Foresight. Fixed and mobile telephony.
7. Choosing methods.

8. Forecasting and technological innovation Google strategy in “over-the-top” video and telecoms
strategy. (YouTube and Nexus 1).
9. Dealing with failure.

10, 11 & 12 Student presentations (1): Videophone,

Telemedicine; 3D TV

16
A. Schuh
Agenda

 Introductions

 Course overview

 Limits of forecasting

 IBM’s Watson: “I think there is a world market for

about five computers”

17
A. Schuh
Can we forecast at all?

NO YES

-Epistemological -Real-world stories

-Psycological biases of success
-Practical constraints
- Validation: bad forecasts

18
A. Schuh
Epistemological reasons why forecasting is
impossible
 Karl Popper  Poincaré
 Originally a criticism of Marxist  Precursor of chaos theory
theory of history
 Interplay of elements
 Technology is fundamentally
 The longer the forecast, the
unpredictable. Iterated
higher the knowledge of all
expectations: if you expect X in
processes associated in the
the future, you are already know
system
about X
 Butterfly effects – impact of
 Example (Nassim Nicholas Taleb):
small changes in in initial
stone age forecaster, report to
condition in complex systems
tribe chief -- Wheel
make difficult to predict
 If doesn’t forecast, useless outcomes
 If does, has invented it
 We don’t know what we don’t
know

19
A. Schuh
Epistemological reasons in practice

 Need to be humble – innovation is inevitably intertwined

to uncertainty

 Limit time horizon

 Understand system

 Be prepared for “black swans”

 However, incompleteness doesn’t prevent identifying

likely outcomes

20
A. Schuh
Psychological biases

 Tunneling: tendency to neglect  Implications

source of uncertainty outside  Optimism bias
the one already focusing.
 Risk of trojan
horses
 Herding: tendency to follow  Bandwagon –
socially validated behaviour. by the time you
join is too late
 Anchoring: exposure to a
reference point affects  Awareness of
estimates, if it is random and biases can reduce
we know it. their magnitude

21
A. Schuh
Practical constraints

 Data & Analysis

 Data limitations: History (precursors/analogs); Environment - Changing
social and demographic trends

Analytic limitations examples

Competition comes from Videogames didn’t come from Mattel; electrical

unexpected sources lights not from gas light companies
Ultimate users are unforeseen Airconditioning: devised for factories

Ultimate form of the product is not Large TVs: 1970s, Kloss front projection
anticipated
Customer perception of benefits is Microwave ovens, launched in 1948, take-off in the
unclear 1970s
Cost structures don’t change as Gallium arsenide chips
forecast
Technology improvements failed to G
materialize

22
A. Schuh
Causes of bad predictions

 Main ones
 Hubris
 Optimism (on new technology or on prospects of existing
technology)
 Poor analysis
 Self-serving (strategic signaling)

 Remedies are possible

 Focus on situations where forecasting can add value
 clear causal relationships
 clear object (a single product/market, not an ecosystem)
 Improve forecaster’s competence
 Domain-specific (technology/market)
 Techniques

23
A. Schuh
No shortage of bad predictions
• "There is no reason anyone would want a computer in their
home."
-Ken Olson, 1977

• "I don't know what use any one could find for a machine that
would make copies of documents. It certainly couldn't be a
feasible business by itself."
- Head of IBM to Chester Carlson, 1940

• “This ‘telephone’ has too many shortcomings to be seriously

considered as a means of communication. The device is
inherently of no value to us”.
Western Union memo, 1875

Source:XXX

24
A. Schuh
And some great ones

David Sarnoff (1916)

"I have in mind a plan of development which would make radio a
'household utility' in the same sense as the piano or phonograph.
The idea is to bring music into the house by wireless.”
(management answer: “who would pay for a message sent to
nobody in particular?”; radio set to cost US$ 1460 in current prices.)

25
A. Schuh
And some great ones

Microsoft (Albuquerque Group, 1978)

“A computer on every desk and in every home running Microsoft
software”

26
A. Schuh
And some great ones

Gerald Whent and Chris Gent

Mobile (1984) & SMS (1991)

27
A. Schuh
Agenda

 Introductions

 Course overview

 Limits of forecasting

 IBM’s Watson: “I think there is a world market for

about five computers”

28
A. Schuh
Thomas J. Watson Sr

 Chairman of the Board of

IBM from 1914-1952
 IBM
 World leader in data
processing
 One of the world’s
most internationalized
companies
 15,000 employees, over
US$ 1 bi in revenues at
current prices
Thomas J Watson Sr and a IBM 701 in 1952
 In 1943 said: “I think
there is a world market
for about five computers”

29
A. Schuh
A folly or a great forecast?

• And why…

• Best answer wins a

ticket for
TEDxMadrid on
Saturday

30
A. Schuh
Folly or great? Great!

 The computer as we know it hadn’t been

invented

 Watson’s prediction held for 10 years

 Didn’t prevent IBM to capture world

leadership in computers for more than 20
years
Source: Gordon Bell, ACM 97, “The Folly Laws of Predictions”.

31
A. Schuh
Computers in 1943: Harvard Mark I

32
A. Schuh
Computers in the early 40s

 Mainly analog
 Electromechanical
 I/O via punch cards
 Storage by relays
 Decimal
 Programmed by hard-wiring/switches or Special
purpose – not programmed at all
 Each with its own architecture

33
A. Schuh
Computers in the early 40s

 Hand made, built to order

 Primarily at Universities
 Harvard: Mark I (H. Aiken + IBM), 1939-43
 Iowa State: Atanasoff–Berry Computer, 1939-42
 U of Pennsylvania: ENIAC, 1943-1946
 MIT: Whirlwind (1944-51)
 ERA: Atlas I, (1947-50)

34
A. Schuh
Features of computers in the early 40s

 Big
 Aiken’s Mark I – over 750 thousand components
 16 meters long
 5 tons – (a featherweight: ENIAC, 30 tons)

 Slow
 Mark I, 5 seconds to multiply 8-digit number
 ENIAC, 1950 operations/second

 Unreliable
 Expensive
 Mark I: US$ 25 M grant by IBM to Harvard to develop it
 UNIVAC (1951): US$ 8 M at current prices
35
A. Schuh
Computers applications in the early 40s

 To compute
 Cryptoanalytic
 Ballistic tables
 Project Manhattan

 Clients/sources of funding
 Navy
 Army
 Atomic Energy Commission

36
A. Schuh
Watson’s prediction held for 10 years

 1st commercial computer: Remington Rand UNIVAC,

1951 – 46 units sold (6 before 1953)
 1st used by a corporation: GE, 1954 (payroll)

 Other vendors with very limited batches

37
A. Schuh
Didn’t prevent IBM to capture world leadership in
computers for more than 20 years …
 Pioneer: Remington Rand  Major changes
 Big + Vision: leader in  Technology: digital,
computers electronic, high reliability,
 Hired Gen. Leslie Groves to memory…
lead development in 1947  Production: importance of
 Acquired Eckert+Mauchly in scale
1950, ERA in 1952  Marketing: customer
 UNIVAC PR gimmicks – e.g. awareness amongst large
Eisenhower election businesses for new uses
 Merged with Sperry in  Complementary assets:
1955(bomb sights, etc) programming
 3-year lead in commercial
computers

38
A. Schuh
…Indeed, it arguably helped it. (1/2)

 How IBM did it

 Kept and eye
 Tech developments (Government funded: Defence Calculator,
1952)
 Inroads by Univac on the business market
 Move in as market showed signs of potential
 Leveraged capabilities in production, marketing, customer service
 Innovated
 Modular approach/peripherals
 Business model: leasing
 Software (Fortan, 1957; “user groups”)
 Accelerated: Model 650 (1954); 1401 (transistor, 1960); S/370

39
A. Schuh
…Indeed, it arguably helped it. (2/2)

 Remington Rand
 Likely overinvested: too much too soon
 R&D externalities and leakage
 Demand creation
 Organization issues reduced ability to respond timely
 Prepared for a strong position in what turned to be a low-growth
segment (Military)
 End result
 By late 50s, IBM had 125 installations vs UNIVAC’s 46
 4x P/E
 By 1963: IBM with ~70% market share and “IBM and the seven
dwarfs”

40
A. Schuh
To sum it up

 Forecasting is critical to identify “strategic windows” of entry

 Proper forecasting was instrumental in capturing leading role

 Must be revised frequently as events change – it’s a dynamic game.

 In computers, major changes since 1943
 Technology
 Applications
 Customer groups
 Moving target – “computer” didn’t mean the same 10 years later

 Forecasting must be coupled to strategy – in IBM’s case, from fast

follower to innovator
 Kept options open
 Once combination of market and technology indicated higher
potential, moved decisively

41
A. Schuh
Another Watson’s view

“There doesn't seem to be any real limit to the

growth of the computer industry.”
- Thomas J. Watson, Jr.
Chairman of the Board of IBM, 1968
42
A. Schuh