Memory & Emotions in Data Patterns - PowerPoint Presentation

Slide1

Memory & Emotions in Data Patterns

Can You Hear the Shape of a Drum?

Patrick A. McNutt, FRSA

Visiting Fellow, Manchester Business School, UK

& Smurfit Business School, Dublin, Ireland

.

March 2018

www.patrickmcnutt.com

Follow @

tuncnunc

Note: Work in Progress Slides

1-25

Slide2

Omega Ω

Circles

&

Meso

-Data

Can You Hear the Shape of a Drum?

Patrick A. McNutt, FRSA

Visiting Fellow, Manchester Business School, UK

& Smurfit Business School, Dublin, Ireland

.

March & April

2018

www.patrickmcnutt.com

Follow @

tuncnunc

Note:

Work in Progress Slides 25-35

Slide3

Modules available

1.

Strategy & Competition

at

Manchester Business School

introduces online transaction costs and

non-cooperative game theory

2.

Business Economics

at

Smurfit Business School

introduces a winning unbeatable strategy set

3.

Masterclass

on Cognitive Business Strategy

introduces ‘thinking about thinking’

mistake-proofing strategy

www.patrickmcnutt.com

Slide4

Introducing

‘

meso

-data’

Small data …

…

meso

-data…….. Large

Data.

The

essence of

meso

-data is that

you

think you have memory but it has

you

2017 Presentations

:

http

://www.patrickmcnutt.com/news/beyond-individual-choice-meso-data-creativity

/

Fundamental

equation:

Memory + Emotions =

Meso

-Data

Algorithms

track and capture

our behavioural

patterns

2012:

Quidco

used GPS to inform you of discounts in nearby stores

Indoor Atlas

in 2018 goes beyond individual choice in the store.

Data with Memory and Emotions

goes beyond individual choice

Algorithms become

sufficiently intelligent

because

we outsource memory and

because

we betray our emotions to smart devices.

Slide5

Slide6

Hypothesis

Data patterns

mimic

behaviour but

meso

-data patterns create a

manifold

Sufficiently Intelligent Algorithms (SIALs) rely on decoding our data patterns

can SIALs

mimic

human behaviour?

YES: only if they pass the Turing Test

{

The Imitation Game

}

do SIALS create a

manifold

?

YES: only if they ‘seed’ a random event from a smaller unpredictable pattern

{

The Daily Routine

}

Slide7

Mimic

Pepper the Robot filters

habits & routines as ‘cumulatively unfolding

processes’

so as to influence demand (

smart strategy

)

Example: Pepper and banking

Manifold

Embeds

mutual

interdependence between data and individual

into

action-reaction chain of

events, and as you get closer to the transaction, in a moment in time, SIAL

resembles

an individual.

philosophically

:

SIAL

as ‘something representing something abstract’

(

J.L.Austin

)

Example

: shopping online…Action

plus time

to repeat the pattern

Slide8

Mutual interdependence

IN

a Game

onsumer

v

sial

Look at yourself in the mirror and you recognise yourself. You become self-aware. You have a conscious, a mind of your own.

With data-driven

strategy, we have a ‘creative type’, SIALs, influencing the

onsumer

via ‘friends’ or 3

rd

party machine learning to arrive at the sustainable

outcome.

SIALs that nudge behaviour towards a predictable [future sustainable] outcome generate a loss in

onsumer

sovereignty and infiltrate the

level of consciousness

of the

onsumers

taking decisions.

Slide9

The Law of Change

I

nversion

Loss

aversion refers to

a rational individual’s

preference

in avoiding losses to acquiring gains.Would you rather receive a £

5

discount or avoid a £5 surcharge

?

Change inversion

refers to a rational individual’s preference

to embrace tangible gains

(of robotic and machine

learning) while

conceding intangible losses

(of

data memory

and

sovereignty)

Slide10

Monetization of FB, WhatsApp, Instagram,

LinkedIn…

Tradable

assets

Hypothesis: With

all the information available it makes ‘who you are & what you do’ a

tradable

asset

Online rational ‘

onsumer

’ trades and exchanges personal data and personal search patterns at zero transactional cost.

The SIAL acquires the tradable asset at zero cost (no exchange value) but with a NPV that is infinite.

Example: On average: FB makes

$3 in EU and $13 in US per month from

onsumer

data

patterns

https://www.telegraph.co.uk/technology/2016/11/03/how-much-money-does-facebook-make-from-you

/

Slide11

Are You Ready to Betray Your Emotions?

Preparing for the Future

(via Experience & Belief

)

Consider your response to a few simple behavioral experiments…….

Slide12

First Example

SCENARIO:

As predicted you are in your favourite Costa having a coffee. You sense you are being watched. You notice Mr G wearing the new google digital glasses. He is looking at you and a red dot is on.

How do you react?

A. You think: ‘I should go to the bathroom and freshen up my look’

B. You approach Mr G and introduce yourself.

C. You think: ‘There ought to be a law against it’.

D. You share your experience on social media.

Slide13

Second example

SCENARIO

: After listening to a game theorist, you begin to realise that your favourite loyalty card provider has been capturing your daily routine and buying habits and trading your data with third parties.

How do you react?

A: You are amazed at the advances in technology and think about your data as a tradable asset.

B:

You are

interested and take a course in data analysis and gaming.

C: You are nervous about the privacy issues and become more circumspect about your habits and patterns

D

. You share your experience on social media

.

Slide14

Third example

SCENARIO:

Preparing for the BBQ, you are having a glass of wine in your garden and you hear a buzzing noise. Oh no, wasps! But soon you realise that the wealthy reclusive neighbour is operating her small drone.

How do you react?

A: You are amazed at the continued advance in technology but relieved it was not wasps.

B: You wave at the drone ‘look up there’ and continue drinking your wine

C: You are annoyed so you decide to invite the neighbour to your BBQ

.

D

. You share your experience on social media

Slide15

Scorecard:

Preparing for the Future

(Adapted from Stanley Bing)

If you ticked all A’s then you are in the future already but it is not what it used to be!

If you ticked a combination of A’s and B’s then you are

ready

for the future.

If you ticked all C’s then you are

not

ready for the future.

Slide16

Scorecard

Betrayal of Linked Memory

If

you ticked all D’s then you are

beyond individual choice

creating the

meso

-data manifold under which 1. SIAL ‘copy and paste’ your memory

2. You have betrayed your emotions

Two Examples

:

(a) Bidding Against Yourself With Online Shopping

BIN pricing < END pricing

(b) ‘Linked memory’

The Daily Routine

Slide17

BIN price < END price

Bidding Against Yourself!

Online prices do not include latent transaction costs (TC)

Example: imputed TC calculated at

hourly wage x no of hours searching

=

minimum value of tradable asset.

Onsumers cannot interact with each other online at point of transaction. So there is experimentation to find a better price and this leads to change in behavior ‘at the margin’

This allows SIAL to experiment with a new BIN price converging to the END price and firm’s online prices move along the MR line not the demand line. This increases the elasticity of demand online with MR pricing.

Telco can nudge you from a 4GB to 5GB plan at marginal price of (additional) €5 per month

.

Value of Tradable asset at 5GB > MC = 0

Slide18

Algorithmic pricing:

BIN price < END price

Game SIAL v

Onsumer

1. If N

players in non-zero sum game

Elasticity falls and prices rise

2. If N+1

players in zero-sum game

Elasticity increases and net revenues

rise

Behavioural Signs and Symbols

Aumann

&

Sorin’s

bounded

recall -

if your repeat your behaviour many times then SIAL attaches a small probability

Maximin

&

Regret (‘last seat left’ or ‘3 other people are looking’ at this hotel room rate package)

=>…….

Online rational ‘

onsumer

’ pays a higher price at the end of an online shopping transaction than the opening bid price

Check with your own experiences: (

i

) Airline

ticket (ii) Hotel room

Slide19

FOMO

1. Rational

onsumers

interpret ‘missing information’ in the worst possible way.

2. Rational

‘

onsumers

’ acquire a

game DNA

as they

bid against themselves

ala

fictional story of ‘Ralph’s Pretty Good Grocery’

Law

of One Price Violated

BIN price < END

price

Large

data

provide

patterns

Small

data sets host the signs and signals that

can be nudged towards

pattern recognition

Small data …

meso

-data…….. Large Data

Slide20

The Daily Routine

The Hedgehog (large data patterns)

v

T

he Fox (small data patterns)

Algorithms

become

sufficiently intelligent

because

we outsource memory and

because

we betray our emotions to smart devices by ‘linked memories

’.

Example:

Send TXT ‘Just met Jane, going to

Sbux

instead’ or Tweet the message or Instagram the event.

Slide21

Slide22

Proustian

moment in time

Large

data

provide

patterns

Small

data sets

like

The Daily Routine

host

the signs and

signals, the symbols and surprises

that

link outsourced memory

and allow the SIAL

to nudge behaviour

towards

pattern recognition

Small data …

meso

-data…….. Large Data

A

meso

-data

set would contain ‘seeds’ (signs, symbols, signals, surprises) that can be used to code behaviour and decode data

patterns

The patterns

become

predictable with

meso

-data.

‘seed’ = a chance meeting with Jane ‘linked with’ Starbucks

or Indoor Atlas signalling (explain to audience)

Slide23

Slide24

At a

Proustian

moment in time

1. Would you prefer to eat in a virtual restaurant with no kitchen or in a restaurant with a kitchen?

2. Would you prefer a virtual surgeon or a consultant

to operate on

you?

3. Would you give your private personal data to a stranger passing by on the street?

4. Would you prefer a robot or a pilot to fly the EK A380?

Slide25

…………..THE

MESO-DATA

MANIFOLD

……

Slide26

Beyond individual sovereignty at a moment in time

1.

As your dinner is served you are informed that it was prepared in a virtual restaurant off premises.

2. At you arrive in the operating theatre, the anaesthetist informs you that a virtual surgeon will operate

on

you.

3. Your private personal data is encrypted by a stranger passing by on the street.

4. As you take your seat for the long-haul flight you are reliably informed that a robot will fly the EK A380.

Slide27

For business strategy exploit co-existence of online and offline consumer behaviour

Co-existence facilitates SIALS as ‘creative’

rather than

as ‘destructive’

Business should focus on

meso

-data

(

i

) to

extract

hidden information & hidden action;

(

ii) to identify frozen markets;

(

iii) to define the temporal distance, moment in

time before a decision (to buy) is made

temporal distance

is ‘moment’ between

action at time period t and

consequences at time

period

t+1

Slide28

Dear

Onsumer

Your patterns are g8t.

I am always with you of late.

That's why we,

Will always be,

Digital Serf & Master..…

Yours 466453.....Al.

Gorithm

Slide29

Seek Not Thyself Outside Thyself

‘Ne

te

quae

siveris

extra’

Thank you for listening………

Ralph Emerson

‘Habit is a great

deadner

’

Samuel Beckett

Waiting for Godot Act II

Slide30

Future Research

A methodology for

meso

-data v deep learning.

Mathematics of both omega circles & manifold in pattern recognition.

If G1 no nudge and G2 with nudge, then G1 and G2 are equivalent if we can obtain by reflection one or more sub-games.

The mirror test: has SIAL purchased what you would have independently selected?

At a moment in time is behaviour symmetric?

Slide31

Meso

-data

as a manifold

n players in a non zero-sum game

n+1 players in a zero-sum game

Is the

meso

-data set

r

eachable

?

Avoid mistake

Define object permanence

Two forces at work:

moving away v reacting

Moving away affects the speed of observer behaviour, so reacting is a secondary effect

Is moving away (from

BIN)

equivalent to reacting to

SIAL?

Are you

moving away from

BIN or

reacting to

END

Is

moving away

equivalent

to reacting?

Mimic or manifold patterns?

Slide32

Meso

-Data Manifold

Loops underlying the data manifold are discussed in an original article from University of Xi’an:

https

://

www.computer.org/csdl/trans/tk/2013/02/ttk2013020337.html

The paper discovers features in the data that fall on the loopy manifold …and this is not dissimilar in representation to our idea of ‘

meso-data’ with omega circles.

Slide33

Omega Circles

A = {BIN, END}

C = {BUY, Mr Al, EXIT}

 

What if: Ω (C, A) = C

A

?

What if: (ex-post) behaviour in smaller circles creates a pattern that ‘reaches’ into or ‘reflects’ the (ex-ante) behaviour

in

circumcircles? Ask: Decoding patterns from small

meso

-data to big machine data.

As

the END prices ‘reaches’ BIN preferably

diverging

away from BIN, Mr AL introduces a rationing rule

by

assigning excess supply in the form of a Dutch auction. So

the

onsumer

buys a concert ticket but not at the preferred seat location in the

arena.

Slide34

Meso

-Data

With Emotions

the ‘linked memories’

You txt or tweet that you achieved a personal health goal on your

FitBit

Deep learning attempts

to mimic the activity

in the brain

as an Euclidean action-reaction sequence.

1. Philosophical:

Someone that could be someone else

SIAL = You, the

onsumer

SIAL has become You as a person with a viewpoint that you share on social media

2. Mathematical

: Mimic or manifold patterns

?

Meso

-data with linked memories attempts

to understand the

action-reaction order

of the brain

as a n-sphere space.

Slide35

BEYOND INDIVIDUAL CHOICE:

Counter-argument

Individual choice & creativity

Steve Jobs: Why join the navy when you can be a pirate?

Is that sustainable?