Final Flashcards
1
Q
Definition digital economy
A
An economy based on Information and Communication Technology (ICT)
- ICT hard/software: Apple, Samsung, Microsoft
- Online ecosystems: Google, Facebook
- Online platforms: Amazon, airbnb
- Content providers: Spotify, Netflix
- Communication: WhatsApp, Teams
2
Q
Definition digital good
A
A networked zero marginal cost virtual object that has value for individuals or organizations
e.g.: operating systems and software, apps, “content”, video games
networked = based on ICT infrastructure
zero marginal cost = additional units are costless
3
Q
Definition digital service
A
A networked zero marginal cost service that has value for individuals or organizations
e.g.: Intermediation, online sales communications, search, social networks, advertising, online banking and payments
4
Q
Digital content & communication
A
Content: text, sound, images, video
Communication: voice, text, video
transition from analog to digitalised: vastly more efficient production, transmission, storage
5
Q
Physical sales channels
Advantages and limitations
A
- retail stores, banks, real estate agencies, hair-dressers, restaurants
Advantages: personal contact, confidence, quality observed ex-ante
Limitations: local footfall, limited capacity
6
Q
Online sales channels
Advantages and limitations
A
- Online stores, banking, intermediation platforms
Advantages: (in principle) global reach, unlimited capacity
Limitations: anonymous, high possibility of fraud, quality observed ex-post
7
Q
Intermediation services
A
- two-sided (or multi-sided) platforms which connect users that want to trade or interact
- sale of goods: eBay, Amazon
- Transport: Uber, Lime, Glovo
- Social networks: Facebook, LinkedIn
Not new: newspapers connect readers and advertisers since 1704
digital intermediation: larger reach and targets specific individuals
8
Q
Advantages digital over brick-and-mortar
A
- lower cost with storage and workers
- far wider reach
- much wider range of offers possible
- accelerated by covid crisis
9
Q
3 steps of digitization
A
Digital goods and services need:
- Digitisation of data
-> transformation of analog data
-> “native” digital data - Digital transmission:
-> packet-based internet
-> fibre-based broadband, 5G mobile - Digital processing and storage
-> server farms, cloud storage
10
Q
Moore’s Law
A
Predicted that number of components in integrated circuits would double every year
-> later revised to every two years
predictions have hold up
11
Q
Digital economy ecosystem
A
- ICT infrastructure forms the basis
- Internet service providers (ISPs)
- Backbone operators
- equipment and device manufacturers - Providers
- operating systems and computing platforms
- apps and software, storage - Digital Marketplaces
- Intermediate between providers and users - Users
12
Q
The internet value chain
A
- Content rights
- media right owners
- user-generated content - Online services
- Communication
- General/vertical content
- Search
- Entertainment
- Transactions - Enabling technology services
- Support technology
- Billing and payments
- Advertising - Connectivity
- Core network
- Interchange
- Retail Internet access - User interface
- Applications
- Devices
13
Q
Technology and Topology of ICT have changed
A
Technology: how data is processed, transmitted and stored
Topology: how different elements of a system are linked up
- from vertical integration to interoperability
- from monopoly to competition
- from separate service channels to convergence
14
Q
Convergence
A
- convergence to package-based transmission lets all simultaneously be sent through the same wireless or fibre link
15
Q
Production model: commons-based
A
- production undertaken by loosely organised individuals who are not employees of the platform
- made possible by ICT
16
Q
Production model: crowdsourcing
A
- special kind of peer production
- centrally organized
- for specific projects
e.g. open source software, bug bounty hunt
17
Q
Returns to scale
constant, decreasing, increasing
A
Constant:
- more inputs increase output in exactly the same proportion
- theory benchmark for long-run decisions when all factors of production can be adjusted
Decreasing:
- more input leads to less than proportional increase in output
- usually due to fixed factors (management time, production sites)
- typical “traditional” manufacturing
Increasing:
- more inputs lead to more than proportional increase in output
- typical network industries (telecoms, electricity, water) with huge fixed costs and low variable costs
18
Q
Returns to scope
A
- lower costs or other synergies from joint production of multiple goods
Google ecosystem:
-data flows btw. search, maps, mobile phones (Android), App Store
- improves search responses, map indications, UX
Apple ecosystem / walled garden:
- devices
- apple store, iTunes, etc.
19
Q
Scale and scope: effect on pricing
A
to CREATE scale of scope effects:
-Penetration pricing:
-> low prices to drive high early demand
- Bundle pricing:
-> different services sold together to create scope
in PRESENCE of scale and scope effects:
- Zero pricing:
-> reflecting zero marginal cost
the bigger firms are, the more profitable low or zero prices become
20
Q
Scale and scope: Effect on market structure
A
firms with more customers are stronger:
- necessity to fight for scale, even if short-run profits are negative
- endogenous berries to entry -> late comers unable to catch up
-> leads to winner take all markets
- persistent (quasi-) monopolies of some companies with global reach
e.g. Google, Facebook, amazon
21
Q
Ways in which free offers are profitable
A
- teasers to gain subscribers
- freemium models
- two-sided pricing
- exchange for data and advertising
- predatory pricing
22
Q
Teaser model
A
free viewing to see what content is there, but access only with subscription
- create trust for experience goods
- increase willingness to pay
- a digital showroom
Problem: competition with multiple competitors offering subscriptions
23
Q
two-sided platforms and intermediation
A
platform: organised markets where sellers and buyers meet
intermediation: real estate agents, auction houses, night clubs, shopping malls
The platform:
- provides mechanism for interaction
- catalyses cross-platform externalities between parties
- derives revenues from membership or transactions
24
Q
two-sided pricing
A
- users on side A benefit from more users on side B, vice versa
- pricing needs to weigh entry and participation incentives
- often two sides are asymmetric: one side pay very little and other side a lot
-> low prices on side that I more difficult to attract
25
Advertising classical platform business
- platform attracts consumers who are subject to advertising
- advertisers benefit from more consumers, but consumers hate more advertising
extreme two-sided pricing structure:
- consumers pay nothing
- advertisers pay (sometimes a lot)
-> services given in exchange for "eyeballs" or data
26
Broadcasting advertising
- traditional newspaper, magazine and TV
- broadcast: undirected transmission of content
- somewhat target to groups of consumers via choice of newspaper, magazines, TV channel, time of day or program
-> most advertising money is wasted, does not affect purchasing decisions
27
Targeted advertising
- internet: possibility to reach consumers individually
- much bigger bang of the buck
- spend only on targets with likely response
- tailored content and pricing
- possibility of immediate measurement of response
Google + Facebook won more than 30% of US advertising market
28
Data scale effects
- virtuous cycle in data accumulation
more usage and more data <-> more accumulated information and higher service quality
Advantage: drives better personalisation and innovation
Disadvantage: forecloses market entry by innovations and potential competitors
29
Data brokers
collect and match data on pretty much everybody, buy and sell data
30
Two concepts of market power
Foreclosure:
- eliminate the possibility of entry, enacted through unfair means
-> exclusive contracts
-> control over and excessive pricing of inputs
-> scale and network effects
Predatory pricing:
- eliminate competitors by charging low prices
- target firm gives up when it can no longer support the losses
- predator makes losses, but expects more than recoup them through higher profits later on
31
Enter Digital Services market
foreclosure and predation
- zero marginal cost implies that predation does not cause additional losses to the predator
- high fixed costs and low margins discourage entry
- accumulation of data leads to increasing knowledge and quality advantage that is more difficult to surpass
32
Data as essential input in digital economy
- demand and trends can be better forecast, products better designed
- services can be personalized
- data of different consumers are complements
-> usefulness and economic value of data increases much more than proportionally with no. of observations
-> joint processing of data of millions of transactions and observations creates value
33
Senator Hawley's SMART act
- social media addiction reduction technology act
- time on social media capped by default to 30 min
- ban on infinite scroll and autoplay of videos
Reasons:
- more time on social media is correlated with more loneliness, higher level of depression and anxiety
- social comparison via likes, followers and page views lead to low self-esteem, esp. among teenagers
34
Data ownership
conflicting objectives:
- max. efficiency in information gathering points -> ownership by firms
- protection of consumer rights and privacy -> ownership by individuals
Data individually attributed:
Yes, if names, addresses, actions taken, etc.
No, if patterns derived by machine-learning from individual observations
35
Concerns on digital economy
- consumer data is hoovered up in huge amounts
- privacy of individuals is ever less respected by firms and governments
- search, social media and online shopping have become quasi-monopolies
- data scale effects hinder entry of new competitors
- start-ups are bought up by giants instead of launching IPOs
36
Right to be forgotten
- Objective: allow ex-convicts and other people to start a new life
- Problem: esp. hard in the internet age, where information is easily copied, spread, and websites hosted in many countries
- US protects the "right to know" as form of free speech
- EU court of justice 2014: codification for digital age
Included in GDPR as "right to erase"
Implementation: individuals can ask search engines to delete certain pages
37
State surveillance
- digital footprints valuable for it
- USA: NSA analyses emails and social media (Wikileaks)
- Thailand: convictions for anti-royal statements on social media
- China: face-recognition systems, mandatory tracking apps, social points system
38
Cookie law
objective
implementation
usage of data
users can demand
protect against:
- online tracking
- personal profiling
- unsolicited marketing
- harvesting of personal data without consent
Implementation:
- limits on usage on data
- explicit consent needed for harvesting data
Usage of data:
- what data can be used for
- how data is handled (anonymised and protected unless consented)
- how and for what purpose data can be shared
- no use of email addresses for unsolicited marketing
Users can demand:
- access and insight
- rectification
- deletion
39
General Data Protection Regulation (GDPR)
- defines principles for data processing and storage
- type and amount of personal data depend on reason for processing and intended use
- applies to all firms doing business with users in EU
Key rules:
- Lawfulness, fairness, and transparency
- Purpose limitation
- Data minimisation
- accuracy (data correct and up-to-date
- no other use that is incompatible with original purpose
- Storage limitation (no longer than needed)
- Integrity and confidentiality (tech and org safeguards agains unlawful access or loss)
40
California Consumer Privacy Act (CCPA)
Looser than GDPR:
- no requirement of legal basis
- no org requirement for data confidentiality
Provides:
- right to know about info
- right to delete info
- right to opt-out of sale of personal info
- right to non-discrimination for using CCPA
Stronger than GDPR:
- wider defintion of personal info
- explicit prohibition of discrimination
- explicit "do not sell data" option
41
GDPR Implementation issues
- each website has to ask every user for consent
- websites have incentive to present consent banners featuring only "OK" button instead of opportunity to choose
- consent fatigue: users click ok just to get to website quickly
- myopic users click "OK" more often than they would if they considered outcome
Result: websites still collect personal information unhindered, while users have hassle of all those banners and pop-ups
-> even browser extensions that automatically consent to cookies
42
Price discrimination
Practice of charging different prices for the same good
1. degree: charge each client different price -> personalized pricing (seller must know characteristics of client)
2. degree: seller does not know the client, but offers a menu to choose from (consumer self-selects)
3. degree: different prices for identifiable groups (e.g. students, retired; often to protect social groups; stronger groups pay more to finance the good, cheaper for weaker groups)
43
Geographical pricing
- Firms with shops in different locations often charge different prices
-> pricing follows local purchasing power
44
Personalized pricing: Trade-offs
Economic theory benchmark: IF seller were to exactly know the preferences and willingness to pay of a client, they can:
- design product to fit the client (as much as costs permit)
- try to extract all the surplus
- If design personalized -> increases social welfare -> good
- If not enough competition -> all gains go to one firm -> unfair?
45
Why digital economy makes personalized pricing possible
- large amounts of data on individual preferences, wealth, willingness to pay
- immediate testing and feedback to offers
- individual communication channel (clients cannot observe each other's offers; divide-and-conquer approach)
46
Customer screening using menus
- sellers cannot identify customer characteristics before they buy
Options:
- charge one single price: ok if customers are not too different, otherwise leaves money on the table
- design menu of choices and let consumers pick -> reveal private information about preferences, wealth, willingness to pay -> increase profits if customers are very different from each other
47
General features of menus
- main challenge: make sure each type of customer picks option that is meant for them
- consumers with high valuation have strong incentive to lie and pick cheaper option
-> instead of lowering price of high-quality option, seller downgrades quality of cheaper option
48
two-part and three-part tariff
- tariff that does not have a constant ratio price/quantity is called nonlinear tariff
Two-part tariff: fixed fee + constant price per unit
- Utility pricing, Cable TV subscription plus video-on-demand, razors + blades, printers + ink
Three-part tariff: fixed fee + bundled units + constant price for additional units
- Telecoms tariffs
49
Bundling
Goods are sold together even if consumers have different tastes for them; special type of menu pricing
- Software (Office)
- Media (Newspapers, magazines, CDs
50
Why bundling increases revenue
- low prices attract many customers but bring low margins
- high prices have high margins, but only high-value customers buy
- bundling smoothes distribution of values of products on offer, so higher average price can be charged
51
Mixed bundling
Seller offers both the bundle and parts of it sparately (e.g. McDonalds and McMenu)
52
Law of one price
Assumption:
- firms sell homogeneous good post prices
- consumers observe all prices immediately, correctly, costlessly
- consumers only buy at firms with lowest price
Conclusion:
- market settles quickly on a single price
- should be no price dispersion (different sellers setting diff prices)
- any firm that has higher cost is forced to leave the market
-> implies efficient market outcome: all consumers whose valuation is higher than cost are served
53
Law of one price and internet
Expectation:
- ease of access to information as force towards more pricing transparence and more competition
- many price comparison sites sprung up as online shopping took off
Reality:
- may even become more complex because prices change faster (price discrimination)
- search cost not gone
- diff degree of search cost
- strategy from companies to make it diff to compare prices
54
Minimum retail prices
- floors on retail prices imposed by a good's producer
Idea: prevent sellers from competing down the retail price
-> low retail price = low margin for seller
-> feeds back into downward pressure on wholesale price
-> reduces the producer's returns
Illegal in US
- considered anti-competitive conduct that coordinates retail prices
Permitted in EU in some cases
- protection of certain sectors (books)
55
Gatekeepers
Big sellers (e.g. Amazon Marketplace)
- market power through network effects allow to charge high prices
- control access to platform disciplines pricing behaviour of smaller sellers
- control access may imply direct intervention in pricing of other sellers
- observed transaction data of sellers is valuable source of information -> no own pricing experiments necessary
56
Umbrella pricing
In market with gatekeeper:
smaller sellers align their prices with the platform's higher price
Do not compete on price because of fear of retaliation
57
Most favoured customer clauses
Platforms prohibit associated sellers to offer good elsewhere at lower price
- booking.com, Expedia, Amazon
- Presented to clients as "best price guarantee" -> sign that market is not competitive
- clauses meant to restrict price competition with platform
58
Algorithmic pricing
- leads to collusion
- algorithms consistently learn to charge supra competitive prices, without communicating with one another
supra competitive: pricing above what can be sustained in a competitive market
59
Network effects
Effect that the number of users or usage of a service has on the value of that service as perceived individually by each user
- usage imposes positive externality on other users
- contrary to non-network goods, where consumption does not affect other users
- has strong implications for business strategy and competition
60
Metcalfe's Law
total value is quadratic in the number of users n --> each user interacts with everything and everyone on a platform
urban myth because most people interact mostly with small number of family, friends, work contacts, etc.
overestimated value of companies with network effects
61
Odlyzko-Tilly's law
Network value grow is faster than linear, but much slower than quadratic
Pareto distribution: value for given user proportional to ln(n)
62
Indirect network effects
- network effect without users interacting directly
- intermediate by some platform through cross-platform externalities
-> user group A benefits from presence of user group B, vice versa
e.g.:
- demand and supply content for LP, VHS, CD, DVD
- players and developers on game platforms: Playstation, Wii
- Users and developers on operating systems: Windows, iOS, Android
63
Demand for network goods
- depends on existing number of other users
- depends on expectations about future number of users (important if network is new to the market)
Affects:
- shape of growth path
- potential market outcomes
64
Diffusion curve
Typical shape of number of adopters over time: S-curve or logistic curve
- Slow launch: few interactions possible
- Fast takeoff: network effects kick in
- Slow-down: few non-users left to join
65
Customer groups along diffusion curve
- Innovators (care less about network effect)
- Early adopters (pulled in by example of innovators)
- Early majority (adopters follow the network effect)
- Late majority (bandwagon effect)
- Laggards (eventually stop resisting)
66
Critical mass
- the threshold where exponential growth takes off (no. of customers)
If yes: Bandwagon / Domino / Snowball effect
-> stereo sound, VHS, QWERTY keyboard
If no: shrinks back to zero
-> Betamax, DVORAK keyboard
67
Role of expectations for network goods
essential for growth in network markets
tend to be self-fulfilling:
- common commercial strategy is penetration pricing: initially low prices to reach critical mass rapidly, then raise prices
68
Competing networks
Assume incompatible networks: users only derive utility from other users of the same network
No. of networks in equilibrium depends on preference for variety and strength of network effects
Entry affected as network effects are hard to overcome
69
Ergodic system
long-run outcome does not depend on previous events
Non-network markets are ergodic -> market domination does not depend on history
70
Path-dependent market
- outcomes depend on history
- small events might be enough to determine future of market, e.g. small group of early adopters
- after time firms grow a lot, new competitors cannot catch up
- the strong become stronger, weak become weaker
- increasing returns to scale (declining average cost)
- winter takes all market
71
Why several companies can exist in path-dependent market
- preference for variety stronger than network effects + returns to scale
- Gen Z don't want to be on their parents' platforms (fb)
72
Absorbing barrier
number of users that make the market tip
fan of network i:
receives u + ni for network i, otherwise nj
Fan of i prefers other network if u + ni < nj or nj - ni > u
in this case, everybody prefers network j
Result: lock-in, only network j remains, network i shrinks and no anticompetitive conduct has taken place
73
Competition between platforms
- network effects drive market structure towards huge quasi-monopolies (Google, Amazon, Facebook)
-> particularly likely if user groups single-home (competitive bottelneck)
- some platforms have stable competition because groups multi-home
-> Uber drivers use Bolt, FreeNow, etc. - as do riders
- strategy to enter a platform market: divide-and-conquer
-> give groups with high externalities subsidies to join the platform
-> charge the others later: cross-group subsidies
74
Will network effects lead to early or late adoption of new technologies?
Late Adoption: Consumer 1 prefers N but keeps O because expects consumer 2 to keep O -> consumer 2 does not switch because 1 didn't
Early adoption: consumer 1 prefers O but switches to N because expects consumer 2 to switch -> consumer 2 switches to N because 1 did
75
Compatibility of network goods
- extends firm-level network effects to market-level network effects
- raises total surplus created -> good for society
- firms will compete in market rather than for market
-> network effects no longer differentiator
-> less likely to become monopoly
- choose compatibility unless firm believes they can take over the market
- compatibility often imposed: interconnection of telecoms networks
76
Compatibility with legacy products
- built on interaction with legacy user base to sell new product
entrants use this strategy:
- one sided compatibility allows customers of entrant to interconnect with incumbent user base
- incumbent may tolerate to sabotage
-> e.g. Nescafe machines, changed to not work correctly with competitor's copycat capsules
77
Policy issues with standardization
benefit: larger network externalities
Problems of government intervention:
- how should government know which technology brings most future benefits -> competition selects technologies and brings more choice and lower prices
- decision process creates uncertainty and delays -> can kill new technology
- attempts to create national champion distort and segment market outcomes
e.g. HDTV standard battle between USA, Japan, Europe for decades
78
Multi-sided platform (MSP)
enables direct interaction between two or more distinct user groups in which all user groups are affiliated with the MSP
candidates: markets with at least 2 interacting groups (e.g. buyers/sellers)
one or more platforms
Payments to platform: membership, usage, or free -> distribution of payments may matter
79
Cross-platform externalities
- presence of a group benefits the other
- need not to be symmetric: eyeballs positive for advertisers but not vice versa
80
Within-group externalities
direct: negative between sellers because of more competition
indirect: positive between sellers because more sellers attract more customers
81
Topsy-Turvy Principle
Skewness of prices:
- one side pays very little or zero, other side charged heavily
e.g. payment cards, adobe acrobat, video game platforms
82
Linkage
coordination problem that platform needs to get enough members of both groups (users, advertisers) to join
subsidize side that is harder to capture:
- multi-homing: one side uses alternative channels (shopping sites, search, credit cards (cash/debit))
- absence of relationship-specific investments: can move easily (opening stores is costly, shoppers only need transport)
83
Competition policy lopsided pricing
In one-sided markets: price below marginal cost is considered anti-competitive (predation or dumping)
In two-sided markets: can be efficient (get right people on platform)
Problem: difficult to identify fair pricing structure
84
Standard procedure in competition cases
1. Market defintion
2. Find dominance or not
3. Remedies
Fundamental problem in multi-sided market: market defintion unclear
-> which sides to include
85
Newspapers moving online because
- print newspaper advertising has collapsed (does not allow precise targeting)
- increasing share of online ads in total ad revenue
- subscriptions online -> freemium, teaser model
86
Why cost cutting at newspapers is unavoidable but self-defeating
- fire full-time journalists and rely on part-timers or freelancers
Vicious cycle:
- quality and investigative journalism becomes niche occupation of few outlets
- news articles are copied or written by someone unqualified
- business newspapers transcribe companies' press releases without background research
87
Why online advertising is endangering traditional media
- returns to scale
- network effects
- competition from "free" content
- targeting of digital advertising
- control of gatekeepers
88
Winner takes all in digital media world
- zero marginal cost: no tech limits on distribution
- network effects/reputation: growth breeds growth
- indirect network effects: audience draws advertising which provides funding
- quality is costly and get a feedback loop
Outcome: few national outlets can be highly profitable, but not smaller ones
e.g. make money by catering engaged political groups (Fox News)
89
Paying for online media
- for exclusive or curated news
- specific topics or investigative journalism
- unlimited content
-download possible
once readers learn to pay for online news, they are likely to stay
90
Advantages of more information
- better-informed citizens make better choices
- more information leads to more civic engagement
- trust in information and transparency are essential for keeping democracies functioning
91
Misinformation
- easily to spread than verifiable information
- with viral techniques, populist actors can capture huge audiences through online channels
- concept of truth dissolves in onslaught of fake news
92
Media plurality
- multiple media channels with different opinions
- present different voices in society
- move to online media increases concentration of news channels
- Media in EU gets more concentrated (esp. in small countries (high fixed costs))
- online media access more concentrated than traditional media
-> information gatekeepers (Google, Facebook)
93
Risks in digital media
- online surveillance and digital risks to journalists
- insufficient internet access
- ineffective implementation of net neutrality
- online hate speech against vulnerable groups is under-investigated and poorly handled
- high digital risk for political independence is associated with lack of transparency and safeguards
94
Governments and internet access (critical views)
Increase in critical views is larger:
- where press in censored
- with rural repondents
- in countries that allow free web browsing
Exposure to information and critical views raises citizen's awareness of problems
95
Shifting patterns in news consumption
- TV most important source
- news websites overtaken radio
- print newspapers overtaken by social media
- social media + news websites already rival TV
96
EU's code of practice on disinfromation
1. Disrupting advertising revenues of certain accounts and websites that spread misinformation
2. Making political advertising and issue based advertising more transparent
3. Addressing issue of fake accounts and online bots
4. Empowering consumers to report misinformation and access different news sources
5. Empowering research community to monitor online misinformation
97
European digital media observatory
- hub for fact-checkers, academics and other relevant stakeholders to collaborate
- link with media organisations, media literacy experts and provide support to policy makers
98
European action plan on disinformation
- improving detection, analysis, and exposure of disinformation
- stronger cooperation and joint responses to threats
- enhancing collaboration with online platforms and industry
- raising awareness and improve social resilience
99
Digital Services Act
regulation of online services:
- liability for content
- market dominance
- advertising
- safety
- smart contracts
- online self-employment
- future governance frameworks
Public fight about what should be included
100
Android operating system
- Google's attempt to control mobile ecosystem, search, ads, data flows
- get as many hardware producers and developers on board as possible
- opposite to walled-garden business model of Apple
101
Why firms are getting bigger and bigger
- zero marginal cost
- returns to scale
- wide adoption of online news, shopping, social networks
- growth of large digital ecosystem
- global reach (plus global tax evasion)
- lots of acquisitions of smaller or not-so-small firms
102
EU merger regulations
Aim: prevent mergers from reducing competition
- M&As above a certain size must be notified to national competition authorities
- Even larger ones to Directorate-General of Competition in Brussels
Investigation process:
- Potential competitor problems?
- Work out competition problems in details
103
Abuse of Dominance
Dominance = position of strength that allows from to behave in extent independently of its clients and competitors
- not illegal as such
- but abuses of dominance forbidden
-> foreclosure or predation of competitors, excessive pricing or imposition of unfair trading conditions, leverage of dominance into other markets
104
Asimov's Laws
1. a robot may not injure a human being or, through inaction, allow a human being to come to harm
2. robot must obey the orders given by human beings expect where such orders would conflict with the first law
3. robot must protect its own existence as long as such protection does to conflict with first or second law
105
Algorithmic price discrimination
- algorithms deduce individual preferences and willingness to pay
- benefit: better matches, customised offers
- disadvantages: higher prices, surplus extraction
- PD targeting preferences hurts consumers but increase efficiency
- PD targeting misperceptions hurts consumers and reduces efficiency
106
Cartels vs. Tacit collusion
Cartels: explicit agreements between firms to limit competition and innovation, create market power, increase prices and profits
-> Illegal because blocks functioning of market and hurts consumers
Tacit collusion: uncoordinated market outcome where firms do not compete
-> learned that aggressivity leads to punishment
-> not illegal, but bad for market outcomes
Prevention: EU avoid coordinated effects in mergers (collusion easier with few firms)
107
Algorithmic collusion
- algorithms quickly understand to avoid price wars to keep prices high
- facilitates collusion between competitors
108
What to do about algorithms
- voluntary codes of conduct (respect for privacy, sale of data)
- transparency requirements (codes checked by competition authority)
- certain obligations and prohibitions (deletion of personal info)
- liability laws (against illegal or harmful content)
109
Standard Contractual Clauses of EU
- set terms for transfer of personal data of EU
- creation of additional safeguards following decisions by courts
- codified role of data controller and data processor
110
European data governance act
Aim: create process and organization for sharing industrial data
-> data from sensors, self-driving cars, health interventions, industrial and agricultural activities, personal data
- data should be used rather than accumulated
- creation of neutral data intermediaries
- creation of sectoral or personal data spaces (control over sharing)
111
Digital Services Act
- regulation of labor in digital services
- transparency in ads (who pays, why targeted)
- liability and content moderation
- platforms must control identity of sellers
- transparency of algorithms
- greater transparency for online political advertising
112
Digital Markets Act
which firms and definition of gatekeeper
- to limit abuse of market power by gatekeepers
- Systemic platforms (Apple, Google, Facebook)
- not all firms from US
Definition of gatekeeper:
- impact on EU single market
- services unavoidable for smaller players
- dominant position with ability of kill of competition
113
Digital Markets Act
Obligations
- inform about intentions when acquiring firm
- give users ability to uninstall gatekeeper apps
- submission of consumer profiling practices to annual audit
114
Digital Markets Act
Ex-ante prohibited practices
- self-referencing in search and online intermediation
- use of marketplace data against sellers
- ban on exclusive pre-installation
- prohibition of exclusive use of data
- exclusivity in marketplaces
115
Digital Markets Act
Sanctions
-fines
- remedies
- break-up of European operations
116
Digital Markets Act
Greyest of unfair practices
- prevent third-party sellers from accessing essential information on customers
- collection of personal data beyond what is necessary for the service
- preventing hardware manufacturer form providing customers and business users with choice of options for applications/services
117
Democracy Action Plan of December 2020
Aim:
- fight disinformation
- enforce rules on fair competition in online public debates
- protect integrity of elections
-> more obligations and accountability for online platforms for amplifying disinformation
-> from self-regulation to co-regulation (Code of Practice, monitoring, checking)
-> Cyber diplomacy toolbox
118
Advantages of pricing algorithms
- quick, based on latest data
- automated, can be constantly repeated
- easy customisation
- predictive tools
119
Disadvantages of pricing algorithms
- potentially excessive personalisation of pricing and services
- can lead to price spirals
- coordinator between firms without human intervention and explicit intent
120
Code of Practice for Disinformation
- improve provision of more granular and country-specific data
- set-up publicly available, up to date, country-level database of trustworthy sources
- clarify what forms of content the firms must act upon
- provision of data on use and performance of automated systems should be explicit commitment
- access to data for independent research should be binding
121
Media Freedom Act
Aim: achieve balanced and impartial media coverage
- specific treatment of media companies b.c. of their crucial role in effectively ensuring democracy across EU member states
- provide access to plurality of views and reliable sources of information
