Control Choices and Network Effects in Hypertext Systems

1
Control Choices and Network Effects in Hypertext
Systems

• Jim Whitehead
• Univ. of California, Irvine
• Hypertext99
• February 21-25, 1999

2
Control Choices and Network Effects in Hypertext
Systems

• Network effects are an important (perhaps the
  primary) adoption factor for hypertext systems
• Control choices in the architecture of a
  hypertext system directly affect its ability to
  generate network effects.
• control of data
• control of hypertext structure
• control of user interface

3
Network Effects

• An item possesses network effects when the
  utility of an item increases with the number of
  other users of that item.
• Has network effects telephone system email
  system
• No network effects breakfast cereal
• Different from scale effects where cost decreases
  as volume increases
• Breakfast cereal does have scale effects

4
Direct and Indirect Network Effects

• Direct network effects
• When the utility of the item depends directly on
  the number of other people using it
• Telephone system
• Indirect network effects
• An artifact intermediates between an item and the
  effect of the number of people using the item
• Computer hardware and compatible software
• Known as hardware/software systems
• Hypertext systems are hardware/software systems
  (viewer is hardware, content is software)

5
Direct Network Effects

• Utility
• Personal utility is the weighted sum of the
  utility of each communications path with existing
  users.
• Demand
• A person has demand for a system if the personal
  utility of using the system is greater than or
  equal to its price.
• Source Rohlfs, J. A theory of interdependent
  demand for a communications service Bell J. of
  Economics 5(1), 1974.

6
Direct Network Effects (2)

• Any system starts with no users.
• When all people who have demand for a system at a
  given price are using that system, it has reached
  equilibrium.
• But there are many paths to equilibrium.

7
A Model of Network Effects in Hypertext Systems

• In hypertext systems, network effects are
  generated via the interaction of
• Information producers (content providers)
• Information consumers (readers)
• Producers and consumers have different utility
  functions, but both act to maximize their utility
  from using the system.

8
Hypertext User Utility

• Reader utility
• Derived from the amount of information and
  services available on that system.
• Content provider utility
• Derived from the number of people who read
  documents or use services offered by the content
  provider.

9
Feedback Loops

• Simple feedback loops lead to increases in
  readers and content providers.
• Readers are lured to the system to take advantage
  of the content and services offered by the
  hypertext system (readers create a demand-pull
  for content)
• Content providers are attracted to the system to
  have readers view their content (content
  providers create a supply-push for readers)
• Content entices readers, readers attract content,
  and so on a virtuous cycle.

10
Control Choices

• Each class of hypertext system made different
  control choices in its architecture
• Monolithic control user interface, hypertext
  structure, and data
• Open Hypermedia control hypertext structure and
  data, but do not control user interface (each
  cooperating application has its own UI)
• Web control user interface, but do not control
  hypertext structure or data

11
Network Effects for Monolithic Systems (1)

• The choice to control all aspects of the system
  leads to limited network effects.
• Readers Attracted to these systems by the rich,
  highly useful content, but the amount and variety
  of this content is limited.
• Content providers Due to the need to learn new
  editors, and because there are relatively few
  initial readers, little incentive to provide
  content.

12
Network Effects for Monolithic Systems (2)

• Bounded reader population
• Lack of remote access to content limits readers
  to those who have access to the local file
  system.
• Bounded content provider utility
• Bounded population of readers limits content
  provider utility, while need to learn new tools
  increases cost, leading to low motivation for
  content providers to add new information.
• Eventually leads to a lack of interest from
  readers.
• Low generation of network effects.

13
Network Effects for Open Hypermedia Systems (1)

• Readers
• Drawn to individual applications in order to
  access documents within the system.
• Motivated to use open hypermedia capabilities to
  gain advantage of key OHS feature
  inter-application hypertext linking
• Content providers
• Have incentive to add links because they are
  immediately useful to provider and other users
• The ability to link together data is limited only
  by the number of hypertext-aware applications.

14
Network Effects for Open Hypermedia Systems (2)

• Several known problems with open hypermedia
  systems
• Editing problem, versioning, lack of user
  interface consistency, limited screen real
  estate, etc.
• Problems limit network effects
• Reduce incentives for readers
• Increase the maintenance burden for content
  providers.
• Limited distribution support caps the total
  possible number of readers.

15
Network Effects and the Web (1)

• Reader utility
• Provided by large amount of documents and
  services available via the Web
• HTML provides visually pleasing displays
• Content provider utility
• Early-on, Web provided gateways to existing
  information sources (gopher, FTP, etc.)
• HTML provides control over appearance of
  information

16
Network Effects and the Web (2)

• Control user interface
• Provides a single, attractive, easy-to-use entry
  point into the system.
• No control of hypermedia structure
• Allows hypertext to grow without scalability
  limits.
• No control on hypermedia data
• Accommodates a wide range of information stores
  (e.g., databases, document management systems),
  enabling more information providers.

17
Conclusions (1)

• Scalability problems limit network effects
  generation
• caps on readers ? caps on provider utility ?
  arrests path to equilibrium
• Scalability-limiting control choices
• Data control tight control over data creates a
  data access chokepoint
• Hypermedia structure control tight control over
  hypermedia structure creates a link traversal
  creation chokepoint

18
Conclusions (2)

• Control over the user interface increases reader
  and content provider utility.
• It incentivizes readers due to visually pleasing
  displays
• Allows for more control over the presentation by
  content providers
• Designers need a lot of control over presentation
• Increased utility creates greater demand, more
  adoption, more network effects
• However, dont control too much Gopher pitfall