Some Lessons from Successes and Failures of Electronic Trading

Download Report

Transcript Some Lessons from Successes and Failures of Electronic Trading

NBA 600: Session 2
The Internet and its Use
23 January 2003
Daniel Huttenlocher
Today’s Class
 Reflect on Tuesday’s lessons
 Competitive forces and online businesses
– Consider Porter’s 5 forces
– Illustrate with a mom & pop business
• Importance of online community
 Structure of Internet
– How it is constructed and connected
– How it is paid for
– Implications for electronic goods and services
 Credibility and expectations online
– Why it is important, how it can be achieved
2
Reflect: On Staying Informed
 Where do people get information today?
– Personal or work
– At home, at office, elsewhere
– News, govt., health, product, work, gossip
 How do people exchange information
directly with one another today?
– Friends, family or coworkers
– At home, at office, elsewhere
– Face-to-face, email, IM, gaming…
 How are these changing?
3
Reflect: Value vs. Revenue
 Many Internet services that people rate as
valuable are not generating much revenue
– How can the value be captured by service
providers?
• E.g., instant messaging on the Internet vs.
short-text messages on mobile phones
 How important is AIM to AOL?
• Tension between giving away and charging
– As more services move online or move digital
what are revenue opportunities?
– Differences in what people are willing to pay
for online and offline
4
Thought Experiment
 Creating a “mom & pop” Web business
– E.g., goal of $1-2MM annual revenue
 What are users willing to pay?
– $25/yr, $50/yr, $100/yr, $250/yr
 What percentage of Internet population
are possible users? (220M active in world)
– 1%, .1%
 What percent of those will sign up?
– 10%, 1%
 Case: Chess Club – online games/rankings
5
Internet Chess Club
 Over $1.25MM annual revenue
– Over 25,000 paying members
• About .01% of possible worldwide Internet users
 .1% eligible and 10% of those signed up?
– $50/yr fee
 Costs
–
–
–
–
–
–
Initial software development
Software and server maintenance/upgrades
Network connectivity
Billing
User support
Maybe half-dozen employees total
6
7
Chess Club: Competitive Position
 Porter considers five forces
– Rivalry among competitors
• A good chess club provides good players –
“network effect”
• Others may undercut price, but their value
depends on having good players as users
– Threat of new entrants
• Difficult for others to get critical mass of users
and to build user rankings
 What about sites with many users like Yahoo?
– Threat of substitutes
• Will easier ways of bringing chess enthusiasts
together come along (e.g., wireless devices)?
8
Chess Club: Competitive Position
 Porters forces continued
– Bargaining power of suppliers
• No real suppliers except commodity hardware,
software and networking (chess software
developed in-house)
– Bargaining power of customers
• Need to find the cost comparable to other chess
activities, other forms of entertainment
 Appears to be sustainable business
– Can it be grown?
 How general – what similar businesses to
be built
– Online community
9
Structure of Internet: 2 Main Ideas
 1: The Internet is a collection of networks
– Held together by standard “protocols” (TCP)
– Like road networks
• Local, county, state, national
• Agree on where to connect and how to drive
 2: Packed switched data
– Information broken up into small packets each
addressed separately to the recipient
• Unique addresses – “IP address”
– Like filling up separate cars and sending to
same place with independent drivers
• Each driver makes own local routing decisions
10
Inter-Connection of Individual
Networks
11
How Data Gets Where Its Going
 User generally provides a hostname
– E.g., www.cornell.edu, www.cnn.com
 Name translated to an IP address
– Uniquely specifies where data goes to reach the
hostname
– Domain Name System (DNS) handles lookup
• Small number of “root servers” around globe
• Local servers handle most requests (e.g., CUDNS)
 IP address is used to route data
– At each connection of networks, best place to go
to get closer to that address
• Routes change dynamically with traffic conditions
12
Routing Example www.yahoo.com
Tracing route to www.yahoo.akadns.net over a maximum of 30 hops:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
10 ms
<10 ms
<10 ms
<10 ms
<10 ms
10 ms
10 ms
10 ms
10 ms
10 ms
10 ms
10 ms
10 ms
20 ms
10 ms
<10 ms
<10 ms
<10 ms
<10 ms
<10 ms
10 ms
10 ms
10 ms
10 ms
20 ms
21 ms
20 ms
20 ms
20 ms
20 ms
<10 ms bb3-2-msfc-vl64.cit.cornell.edu
<10 ms core2-msfc-vl8.cit.cornell.edu
<10 ms cornellnet4-dmz1.cit.cornell.edu
10 ms at-gsr2-syr-1-2-cornell-OC3.appliedtheory.net
11 ms at-gsr1-syr-6-0-OC12.appliedtheory.net
10 ms so-4-1-2.nycmny1-hcr3.bbnplanet.net
10 ms acr2-so-3-1-0.NewYork.cw.net
10 ms agr4-loopback.NewYork.cw.net
10 ms dcr1-so-7-3-0.NewYork.cw.net
20 ms dcr2-loopback.Washington.cw.net
20 ms bhr1-pos-10-0.Sterling1dc2.cw.net
20 ms csr12-ve241.Sterling2dc3.cw.net
20 ms 216.109.84.166
20 ms vl44.bas2.dcx.yahoo.com
20 ms w10.dcx.yahoo.com
13
What’s Shown in Example
 Getting to the end location is fast
– 20 ms (millisecond) response time
 Large sites have multiple locations
– Yahoo uses Akamai to determine where to
send my requests – in this case to DC area
• Network does not specify geo-location explicitly
it is inferred from where I’m connected to net
 Data travels over many networks
– Applied Theory is CU’s provider, C&W is
Yahoo’s provider
• Data flows over BBN (Genuity) network too
14
How Fast Data Travels
 Speed of light in fiber about 200x106 m/s
– About 2/3 speed of light in a vacuum
 Distance coast-to-coast about 4300 km
– Double for round-trip: time to send a request
and get an answer back is what matters
 So “best possible” with fiber about 43 ms
– Interactive speeds are about 80-100 ms
• E.g., “flicker fusion” of images at 10-12/sec
 Measured times about 60-70 ms for good
service provider – including routing time!
– Possible to build national interactive services,
but not international without multiple locations
15
Different From Earlier Networks
 Railroads, telephones (pre 1990’s)
– Centralized ownership/control rather than open
standards and inter-connected competitors
– Centrally planned capacity rather than ad hoc
growth and negotiated connections
– “Circuit switched” rather than “packet
switched”
• End-to-end phone connection – switchboards
• End-to-end data – entire train not separate cars
 Note: modern phone networks moving
more towards Internet-style structure
16
How Internet is Constructed
 Each network provider (ISP) builds its own
network
– Chooses what other network(s) to connect to
– Chooses what traffic to accept from connected
network(s)
 Pairwise peering arrangements govern
what inter-network traffic will be carried
– Sometimes involves charges, sometimes
involves trades – local decisions by two ISP’s
 Each provider motivated to provide
connectivity needed by their customers
17
Payment Structure of the Internet
 End-users pay for bandwidth
– $20/mo 56kb dialup (consumer)
– $40/mo 128kb upstream - 768kb downstream
broadband (consumer)
– $900/mo 1.6mb T1 (commercial)
– $40k/mo 155mb OC3 (commercial)
 Commercial users tend to use their full
bandwidth 24/7, consumers not
– Asymmetric broadband appears to be
disproportionately cheaper as a result
 Each of these categories generates $100’s
millions of monthly revenue
18
Connectivity of Internet and Web
 Evolutionary rather than designed
– Shows many patterns similar to natural or
“organic” growth phenomena
• E.g., neurons, “six degrees of Kevin Bacon”
 Good routes evolve through needs of endusers
– Any two hosts about 15 hops (degrees) apart
 Aside: analogous structure in hyperlinks
between Web pages
– This structure is used for search engines, e.g.,
Google
19
Geo-Location in Internet
 The cabling and “routers” (connections) of
the Internet are in physical locations
– Often in or near big cities where the traffic is
 These physical locations are not evident in
the IP addressing scheme
– Companies sell services that try to determine
geo-location from IP address
• For marketing, security, legal and other uses
 Large service providers need to combat
physical location
– Packet transit times in the network too slow
– Potential for congestion with single site
20
Difficult to “Map” Internet
 Trace routes from certain locations (host
machines) in the Internet
– Routes change dynamically
– Only see routes from a few places to many
– Like trying to map roads by exploring specific
destinations from a few starting locations
 Internet changing every day
– On order of 150,000 “routable networks”
• I.e., service providers with peering agreements
 Easier to depend on gross characteristics
than specifics of Internet structure
21
Reflect: Conclusions We Can Draw
 Relentless, organic, drive to connectivity
– Services that do not connect to the Internet
are at risk
• Consider case of AOL’s free AIM client
• What does this say about text messaging
 Consider Blackberry/RIM or Bloomberg messaging
niche businesses
 Global, highly interactive services need
multiple physical locations in the network
– Simply slow if interacting users far apart
 Upstream bandwidth will remain costly
22
Credibility Online
 Crucial to success of any online business
– Can’t use many traditional means: location,
offices, attire
 Brand important, but for new business
hard to build
– For existing business important not to damage
• Low credibility online presence may hurt
 What is credible online?
– People say: identity, sponsors, service, privacy
– Stanford study reveals layout and design most
important – old “say vs. do problem”
23
What is Credible Online
 Consider Orbitz – new business
– Strong pedigree (sponsors), heavy advertising,
fairly simple almost “retro” design
– Has quickly grown into large business
 Consider Internet Chess Club
– Existing business, little pedigree, no
advertising, little design
– Successful small business
• Online community – appearance less a factor?
 Effect of broken sites
– Recognized brand versus not
24
Summing Up
 Porter’s 5 competitive forces and Internet
– Network effect – value of large user base and
its impact on competitors or new entrants
– Lack of non-commodity suppliers for many
net-based businesses
 Technical architecture of the Internet
supports decentralized as-needed growth
– Local demand for connectivity driving growth
– For what businesses is that connectivity key
 Building credibility is important but poorly
understood – need good layout and design
25
Next Time
 Internet and New Economy debate
– Readings: Porter, Tapscott, Magretta
– Skim: Nordhaus
– Be prepared to take a position and defend it
 First individual assignments next Tuesday
– Due following Tuesday Feb 4, before class
– Choice of topics:
• Challenges and opportunities for consumer
mobile text messaging services in the US
• Porter-Tapscott debate, changes to strategy in
the Internet age
26