GSB M395 : Internet Marketing
Debate 3 : Usage Fee
Topic: Implementation of Per Packet Usage Fee
Statement: Be it resolved that the lack of per packet usage
fee will increasingly cause a tragedy of commons with brownouts
and increasing congestion, and that consumers of Internet services
should be charged a usage fee reflecting the packet volume resulting
from their requests.
Teams:
Moderators: Mary Jo Torres, Terrence Ting, Sanjay Verma, Kentaro
Yamagishi, Yutaka Yasuda
Pro: Neeraj Gokhale, Douglas Hendrickson, Seiji Iwakaski,
Alex Liff, Edward Liu, Thomas Marsh, Anita Ng
Con: Gregory Penner, James Porter, Megan Ralston, Claudio
Rangel, LeMarque Sheppard, Beibei Song
Background Material
Top Ten Facts on Internet
- Internet started as a federally funded project to develop
fault tolerant systems for reliable communication during war in
1968.
- There have been two federally funded upgrades of the main
Internet backbone, one in 1986 And the other in 1988.
- Since the introduction of ISP's the average time of a call
within US has been increasing as opposed to the historical trend
whereby the average time of a call had been decreasing.
- Pricing among the regional, local and national ISP is still
a flat fee pricing system.
- There is no single backbone now, it is a conglomeration of
backbones developed by AT&T, MCI, Sprint, MERIT etc. These
backbones connect to each other through NAP's- Network Access
Points.
- Donna L. Hoffman
estimates that 28.8 million people in the United States 16 years
and over have access to the Internet, 16.4 million people use
the Internet, 11.5 million people use the Web, and 1.51 million
people have used the Web to purchase something.
- The current NSFNET backbone costs about $1 million per month
and carries 60 billion packets per month. This implies a cost
per packet of about 1/600 cents. If there are 10 million users
of the NSFNET backbone then full cost recovery of the NSFNET subsidy
would imply an average monthly bill of about 10 cents per person.
- In terms of file sizes, a 700 page book in ASCII is about
1 megabyte. On the other hand, a single non-compressed GIF image
is about half a megabyte and a compressed JPG image is about a
tenth of a megabyte. A 13 second compressed movie is over 4 megabytes.
- The traffic on the Internet has been increasing exponentially.
Internet experts believe that because of a deluge of video and
audio applications on the Internet, soon the peak time congestion
on the backbone would lead to unacceptable performance during
peak time.
- As of 1996, all federal funding for Internet backbone has
been stopped, and Internet is completely private enterprise now.
The Debate
Link
to pro side argument.
Link
to con side argument.
Summary of points raised in the debate
Issues | Pro Side
| Con Side |
View on congestion | Already congestion in the Internet, and a big potential of further congestion in near future because of new video, audio and broadcast applications.
| Average usage is only 5%, on the Internet backbone.
|
Effects of per packet fee pricing | Leads to efficient usage of Internet resources
| Discourages innovative audio and video applications
|
Does the tiered pricing structure reduce peak traffic?
| Yes | No, just shifts the peak
|
Cost Drivers of a packet based fee structure
| Cost of packet transmission | Cost to monitor packets
|
Impact on expansion of the net |
Providers will have money to expand the infrastructure and companies
will have incentive to invest in technology (to suppress data, etc) to make
downloads less timely for consumers.
|
Per packet fees will slow the growth of the Internet. Stifles
multi-media application development, and discourages software downloads,
push technology. Widens gap between haves and have nots. Web becomes tool
for big business. Jeopardizes borderless world dream.
|
Feasibility of Per Packet Fee Pricing |
Logistically similar to phone or electricity service. Sender pays
and packets are counted at first switch. Different rates for different
priority content. |
Administrative costs to support packet based model will be
overwhelming. Each packet will need to be traced to point of delivery.
User has no real sense of what sending a message will cost.
|
| | |
Reflections on the debate
- Congestion of local phone lines: One of the issues
not covered in the debate was the issue of connect time vs.
packet count. The reason why the issue of congestion has come
into prominence today is because local phone companies are facing
unprecedented amount of congestion due to ISP's charging a flat
fee. If the ISP is in the local calling zone for the subscriber
then he has no incentive to bring down his/her phone connection
with the ISP. A usage based pricing scheme that focuses on packet
count, would not solve the problem of idle open connections. This
is so because connect time and packet count are two different
measures that have no correlation with each other. One could have
an open line but have no packet traffic at all. Therefore a (packet)
usage based fee would not solve the problem of congestion at the
local phone lines. Time based fee seems to be the only fundamental
solution to this problem. ISP's have recognized this problem,
and as of today they are practicing non-monetary disincentives
e.g. AOL closes your connection if the connection has been idle
for more than 30 minutes.
- Innovation viz. a viz. per packet fee: It seems that
a pure usage based scheme might discourage the innovation in multimedia
and broadcast applications. This is so because the marginal benefit
of such applications might be less than the marginal cost. If
this is the situation then a hybrid scheme could be used which
has a flat component and a usage based component. The flat fee
would then subsidize the marginal cost so that the marginal benefit
is more than the MC. For example, the following cost structure
might be reasonable that would lower the marginal cost of multimedia
applications.
Pricing Scheme | Flat Component
| Per Packet Fee |
Pure usage based | 0$
| 0.06c |
Flat fee | 20$
| 0 |
Hybrid Scheme | 10$
| 0.03c |
- Smart Markets: Under the smart market scheme, users
will be charged a per packet fee according to the degree of congestion.
The price of the packet will be based on the highest unsuccessful
bid at the instant when the user tries to send a packet. As
a result, the marginal cost of using the Internet will be a function
of current congestion. This is a theoretically good scheme. If
we assume that all kinds of congestion situations would be encountered
over time, then every bid has a chance of succeeding at some point
in time. For example, a person who is not ready to pay anything
for Internet usage will be able to use it at 2am, when there is
no congestion on the net. Although this scheme is conceptually
attractive, it seems to have technical difficulties of implementation.
Debaters in Action (photographs)
The End