Transcript Beyond Layers - Silicon Flatirons

Beyond the
Layered Model
J. Scott Marcus
Senior Advisor for Internet Technology, FCC
The opinions expressed are solely those of the author, and do not necessarily
reflect the views of the FCC or of the European Commission.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 1
Beyond Layers
 Several theorists have proposed a layered
approach to telecoms regulation, including:





Werbach
Whitt
Sicker/Mindel
Nakahata
Solum
 Diagnosis of problem generally correct.
 Underlying objectives are reasonable.
 What about the recommended cure?
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 2
Beyond Layers
 As a thought model, the layered approach
provides interesting insights.
 The Layered Model provides one useful
caveat: Beware obligations that cross layers.
 Otherwise, the Layered approach offers
surprisingly little guidance to the regulator.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 3
The Diagnosis
 The Communications Act as amended comprises
regulatory silos:
 Wired telecommunications (Title II)
 Wireless (Title III)
 Cable (Title VI)
 The convergence of computing and communications
renders those silos irrelevant.
 Broadband
 Voice over IP
 The mismatch of law to reality causes problems.
 Irrational conclusions.
 Regulatory arbitrage.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 4
Layered Approach
Content
Applications
Logical Transmission
Physical Transmission
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 5
Ostensible Goals of the
Layered Model
 Closer mapping to reality as all networks
converge to an Internet-based model.
 Straightforward regulatory categorization of
services.
 Regulation that responds naturally to
technological evolution and convergence.
 Avoidance of regulatory silos.
 Straightforward determination of regulatory
consequences.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 6
The Layers of the Internet
Layers interact with peer layers
Application
Transport
Network
Data Link
Physical
Network
Data Link
Physical
Application
Transport
Network
Data Link
Physical
Layers derive services from successively lower layers
Closer mapping to reality as
all networks converge to an
Internet-based model?
 Are all services based on the Internet today?
 Can we be sure that all services will be based
on the Internet in the future?
 Can we say with certainty that network
products and services will never evolve beyond
the Internet?
 Risk of regulatory hubris.
 So much for regulation that responds naturally
to technological evolution and convergence...
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 8
Hybrid Services
 The most challenging services represent a mix
of Internet and traditional elements:
 AT&T Internet-based calling card services
 Vonage
 Skype dial-out and dial-in
 How useful is the distinction between the
Logical and the Physical transmission layers
for such services?
 Challenges both correspondence to reality and
ease of categorization.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 9
Straightforward regulatory
categorization of services?
 Layers were intended to be an aid to specification, not
necessarily a detailed guide to implementation.
 Products and services routinely and systematically
violate layering.
 Layers are often used recursively (e.g. tunneling).
 If layers were tied to economic and regulatory
consequences, we might expect systematic
manipulation and arbitrage.
 Determination of layer is not notably easier than
determination of telecommunications service versus
information service.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 10
Avoidance of regulatory
silos?
 Old system had old silos.
 Wired telecoms
 Wireless
 Cable
 Implies asymmetric treatment for broadband?
 New system has new silos.
 Logical (Internet)
 Physical (Non-Internet)
 Implies asymmetric treatment for IP telephony?
 Risks needlessly or inappropriately violating
technological neutrality.
 Vertical silos -> horizontal silos
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 11
Straightforward determination
of regulatory consequences?
 Computer Inquiries declined to regulate enhanced services:
 Competitive entry would be straightforward, provided that any
underlying bottleneck facilities were available on nondiscriminatory
terms.
 Competition could prevent competitive harms.
 Promotion of innovation services for consumers.
 Internet services have been unregulated, not because they
constitute logical transmission, but rather as a consequence of the
prospect of competition.
 Economic tests can work in a converged world; semantic
definitional tests often fail.
 Triggering regulation based on an incidental consequence rather
than an underlying root cause is a step in the wrong direction.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 12
Straightforward determination
of regulatory consequences?
 Advocates regulation of last mile facilities because they
represent competitive bottlenecks.
 A lightning rod for objections.
 Must be balanced against other objectives:
 Encouragement of investment.
 Support for nascent technologies.
 Possible to reach this conclusion without recourse to the
Layered Model.
 The Layered Model offers little or no specific guidance
otherwise as to how to regulate within a Layer.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 13
Market Power in the Last Mile
 Most papers on the Layered Model argue for regulation
of last mile facilities because they represent bottleneck
facilities.
 Is this really true in all geographic areas?
 Is this really true for all time?
 What mechanisms does the Layered Model provide for
avoiding regulation in geographies where market
power is not an issue, or when it no longer is an issue?
 If market power were to emerge at a higher layer,
should we necessarily decline to take action?
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 14
Market Power in the Last Mile
 Stevens Report and Layered model seek to:
 Regulate underlying telecommunication services.
 Leave Internet ISP services unregulated.
 “Assess market power separately for each layer.” (MCI
comments on IP Enabled Services NPRM)
 Why not instead assess market power separately for each
market?




Layers may not correspond to markets.
Markets may not correspond to layers.
Classification of services into layers may be intractable.
Identification of markets, by contrast, could draw on existing
DoJ/FTC guidelines and longstanding jurisprudence.
 Economic metrics rather than semantic definitions.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 15
Market Power in the Last Mile
 Assume arguendo that it were deemed desirable to use
regulation to address bottlenecks that might exist in
conjunction with last mile facilities.
 A solution based on European principles leads much
more directly to the same conclusion:
 Explicitly identify the market using economic tests.
 Determine who, if anyone, has significant market power.
 Apply remedies (a minimally adequate set to address the likely
competitive harms) solely to those parties.
 Remove corresponding remedies where significant market
power is not present, or when it is no longer present.
 Provides far greater granularity and flexibility.
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 16
VoIP and the Layered Model
 The FCC’s public forum on VoIP (12/2003)
identified five key issues:





Intercarrier compensation
Funding for universal service
Lawful intercept
Access to emergency services
Access for those with disabilities
 Three scenarios:
 Red: Free World Dial (FWD)
 Blue: AT&T Calling Card
 Purple: Vonage
 WHAT does the Layered Model tell us to do?
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 17
An Alternative Layering
Content
Applications
Transmission
• Does this really add value?
• How would we regulate content differently
from applications?
• Is this perhaps already obvious?
February 13, 2005
Silicon Flatirons, Boulder CO - J. Scott Marcus
Slide 18