Quality differentiation tries to achieve a more efficient allocation of traffic during peak times by giving priorities to certain types of traffic. Economic theory predicts that product or quality differentiation increases total welfare. In many circumstances, this implies that ISPs’ profits also increase (see Economic first principles in section 4.1 for more detail). Thus, in general, quality differentiation has the potential to increase ISPs’ profits and should thus be analyzed as an element of a viable business model.
44 Assessment of a sustainable Internet model for the near future
Of course, as previously explained, quality differentiation can come in different shapes. Among the many different possibilities, the two stylized strategies described in the following can be considered as focal points. The first strategy would implement a dynamic best effort model for traditional services and a new flexible model for innovative services. The second model would implement a number of quality classes that content providers can choose from.
The first strategy, henceforth called “Best Effort Plus,” is inspired by the US debate on net neutrality regulation. There, the discussion currently evolves around a model where the best effort network as we know it would be preserved for traditional services and truly innovative services would enjoy some regulatory holidays.57 Thus, one more or less realistic future scenario for quality differentiation could be that there exists a uniform quality class of best effort for all “traditional” applications on fixed infrastructure. However, ISPs would have the ability to develop and market innovative new online services (such as e-health, or new entertainment and gaming options) outside the best effort class as long as those services do not interfere with the “continued development of Internet access services” (best effort).58 Consequently, those novel services would be marketed to end users independently of the general best effort Internet.
Of course the definition of what constitutes a novel service would be crucial in order to ensure that innovative new online services are not used to circumvent the non-discrimination rule for traditional services. When discussing the effects of such quality differentiation, it is assumed that a clear definition of what constitutes a traditional service and what constitutes an innovative new online service can be achieved.
The definition of quality outside the best effort network would be bilaterally negotiated between the ISPs and the content providers of novel services. In this setup, eyeball ISPs could offer individual and exclusive quality levels to developers of new services. However, it should be kept in mind that only a limited number of
57 The FCC has summarized the issues on which there appears to be consensus among stakeholders: First, ISPs should not be able to block lawful content. Second, there should be some form of protection against discrimination of content. Third, ISPs should be allowed to engage in reasonable network management as long as they are transparent about their network management practices. Whether or not specialized or managed services should be excluded from the regulation on broadband Internet access services and how to deal with mobile broadband access still appear to be open issues and are broadly discussed. See http://www.fcc.gov/Daily_Releases/Daily_Business/2010/db0901 /DA-10-1667A1.pdf.
58 This approach is for example also described in the joint proposal of Google and Verizon, see http://googlepublicpolicy.blogspot.com/search/label/Net%20Neutrality.
individual and exclusive quality levels can technically be provided for. Thus, in this future scenario, either a number of content providers of novel services use a limited number of different quality levels non-exclusively or a few providers have exclusive agreements.
In the second strategy, a business model might emerge where content providers or end users can choose among a number of pre-defined quality classes. We refer to this model as “Quality Classes.” Different quality classes would be defined on top of the current best effort level and every content provider or end user would have the ability to choose among different quality classes. Since all current and future content providers including those which are not directly connected to the eyeball ISP would be able to choose from different quality levels, it is likely that a form of standardization or interoperability process would emerge.
This strategy is inspired by the current ongoing industry activities – in particular, also within the mobile infrastructure industry – to define standards for different quality classes. According to industry sources, a number of ISP associations, such as, for example, the GSM Association, are already in the process of trying to define a number of different quality classes for the mobile segment.59 Due to the fragmented nature of the Internet, it appears that there are currently a number of independent standardization efforts. One major challenge of such a business model would be getting a sufficient number of relevant players to agree on certain standards. It appears that to date no single solution as to the number and definition of quality classes has emerged from the standardization efforts.
Whereas the content provider is likely to invest in higher quality classes for individual applications, the end user typically consumes a variety of different contents. Each of those contents might need a different quality of transmission in order to optimize the quality of experience for the end user. Thus, ideally, the consumer would purchase different quality levels for different types of contents and services that she consumes. However, if ISPs do offer different quality levels for different services to end users, this could potentially be a very confusing and complex marketing exercise. In contrast, the decision problem for the content provider is significantly simpler as he would choose a certain quality level for each of its services. Thus, in a scenario where end users pay for higher quality classes, they are going to be confronted with flat rates for quality classes that are optimized for a certain content consumption pattern.
59 See for example http://gsmworld.com/our-work/programmes-and-initiatives/ip-network ing/ipx_pci_trials.htm. According this information, the negotiations on the mobile industry (GSM Association) resulted in a system called IP Exchange which is currently in
“Pre-Commercial Implementation Trials.”
46 Assessment of a sustainable Internet model for the near future
The two described future quality scenarios share some similarities but differ also in a number of aspects. Table 5 summarizes the differences. Those two scenarios together with the uniform quality best effort standard of the status quo form the basis of the analyzed business models.
Table 5: Differences between quality scenarios
“Best Effort Plus” “Quality Classes”
Access to different qualities
Unlimited access to best effort
Limited access to higher qualities (only novel services)
Unlimited access to best effort
Unlimited access to higher qualities
Exclusivity Non-exclusive best effort Potentially exclusive higher quality classes
Non-exclusive best effort and higher quality classes
Standardization/inte roperability
Not at the core of the business model
Integral part of the business model
Source: ESMT CA.