Based on the results in this thesis, the most important recommendations to Telenor are
General recommendations towards adopting an NNN regime
o Analyze the attractiveness of advertising and be sure that it is not high enough to make the competitiveness of the NNN regime reduce the overall profit of the enterprise from introducing the SS lane.
o Be certain of your competitive position in the value chain, and that the backbone ISPs will not eat away the benefits of an NNN regime. In general this should not be a problem if the backbone ISPs do not have a monopoly.
o Be sure that the IUs are dependent enough on CPs and QoS so that they are still willing to pay when the CPs are charged in the SS lane
General recommendations to persuade policy makers to allow the NNN regime
o It is very important to show that the attractiveness of advertising is sufficient enough for the IUs and CPs to benefit from the NNN regime.
o Focus on the following arguments, as proven in section 5.5.3 and 5.6.3
The overall social welfare always improves in the unregulated NNN regime
Investments, Innovation and average QoS always improves in the NNN regime
Assuming sufficient QoS in the backbone of the Internet, the access ISPs investments is the value creator of the ecosystem. Higher investments means more value to the whole ecosystem, and the access ISP will only claim a fraction of the extra value by itself
Even though the benefit for individual players is
ambiguous, all players benefit around the medium ranges in the scenarios
As the IUs become more and more dependent on the CPs and QoS, they increasingly benefit more from the NNN regime
The analysis confirms the three arguments proposed by the NNN-proponents
General recommendations to persuade policy makers to allow an unregulated and flexible network
o Focus on the following arguments, as proven in section 5.5.3 and 5.6.3
Splitting the network in two makes the network much less effective, doubling the overall congestion levels
Regulation might severely impair individual players, and especially the IUs are exposed when they are willing to pay more for investments than they gain in utility Too much regulation might reduce social welfare, and is
therefore potentially dangerous
Even though some regulation usually increases the social welfare, individual actors might be severely impaired. Even some regulation can therefore be dangerous if one aims to protect the interest of individual actors
o Bourreau et al. [2] have shown that the access ISPs have no incentives to sabotage unless the IAS lane unless the
attractiveness of advertising is sufficiently high. However, in when the attractiveness is sufficiently high, as shown in this thesis, the NNN regime is not beneficial to begin with, so given that the access ISPs prefer an NNN regime, sabotaging the IAS lane is not likely assuming the access ISPs only interests are to maximize their own profits
The results of this thesis are dependent on the assumption that the QoS of the backbone is sufficient so that no extra investments by the backbone ISP are needed in order to manage the extra traffic in the NNN regime. This is not unlikely, as the eyeball network is much more expensive and harder to build than the backbone network [24]. However, this is likely to vary from region to region, and where the QoS of the backbone is not sufficient they either have to accept a lower QoS than possible, or invest which will probably lead to an increase in prices for the CPs that use them.
The results are also dependent on the assumption of settlement free peering. On average this should be true as the access ISPs and backbone ISPs are equally dependent on each other. However, the difference between individual players can be substantial, meaning that many non-tier 1 ISPs have to pay extra for transit or paid peering in order to obtain the necessary increase in QoS in a NNN regime. This
analysis is in connection with the different actors, and thus differences within the type of actors is not part of this thesis. Intuitively the overall result should be the same. However, the smaller ISPs that have to pay extra for transit or peering are likely to benefit less from the NNN regime. On the other hand, ISPs that can demand payment for transit or peering would likely be better off in a model where this factor is included.
This model assumes that the SS lane is prioritized and that the packages are put first in line, while the IAS packages are next in line to the SS lane packages. There are other ways to manage the internet, which could alter the result of this analysis. However, it is not possible to make the QoS of the SS lane any better than what is assumed in this analysis, and thus, unless the ISPs sabotage, the QoS of IAS cannot be worse either. Regarding the concerns of the NN proponents, this analysis could be seen as a worst case scenario for the IAS lane, as long as the ISPs do not sabotage. Bourreau et al. have shown in a similar setting, but without E2E, that sabotaging IAS might be preferred in the NNN regime when the advertising rate, r, is sufficiently high [2]. Intuitively this is in agreement with our analysis since in this scenario the access ISP are worse off in the NNN regime because not enough CPs pay for the huge investment requirements.
Lastly, these results assume complete information, which means that the ISPs can set their prices with full knowledge of demand. That all players have complete information is most likely not realistic, and incomplete information generally in favor of the buyers when the sellers set the price, as is the case in this model [37]. The outcome in this setting, however, is not certain. Generally this means that in a real setting the IUs and CPs, the buyers, should be even better off than the sellers, as the backbone and access ISPs would not be able to extract their full potential margins because of the uncertainty. However, we could also expect this effect to be reduced by the fact that the ISPs do business with most of actors and individual identities within the type of actors and should thus be able to provide fairly good estimates to compensate for the lack of full information. Therefore, the results in this thesis should not be regarded as unreliable, but rather slightly optimistic on the access ISPs behalf.