3.104 As noted in the stakeholder responses above, there were also conflicting views relating to our proposal to limit the initial application of the Code to fixed ISPs with more than 400,000 subscribers. The key points of those who disagreed with this proposal were:
• It would be more appropriate to base the threshold on actual infringement levels;
• It is discriminatory to apply the Code only to large ISPs;
• The existence of a threshold will lead to the distortion of competition by encouraging subscribers to switch to other ISPs;
• Ofcom should provide certainty about how coverage will be extended in the future or even establish a mechanism for bringing new participants within scope in the future; and
• The Code should cover KCOM in the Hull area due to its monopoly status. 3.105 A number of respondents disagreed with our proposal to base the threshold on
subscriber numbers, claiming that a threshold based on actual observed levels of infringement by ISP (generated perhaps as part of a pilot scheme or other early experience of the notifications process) would be more robust and more in keeping with the intention of the DEA provisions.
3.106 We have considered this approach but do not think it is practicable. As far as we are aware there is no independently audited data on infringement levels by network which would provide a robust basis for determining on its own the application of the Code to ISPs on basis of actual levels of infringement. 35 In order to obtain such data, we would have to conduct a pilot scheme whereby ‘dummy’ CIRs are generated and then processed by ISPs in order to validate them. However, under the DEA
provisions no obligation to process CIRs exists unless an ISP meets the qualifying threshold, and so such a threshold could not be derived from a pilot scheme in this way without the voluntary, widespread co-operation of ISPs. It is doubtful whether such co-operation would be forthcoming but, in any event, such a scheme would in all likelihood add significantly to the overall time and cost of implementation and accordingly we have rejected it.
34
As set out in detail in the costs consultation, the cost-sharing scheme which underpins the implementation of the initial obligations is designed to enable qualifying ISPs to recover from qualifying copyright owners a proportion of the costs of compliance which are efficiently and
reasonably incurred. The costs consultation seeks to identify those costs. The current assessment of such costs in the costs consultation does not include the additional costs associated with matching IP addresses allocated via NAT technology. On the basis that such costs will be very high but will yield little benefit given the negligible risk of infringement via this type of access, we do not consider that these are costs which would be reasonably and efficiently incurred by a qualifying ISP. Accordingly, in the absence of reasonable means to match an IP to a subscriber, the qualifying ISP is not required to send a notification.
35
The Hargreaves Review of Intellectual Property and Growth noted the general lack of statistically reliable data on levels of copyright infringement (http://www.ipo.gov.uk/ipreview.htm).
32
3.107 In the absence of audited data sufficiently robust to establish independently a
threshold based on infringement levels by network, we think that subscriber numbers provide a reasonable proxy for likely infringement levels. As the premise of the DEA provisions was that online copyright infringement is a mass market problem,36 we believe it is appropriate to set the initial threshold on the basis of subscriber numbers, at a level that captures over 90% of subscribers to fixed internet access services.
3.108 We note that evidence submitted by copyright owners about the distribution of alleged online copyright infringement activity indicates that there is a broad correlation between the number of subscribers that an ISP has and the level of alleged infringements on their service. While we acknowledge the limitations of this data (specifically that it has not been independently audited), we consider it does provide support for the view that it is proportionate and reasonable to base the qualification threshold on subscriber numbers.37
3.109 With regard to the level of our proposed threshold, we need to make two clarifications. First, in the May 2010 Consultation we stated that the 400,000+ subscriber threshold would cover seven ISPs representing 96.5% of the residential and SME broadband market. This statement was based on incorrect information provided by one of the ISPs. Having corrected this error and included the most recent available data at the time of this statement, we estimate that the threshold covers six ISPs representing 93.5% of the broadband market.
3.110 Second, some respondents were unclear regarding the unit of measurement for the threshold. For clarity, the analysis underlying the proposals in the May 2010
Consultation was based on the number of broadband-enabled lines over which the ISP provides an internet access service to a subscriber (i.e. lines providing
broadband services using DSL, cable modem, FTTx or LLU technology) and not on subscriber numbers. We consider that this unit of measurement is appropriate since it is information that is readily available for all ISPs and is typically used by them to state the size of their subscriber base. We are not aware of any fixed ISPs which would satisfy a 400,000 subscriber threshold but not a threshold of 400,000
broadband-enabled lines over which its services are provided. In practice, therefore, this distinction would not affect which ISPs will be qualifying ISPs for the purposes of the application of the Code.
3.111 Therefore, in the interests of clarity at the time of publication, there are six ISPs that would be covered by this threshold:
• BT (including Plusnet) – 6.3 million;38
• Virgin Media – 4.4 million;39
• TalkTalk – 4.1 million;40
36
See for example the Government’s Impact Assessment for the Digital Economy Act
(http://webarchive.nationalarchives.gov.uk/20100511084737/http:/interactive.bis.gov.uk/digitalbritain/ wp-content/uploads/2010/04/Digital-Economy-Act-IAs-final.pdf).
37
See fn 22. This evidence was also cited in the second witness statement of Rachel Clark in defence of the application for judicial review made by BT and TalkTalk and the Government’s reliance upon it was endorsed by both by Parker J and the Court of Appeal.
38
BT results, 31 March 2012,
http://www.btplc.com/Sharesandperformance/Quarterlyresults/PDFdownloads/q412KPIs.pdf 39
Virgin Media results, 31 March 2012,
• Sky – 3.9 million;41
• Everything Everywhere (Orange and T-Mobile) – 0.7 million;42 and
• Telefonica O2 (including Be) – 0.6 million.43
3.112 We are not aware of any other ISPs that fall into the category of qualifying ISP at this time. As shown in the table below, even if we reduced the threshold significantly it would have only a small impact on the overall coverage of the scheme. For example, dropping the threshold to 50,000 lines would expand coverage to five further ISPs (none of which has more than 150,000 lines) and increase coverage of the market by only 2.3%.
3.113 These smaller ISPs would face proportionately higher costs to deliver their
obligations under the Code. The smallest ISPs, for which investment in automated processing may be uneconomic and unaffordable, might have to rely on less efficient and more costly manual processes to execute their obligations. The operating costs to both copyright owners and ISPs of sending CIRs under such manual processes would be significantly higher, potentially constraining or even reducing to negligible levels copyright owners’ demand for CIRs to be processed by those small ISPs. In the absence of robust data suggesting copyright infringement levels on the networks of these small ISPs is disproportionately high, compared to their market share, and taking account of the objective of effecting a change in mass market behaviour, it does not appear to us that it would be proportionate to include smaller ISPs within the scope of the Code at this time.
Table 1: Impact of different threshold levels on coverage of the scheme
Source: Ofcom.