Skip to Main Content
The Liability of Internet Intermediaries The Liability of Internet Intermediaries

Contents

The Liability of Internet Intermediaries The Liability of Internet Intermediaries
1

Overview 14.04

2

Blocking technology 14.09

2.1

IP blocking 14.10

2.2

DNS blocking 14.16

2.3

URL filtering 14.23

2.4

Deep packet inspection 14.30

2.5

Hybrid models 14.34

2.6

Circumvention methods 14.35

3

Injunctions for copyright infringement 14.47

3.1

Overview 14.50

3.2

‘Service provider’ 14.54

3.3

Primary infringement 14.56

3.4

‘Whose services are used’ 14.58

3.5

Actual knowledge 14.68

3.6

Examples of application 14.76

3.7

Proportionality 14.112

3.8

Site blocking injunctions 14.145

4

Injunctions for other types of wrongdoing 14.149

4.1

Trade mark infringement 14.150

4.2

Defamation 14.156

4.3

Confidentiality and privacy 14.164

4.4

The administration of justice 14.167

14.01 A blocking injunction is an order requiring an internet intermediary to implement technical measures directed at preventing or disabling access to a specified internet location. The essential purpose of such an order is to restrict the ability of individuals located in a particular jurisdiction to access or download material the subject of the order using the respondent’s service. Although theoretically available against any service provider, blocking injunctions are most commonly sought against ISPs, who possess a comparatively high degree of control at the network layer and may operate systems that can drop or re-route packets based on criteria including the destination of a request. Blocking injunctions require the respondent to adopt these technical measures in an attempt to prevent others from using its service to infringe the applicant’s rights. Section 2 provides an overview of the most common technical measures employed by ISPs.

14.02 This chapter discusses the availability of blocking remedies against intermediaries in the United Kingdom. Such remedies are presently available on a statutory basis where a third party has used a service provider’s service to infringe copyright, performers’ right, or database right. Relying on those provisions, more than 25 orders have been made against the five main British ISPs requiring them to block access to approximately 120 websites (and thousands of URLs for mirror websites, proxy servers, and successor URLs).1 Blocking orders have been obtained by applicants representing the film, music, publishing, and sports industries, and the number of applications and blocked URLs continue to grow exponentially. Section 3 discusses the emerging law and practice in relation to these orders, which relate mainly to BitTorrent tracker indices, and film and television streaming services.

14.03 Trade mark proprietors have successfully argued that an equivalent remedy should be available in relation to trade marks—and perhaps other types of intellectual property rights—by a combination of article 11 of the Enforcement Directive and section 37(1) of the Senior Courts Act 1981.2 As at November 2015, only 2 orders had been made and 8 websites blocked pursuant to this jurisdiction. Other scholars have suggested that blocking injunctions are available to protect any kind of legal right as an incident of the courts’ equitable protective jurisdiction. These arguments are discussed in section 4.

14.04  Purpose of blocking. Blocking access to an internet location aims to prevent or bring to an end infringements of a person’s rights involving material accessible at that location. Its essential purpose is to make access to unlawful or tortious material from a designated jurisdiction substantially more difficult, which may dissuade individuals from seeking to access that material or (where it is available lawfully) encourage access to lawful sources of the material. Secondarily, the remedy aims to discourage the operators of websites targeted by blocking measures from accessing that audience and continuing to disseminate infringing material.

14.05  Target of blocking. Blocking is intended to increase the expected cost of infringement (eg by making it harder to find, access, or disseminate blocked material). The target of a blocking remedy is neither the primary author nor host of the material complained of, but rather a local intermediary which carries the tortious material to its users within the jurisdiction. Whether blocking is effective in practical terms depends on a number of variables:

(a)

First, the type of material being blocked: blocking is naturally more suited to material that, because of its file size or centralised distribution, is difficult to transmit or access by other means.

(b)

Second, the nature of the audience for blocked material: blocking is generally more effective to deter those who would infringe as a matter of convenience than those who would engage in deliberate wrongdoing or actively seek out material.

(c)

Third, the other remedies that will be applied: blocking tends to be more effective when used in combination with other methods of enforcement, such as notice-and-takedown and de-indexing orders.

14.06  Comparison with notice-and-takedown. Outright removal of material doubtless offers several advantages over blocking access to material. First, removal is universal: once data have been removed, they are inaccessible regardless of which ISP or other service provider is used. Blocking is specific to the targeted service, meaning that separate blocks must be introduced by each IS intermediary within each jurisdiction. Second, removal destroys the original material complained of; blocking allows the material to remain, as a result of which it might be further disseminated by others. Third, subject to the risk of mirror websites or repeat postings, removal is absolute; blocking is imperfect and can be circumvented by various technical means.

14.07  Territorial rights. Blocking can be well suited to the protection of territorially circumscribed interests, such as intellectual property rights and reputation. Rather than remove material whose dissemination may be lawful in some jurisdictions, blocking is a more targeted remedy that aims to prevent access to the material from only those places where doing so would infringe the applicant’s rights. In this way, blocking avoids some of the extraterritorial spillover effects that would result from a national court ordering global removal of internet content whose lawfulness abroad has not been determined.

14.08  Sources of law. In the United Kingdom and other member states of the European Union, blocking remedies are derived from two main sources in European law: first, article 8(3) of the Information Society Directive, which obliges member states to make available injunctions against intermediaries whose services are used by a third party to infringe copyright or a related right; and second, article 11, final sentence, of the Enforcement Directive, which requires a similar remedy for infringements of any intellectual property right.3 The interpretation and application of these provisions in national law is considered in sections 3 and 4. The approach in other member states is the subject of chapter 15.

14.09  Extent of blocking. Website blocking is not a binary concept. The complexity of modern communications networks means that blocking is rarely as simple as switching access ‘on’ or ‘off’. Because internet networks are profluent by design, occluding access to network resources requires a diverse set of technologies to operate at different layers of network architecture. Some of these technologies achieve the same result by different methods; others are complementary in that they each close off different methods of access. Short of severing a connection at the physical layer, access can never be entirely prevented: the variables are for whom access will be made more difficult and how much more difficult it will be made. This section provides an introduction to five basic filtering models, and their main vulnerabilities. It is not intended to be comprehensive or technically detailed, but aims to provide a basic appreciation of the available technology and its limits.

14.10  IP addresses. The crudest form of website blocking instructs ISPs’ networks to discard requests for an identified Internet Protocol (‘IP’) address. All internet hosts are assigned an IP address which uniquely identifies them (or, in the case of Network Address Translation, their local subnet)4 to other computers, similar to a postal address or (in the case of a shared server) a postcode.5 Network communications from one computer to another are processed by routers which sub-divide messages into data units called ‘packets’ and forward them to the destination IP address, where they are reassembled and processed in order to reconstruct the original message.

14.11  Implementation. IP address blocking is usually implemented in two ways. Under the first method, routers are instructed to discard messages destined for a particular IP address range. This is done by creating an entry in the routing table which associates that IP address block with a ‘blackhole route’ (such as 0.0.0.0) or the loopback address (127.0.0.1), causing the packets to be dropped or redirected.6 This is commonly done to mitigate denial of service attacks. Under the second method, an entry is added to a border router which rejects all outbound requests for the IP address.7 The effect of both methods is to prevent communications with the targeted remote host, to the extent that the destination IP address is visible to network infrastructure. IP blocking is commonly implemented in combibation with a second layer of filtering technology (such as DPI or URL filtering) in order to reduce the risk of false positives.8

14.12  Accuracy of blocking. Although IP blocking is simple to implement and has a low impact on network performance, it suffers from a number of limitations. First, because IP blocking blocks access to an entire host rather than individual webpages or URLs, it will often cause overblocking of non-tortious content. As many as 87.3 per cent of websites share an IP address with at least one other website, while over 66 per cent share with 50 or more other websites.9 When blocking occurs, these websites (which may not be publishing tortious material) will also be affected. For example, in 2011 an Argentinian court ordered ISPs to block the IP address of a webmail service and blog (leakymails.blogspot.com, which was hosted at 216.239.32.2). This caused around one million blogs hosted by Google’s Blogger service (which also resolves to 216.239.32.2) to become inaccessible in that country.10

14.13 False positives can be difficult to predict because a comprehensive list of the websites sharing an IP address11 cannot be reliably identified in advance. IP blocking may be appropriate in some cases—eg where tortious content is hosted on a dedicated server, so that there is a smaller risk of false positives—and the risk of error can be reduced in a hybrid or two-stage blocking system. However, in most cases IP blocking offers insufficient granularity on its own to function as an accurate method of blocking.

14.14  Ease of circumvention. Second, IP blocking is ‘straightforward’ for website operators to circumvent.12 They can shift hosts, change IP addresses, offer tools to facilitate encrypted access, or use any other number of techniques. Many websites associated with file-sharing use a low Time to Live (‘TTL’) value in their DNS records to ensure that any IP address changes will propagate across the internet within a matter of minutes.13 The susceptibility of IP blocking to operator-directed circumvention makes this method problematic when used to block access to materials that the operator is likely to have a strong motivation to re-disseminate. Of course, most blocking orders provide for updates to be notified to the respondent to ensure that such replacement IP addresses are also blocked.

14.15  Unintended consequences. Third, activating an IP block requires low-level changes to network routing hardware. This requires any prudent ISP to conduct testing and impact assessment before rolling out a block on their customer networks. Because the effect of blackhole routing on complex network architecture can be difficult to assess in isolation, there is a risk of downtime and delay for customers seeking access to lawful information. Because of the propensity for circumvention by shifting IP addresses, new blocking ranges are likely to be added frequently, which further increases the risk of interference.

14.16  Domain Name System. The Domain Name System (‘DNS’) is responsible for translating human-friendly domain names, such as ox.ac.uk, into the IP address of the corresponding server, namely 129.67.242.155.14 Besides convenience, domain names ensure stability by allowing application-layer services to shift locations on a network without requiring network addresses to be updated.15 Domain names are essentially useful topological abstractions.

14.17  DNS resolution. The process of DNS resolution involves recursively querying authoritative ‘nameservers’ until the correct IP address is returned. Initially, a client web browser contacts one of 13 authoritative root servers—core network nodes with a known IP address that contain lists of other authoritative DNS servers for a particular Top-Level Domain (‘TLD’) zone (such as .com, .net, or .org)—which respond with data about the authoritative servers for a particular Second-Level Domain (‘2LD’) zone (such as .co.uk or .com.au). The process is repeated until the destination host is found. The internet’s root servers are a critical part of the internet’s network layer and, although the 13 servers each offer considerable redundancy and geographic dispersal via distributed mirrors, they have been identified as vulnerable points in the internet ecosystem.16

14.18  Implementation. DNS blocking instructs an ISP’s local DNS server to return false results for blocked material. Instead of returning the correct IP address corresponding to a domain name, the ISP’s DNS server will return a ‘does not exist’ error code or (under some implementations) redirect users to a specified webpage which explains that the request has been blocked. It does so by intercepting a user’s DNS request and inserting a forged answer which points to the new webpage. This method of blocking is usually combined with DPI and has been widely used in China and the Middle East.17

14.19  Effect on architecture. DNS blocking departs from the end-to-end architecture of the internet by making the same URL mean different things in different contexts. URLs are designed to be universal and unique, meaning that a single resource is identifiable by the same string of characters worldwide—irrespective of the end user’s ISP or country of residence. This may have wider consequences for software which relies upon IP functions at the network layer. DNS blocking is often implemented incompletely or incorrectly, which adds to these problems.18

14.20  Potential obsolescence. DNS blocking takes advantage of a vulnerability in the DNS security architecture. By default, DNS does not authenticate data contained in nameservers, meaning that a malicious party can forge answers to DNS queries and redirect unwitting users to an imitation website (typically to carry out a phishing scam).19 This is, in effect, how DNS blocking operates. One side effect is that this precludes the use of security improvements such as DNS Security Extensions (‘DNSSEC’). DNSSEC is designed to improve the security of DNS by requiring records to be cryptographically signed by authoritative nameservers.20 The objective is to prevent DNS poisoning and other man-in-the-middle attacks on DNS lookups. DNSSEC would, once implemented, prevent these attacks by requiring each authoritative nameserver in the DNS chain to sign its DNS query responses. If the encryption keys do not match, the response is presumptively forged and the request will be repeated.21 Unfortunately, this behaviour is impossible if blocking requires ISPs to insert forged DNS query responses. That is, in other words, ‘the very behaviour DNSSEC is designed to detect and suppress’.22 Although it would still be possible to ignore the DNS request completely, users would be unable to determine whether a webpage was inaccessible because it was blocked or for any number of other reasons.

14.21  Ease of circumvention. DNS blocking is particularly vulnerable to circumvention by any of the methods described in section 2.6. Users can easily change their DNS servers to evade blocking or can access materials by typing their IP addresses directly into a web browser. It is conceivable that directories of IP address mappings for blocked resources would appear soon after any blocking order was implemented. DNS blocking also fails to deal with non-web content and protocols, such as File Transfer Protocol (‘FTP’), Usenet (‘NNTP’), email, and peer-to-peer (P2P) file-sharing. These other protocols do not rely on DNS lookups and cannot be controlled using DNS blocking.

14.22  Scope of blocking. The granularity of DNS blocking is low. It blocks access to all parts of a domain zone (which may be an entire domain name such as blogspot.com or a sub-domain such as gossipblog.blogspot.com) which may include unrelated content. This is especially true of a domain zone like twitter.com, which may host a multitude of legitimate content in addition to any tortious material—all of which would be blocked by a zone-wide entry. Given these problems, it is unsurprising that both Ofcom23 and the White House24 concluded that DNS blocking was unfeasible for many kinds of material.

14.23  Uniform Resource Locators. A URL uniquely identifies a website resource, which might be a TLD, a specific sub-page, an image, or another file. URL filtering requires ISPs not to serve requests for a specific URL. This allows a high degree of granularity, such as http://twitter.com/#!/SuperInjunctionFeed/status/1234567890, rather than a DNS block for twitter.com or an IP block for 104.244.42.129. This allows the risk of overblocking to be minimised. It is also possible to block higher-level URLs such as directories or entire domain names; for example, URLs beginning in http://twitter.com/#!/SuperInjunctionFeed, which would block access to all tweets by the Twitter user ‘SuperInjunctionFeed’. Although applicants are likely to seek orders of wider scope, URL filtering can be tailored more closely to the precise location of tortious material.

14.24  Scope of blocking. URL filtering suffers from several limitations. First, like DNS blocking, it only blocks web traffic using the Hyper Text Transfer Protocol (‘HTTP’). Protocols such as BitTorrent, which do not require the use of URLs, cannot be filtered using this method. Accordingly, many types of tortious material may migrate beyond the reach of URL-based mechanisms.

14.25  Encrypted traffic. Second, complexities arise in respect of HTTP Secure (‘HTTPS’) requests, which is a protocol used for encrypted web transmissions. Filtering web traffic secured in this way is currently only possible using a proxy server, which effectively performs a man-in-the-middle attack using a trusted cryptographic certificate to decrypt, inspect, and re-encrypt requests before forwarding or dropping them as dictated by the blocking policy.25 This carries a significant performance and security cost and is therefore not widely employed outside military and intelligence applications. The alternative is not to inspect HTTPS traffic at all, which would make it trivial for website operators to evade URL filtering.

14.26  Two-stage URL filtering. Third, some implementations of URL filtering rely on proxy servers (which may be transparent or explicit) for unencrypted traffic. When a request is made for a host which matches a suspicious IP address (as defined in the ISP’s routing tables), it is redirected to a proxy which checks the exact URL against the blocklist. If the requested URL matches a blocked URL, the query is dropped or redirected. If no match is found, the query is rerouted back to the internet. This process introduces two new problems. Principally, when a URL on a popular website (such as Wikipedia or Facebook) is blocked, all requests for the associated IP addresses will be marked as suspicious, which causes severe performance degradation as the proxy server becomes overloaded. Second, when traffic is rerouted via a proxy, it can change the source attributes of packets if not configured correctly. This occurred when a Wikipedia article and image were added to the Internet Watch Foundation (‘IWF’) blocklist of child abuse material in December 2008. All inbound requests to Wikipedia from the UK appeared to come from a handful of proxy IP addresses corresponding to ISPs’ blocking servers, which made it impossible for Wikipedia to determine whether anonymous edits to its articles were coming from the same or different people.26 British users were therefore prevented from editing Wikipedia during this period. In other words, URL filtering caused users to be inconvenienced by ‘two mutually incompatible content regulation systems’.27

14.27  Circumvention by website operator. Fourth, if a website operator wishes to evade a URL filter, they can simply change the URL or domain name at which blocked materials are displayed. For example, suppose the owner of a tortious posting which is hosted at http://celebritysecrets.blogspot.com/2012/03/john-smith-photos.html receives notice of a blocking order. It would be trivial to alter the sub-page by changing the date and name of the posting (eg to /2012/02/js-photos.html) which would evade the block. Notwithstanding any possible liability for contempt, the posting would not be filtered until the new URL was added to the blocklist. It was for this reason that Ofcom concluded URL-specific filtering was an ineffective method of blocking:

It is our current belief that the blocking of discrete URLs, or web addresses, is not practical or desirable as a primary approach. Infringing website operators can readily change the structure of a websites, particularly commonplace database driven websites. We therefore recommend that if site blocking is adopted it should be implemented at a domain level.28

14.28  Potential improvements. Solutions to these problems do exist. For example, an ISP could update filters to account for the changed URLs or use a system of wildcard matching . However, this is likely to introduce an additional processing load on the ISP’s systems. A more flexible model is reflected in the current approach of the English courts, which is to require ISPs to block access to any updated URLs notified by the applicant to the implementing service provider from time to time. This permits a large degree of flexibility and ensures that orders can be updated to react to operator circumvention.

14.29  Stateful inspection. Although stateful URL filtering was once expensive, several new URL blocking appliances use the Border Gateway Protocol (‘BGP’) to perform in-line inspection of URLs without the need for a proxy server; although implementations vary, this tends to be faster, cheaper, and less likely to affect legitimate traffic. New Zealand has implemented a filtering scheme for criminal internet content which uses this technology.29 URL filtering remains the most popular method of website blocking, and forms the basis of the highly successful IWF blocklist.30 That blocklist is distributed to around 60 United Kingdom ISPs, mobile carriers, and filter software producers. All major ISPs have implemented the blocklist since 2007. Although the system has been criticised for its lack of procedural transparency31 and vulnerability to evasion,32 it appears that next-generation URL filtering systems may offer greater effectiveness without many of the downsides of other forms of filtering.

14.30  Packets and payloads. A fourth generation of blocking techniques inspect the payload (content) of data transmissions to determine whether they relate to blocked material. This involves examining the content of each packet of data which is transmitted by an ISP. Payloads can be examined at various degrees of specificity: the most basic ‘shallow’ techniques determine the protocol, port, and sometimes data type. Advanced forms of ‘deep’ packet inspection (‘DPI’) can be complex and offer high levels of detail about the content of a communication by examining patterns in payload data and traffic flow. This information can then be used to determine whether a request is destined for a specific URL or IP address and drop or forward the packet accordingly.

14.31  Performance impact. DPI offers a wider breadth of coverage than URL filtering or DNS blocking because it applies to all forms of traffic and operates independently of proxy and DNS servers. It is also highly granular, in that it can detect and block requests for specific URLs and pages. However, DPI suffers from four main disadvantages. First, the need to capture, analyse, and reassemble traffic flows introduces a network bottleneck that can significantly degrade network performance. The more detailed the analysis, the greater the impact. In some cases, degradation has been estimated at as high as 44 per cent.33 Preventing performance degradation requires an ISP to invest in very substantial network infrastructure, which must be continually scaled to meet customers’ growing demand for bandwidth.

14.32  Ease of circumvention. Second, DPI is vulnerable to circumvention using similar techniques to IP blocking. In particular, virtual private networks can be used to obfuscate or encrypt the payload of communications in ways that make it impossible to perform content analysis. Using current technology, it is exceedingly difficult to carry out DPI on encrypted payloads without access to military-grade hardware or backdoors.

14.33  Universal monitoring. Third, DPI requires ISPs to inspect and monitor all internet traffic. This carries a significant privacy risk in that every outgoing request by a user will be examined rather than just those which relate to blocked content. As Doctorow argues, blocking in this form ‘is inseparable from surveillance’.34 Finally, DPI is generally much costlier for ISPs to implement, since it requires a large capital investment in new equipment and (as noted in paragraph 14.31) ongoing upgrades in order to ‘scale’ that equipment to keep pace with network utilisation.35

14.34  DPI appliances. Several models of blocking use a combination of IP address or DNS blocking and DPI. For example, many off-the-shelf DPI appliances conduct daily lookups of destination IP addresses associated with blocked URLs, then use BGP and IP blocking to determine whether a URL request corresponding to that host also matches the blocked sub-page.36 This combines the accuracy of URL blocking with the speed and simplicity of IP blocking. Other models combine DNS blocking with DPI or URL blocking. The overall advantages and disadvantages of blocking technologies are summarised in Table 14.1.

Table 14.1
Comparison of blocking methods
MethodDelayCostEfficacyEase of evasionPerformance impactRisk of overblocking
IP addressMediumLowLowEasyMediumHigh
DNSFastLowLowEasyMinimalHigh
URLMediumMediumMediumMediumMediumLow
DPISlowHighHighMediumSevereLow
DNS + URLMediumMediumMediumMediumMediumMedium
DNS + DPISlowHighHighMediumMediumMedium
MethodDelayCostEfficacyEase of evasionPerformance impactRisk of overblocking
IP addressMediumLowLowEasyMediumHigh
DNSFastLowLowEasyMinimalHigh
URLMediumMediumMediumMediumMediumLow
DPISlowHighHighMediumSevereLow
DNS + URLMediumMediumMediumMediumMediumMedium
DNS + DPISlowHighHighMediumMediumMedium
Source: Ofcom
Table 14.1
Comparison of blocking methods
MethodDelayCostEfficacyEase of evasionPerformance impactRisk of overblocking
IP addressMediumLowLowEasyMediumHigh
DNSFastLowLowEasyMinimalHigh
URLMediumMediumMediumMediumMediumLow
DPISlowHighHighMediumSevereLow
DNS + URLMediumMediumMediumMediumMediumMedium
DNS + DPISlowHighHighMediumMediumMedium
MethodDelayCostEfficacyEase of evasionPerformance impactRisk of overblocking
IP addressMediumLowLowEasyMediumHigh
DNSFastLowLowEasyMinimalHigh
URLMediumMediumMediumMediumMediumLow
DPISlowHighHighMediumSevereLow
DNS + URLMediumMediumMediumMediumMediumMedium
DNS + DPISlowHighHighMediumMediumMedium
Source: Ofcom

14.35  Overview. No method of blocking is perfect. In practice, any user possessing sufficient determination can access blocked materials using several widely known methods, about which information is readily accessible online. Evasion strategies exhibit varying degrees of cost and technical difficulty. Many of them are vulnerable to countermeasures which can be applied, but a non-zero circumvention rate is inevitable. Because it is important for courts and practitioners to appreciate the technical limitations of blocking remedies, the main methods are outlined below.

14.36  VPNs. A Virtual Private Network (‘VPN’) offers an encrypted tunnel between two points on the internet. It operates by encapsulating page requests and transmitted information in a secure format which is unreadable by intermediate systems and forwarded to the VPN exit node. An IP blocking system sees only a request for the IP address of the VPN, but not the underlying request for blocked content, which is processed and routed via the VPN server. A DNS blocking system is completely ineffective since the connection is usually established with the VPN server’s IP address directly. A URL filtering system cannot determine what URL is being requested because the content of the request is known only to the VPN server. Even DPI-based methods are blind to the content of the communication, which is strongly encrypted. In effect, the VPN server becomes the user’s de facto ISP and supplants their ordinary gateway to the internet. This can be used to evade blocking where a VPN server is selected which is located in another country where no blocking order applies.

14.37  Forward proxies. A proxy server acts as a gateway between a client computer and a remote host. When a proxy is enabled, requests for data are routed via the proxy, which can be located in any country. In the most common configuration, a forward proxy, requests are serviced in such a way that they are made using the ISP of the proxy server rather than the end user. Some gateways, known as anonymous open proxies, enable anyone to route data in such a way that it is possible (though not always reliable) to conceal the IP address of the user who is requesting the data. However, unlike a VPN, traffic is not encrypted between the proxy and the client (unless provided for in the underlying protocol, such as HTTPS).

14.38  Web-based proxies. Other proxy servers are web-based, meaning that they allow users to request a particular URL from the proxy’s website without any additional software configuration. Many popular web browsers, such as Opera and Google Chrome, include web-based proxies by default or via extensions. Some even convert the webpage content into email format and send it to the requesting user using SMTP.

14.39  Lawful use of proxies. Proxy servers have many legitimate applications within organisations, universities, and residential networks.37 However, internet users in countries within which material is blocked can access that material using a proxy server located in a country where no block has been implemented. Because users’ domestic ISPs do not know to treat traffic from the proxy server as falling within the scope of the block, that material would be sent back to users as if it were a response from a normal, unblocked resource. Short of also blocking access to every proxy server, it is difficult using IP, DNS, and URL-based systems to prevent users from connecting to proxies.

14.40  Automatic caching by gateways. Copies of most webpages are stored in the caches of search engines, internet archives, and news aggregators. Internet users wishing to access blocked material may consult these resources by searching for the blocked material and clicking on the ‘cached’ hyperlink instead of accessing the page directly. The resulting webpage is delivered from the search engine’s servers rather than the original host. It would be difficult to block cache services without either (1) also blocking a wide range of legitimate cached content, or (2) preventing access to the search engines themselves.

14.41  Tor and other networks. Anonymising networks route users’ traffic through an international network of participating nodes for the purpose of obfuscating the origin and destination of content requests. Although these networks are sometimes prone to security flaws, they are generally effective to circumvent territorial website blocking for similar reasons to VPNs. They tend to be more difficult to configure and slower to use, but the required technical expertise is diminishing.38 The Onion Router (‘Tor’), a security network originally developed by the United States Naval Research Laboratory, is the most popular example.39

14.42  HTTPS. Even where the contents of subscriber traffic is encrypted using a protocol such as HTTP over Secure Sockets Layer (‘HTTPS’) or Transport Layer Security (‘TLS’), the communications metadata associated with page requests will remain visible to network infrastructure. This is necessary in order for the requests to be routed to their intended destinations. Accordingly, even if traffic is encrypted, it is normally possible to apply IP address and DNS blocking techniques based on the destination of a request.40 However, because it is not possible to examine the contents of the encrypted payload, such requests cannot normally be blocked using URL filtering, DPI, or hybrid systems. Absent military grade hardware (or a backdoor), such systems cannot decrypt the contents of encrypted communications, and so cannot determine the URL being accessed. To avoid this, some ISPs have previously blocked encrypted traffic altogether, but this raises obvious security concerns.41

14.43  Site-specific tools. Some extensions for popular web browsers allow users to circumvent blocks on specific services transparently. For example, Access Flickr permits access to the popular photography-sharing website Flickr.com in Iran.42 In the United Kingdom, a circumvention tool was released shortly after the High Court ordered BT to block access to newzbin.com. A group known as ‘TeamRDogs’, which purported to be affiliated with Newzbin2, released a tool called ‘Newzbin Client’, which used encryption to circumvent the website blocking order made against BT in July 2011.43 Soon afterwards, the developers announced that a new version of their tool would also support circumvention of other countries’ blocks on thepiratebay.org.44

14.44  Non-web traffic. Traffic carried between clients over non-web protocols is intrinsically difficult to block. IP blocking is largely ineffective against such traffic because peer IP addresses change rapidly and may be used for both tortious and lawful transmissions. DNS blocking is entirely ineffective because these protocols make direct requests for IP addresses without resort to hostnames or URLs. URL filtering is ineffective for the same reason. Although DPI-based techniques can be used to filter P2P traffic, it is often difficult to determine from a single packet whether the file being exchanged is tortious or not, and reassembling transmissions entails high performance overheads.

14.45  Definition. A darknet is part of the internet which is private and hidden to search engines and other ‘surface’ web traffic. They tend to use decentralised, peer-to-peer topologies to share access to resources within a closed group of trusted and typically anonymous users. Some networks use a system of URLs ending in suffixes which have no equivalent on the public internet, such as .onion and .bitnet, which can only be accessed using anonymising proxies. Material is hosted by other users within the darknet, transmitted via encrypted tunnels and cannot be blocked using IP, DNS, URL, or most DPI blocking systems.

14.46  Second devices. Most internet users have access to a second, unblocked internet connection in the form of their smartphone’s cellular data service. Historically, those services tended to be slower than conventional broadband and subject to more restrictive data transfer caps. However, for low-bandwidth content (such as e-commerce and text materials), mobile access can allow users to circumvent blocking on their residential internet connection to the extent that access to the material is not restricted by their mobile network operator.

14.47  Statutory basis. Sections 97A and 191JA of the Copyright, Designs and Patents Act 1988 (‘1988 Act’) create statutory blocking remedies consequent upon a finding that a third party has used the respondent’s services to infringe the applicant’s copyright or performers’ right, respectively. Section 97A(1) provides:

The High Court (in Scotland, the Court of Session) shall have power to grant an injunction against a service provider, where that service provider has actual knowledge of another person using their service to infringe copyright.

14.48  Relevant provisions of EU law. Despite being enacted in 2003, claimants did not begin utilising section 97A until relatively recently. The statutory remedy created by section 97A is a final injunction which is, in substance, a remedy modelled on article 8(3) of the Information Society Directive.45 Article 8(3) provides:

Member States shall ensure that rightholders are in a position to apply for an injunction against intermediaries whose services are used by a third party to infringe a copyright or related right.

14.49  Elements of a claim. To succeed the applicant must satisfy four elements. First, the respondent must be a ‘service provider’, as defined in chapter 12.46 Second, there must have been an infringement of the applicant’s copyright (or performers’ right) by a third party. Third, the infringement must have been carried out using the respondent’s service. Fourth, the respondent must have actual knowledge of such infringing use by the third party.

14.50  Legislative purpose. The legislative purpose of section 97A is essentially the same as that of article 8(3).47 Both provisions are directed at an intermediary ‘who carries a third party’s infringement...in a network’,48 even where the intermediary is not liable for doing or authorising the doing of the infringing acts. Article 8(3) is, in one sense, the quid pro quo for exempting intermediaries from liability for acts of temporary reproduction under article 5(1)(a) of the Information Society Directive: the latter provision clarifies that intermediaries will not face primary liability for transmitting infringing copyright materials in a network between third parties. However, article 8(3) provides that intermediaries may be obliged to bring such transmissions to an end by means of a blocking injunction.

14.51  Underlying policy. The policy underlying article 8(3) is expressed in recital (59) of the Information Society Directive, which provides:

In the digital environment, in particular, the services of intermediaries may increasingly be used by third parties for infringing activities. In many cases such intermediaries are best placed to bring such infringing activities to an end. Therefore, without prejudice to any other sanctions and remedies available, rightholders should have the possibility of applying for an injunction against an intermediary who carries a third party’s infringement of a protected work or other subject-matter in a network. This possibility should be available even where the acts carried out by the intermediary are exempted under Article 5. The conditions and modalities relating to such injunctions should be left to the national law of the Member States.49

14.52  Identifying the least cost avoider. Recital (59) makes clear that article 8(3) reflects a policy of granting injunctions against intermediaries where they are the least cost avoiders of third parties’ infringements. That policy is independent of whether or not the intermediary would otherwise be primarily liable for the infringements (injunctions must be available even where the acts of the intermediary fall within an article 5 exemption). There is ongoing debate about whether, properly understood, recital (59) is a limiting criterion or simply an expression of the rationale for article 8(3) if the policy underlying blocking injunctions is that an intermediary should be ‘best placed’ to bring infringements to an end, they may be inappropriate in those cases where there is another party who is better placed to do so or, in other words, where enforcement action by another intermediary would be cheaper and at least as effective.

14.53  Comparison to substantive liability. The scope of section 97A is therefore much broader than doctrines of authorisation and joint liability under English law, since it is not necessary to show that the service provider has itself infringed copyright or performers’ right. A number of cases have considered the application of section 97A in rapid succession. These are discussed below by reference to each of the four requirements identified above.

14.54  Definition of ‘service provider’. Section 97A adopts the conventional definition of a ‘service provider’ from the 2002 Regulations. The meaning of that term is discussed at length in chapter 12. Virtually all internet intermediaries will satisfy the definition, including in particular ISPs, which are the most common respondents to applications for blocking injunctions. Additionally, it seems arguable that search engines, social networks, and other application-layer services would satisfy the definition of ‘service provider’, for the reasons discussed in chapter 12. However, whether they satisfy the other requirements of section 97A is yet to be established. Because the concept of a ‘service provider’ is arguably wider than the concept of an ‘intermediary’ (as used in articles 12 to 14 of the E-Commerce Directive), section 97A is of potentially wider scope than article 8(3), which refers to ‘intermediaries’.

14.55  Definition of ‘intermediaries’. The word ‘intermediaries’ is not defined in the Information Society Directive, and does not directly appear in the transposing English legislation. Recital (59) confirms that the provision was intended to grant relief against those who transmit infringements in a network and who are ‘best placed’ to bring infringing activities to an end. As the Court of Justice has confirmed, this does not require the intermediary to exercise actual or legal control over the use of its service, or to have a contractual relationship with the primary infringer;50 it is sufficient merely to supply the connection or some aspect of the network over which rights are infringed.51 It is therefore unlikely that the notion of carriage in a network would be understood literally so as to be confined to ISPs: understood purposively, the provision is directed at the least-cost avoider, rather than at a particular class of service provider that transmits data over their own network infrastructure.

14.56  Copyright and perfrmers’ right. Before the Court will have jurisdiction to make a blocking order under section 97A (or section 191JA), the applicant must establish the existence of at least one infringement of copyright (or a performers’ right) carried out by a third party.52

Although a single infringement is theoretically sufficient, the scale and nature of wrongdoing will of course be matters relevant to the assessment of proportionality. As the Court of Justice explained in UPC Telekabel, ‘the concept of infringement...[includes] the case of protected subject-matter placed on the internet and made available to the public without the agreement of the rightholders at issue’.53 Ultimately, whether this has occurred is a question going to primary liability—in particular, the scope of the right of communication to the public.

14.57  Database right. On a literal reading of regulation 23 of the Copyright and Rights in Databases Regulations 1997, blocking injunctions under section 97A are available for infringements of database right which are carried out using the services of a relevant service provider. Regulation 23 provides that sections 96 to 102 of the 1988 Act ‘apply in relation to database right and database in which that right subsists as they apply in relation to copyright and copyright works’. That regulation was substituted by 2006 regulations intended to transpose the provisions of the Enforcement Directive into United Kingdom law.54 It follows that section 97A was intended to apply mutatis mutandis to database right.

14.58  The concept of ‘use’. The primary infringement must be committed using the defendant’s services. This follows from the fact that the scope of article 8(3) is limited to intermediaries whose services are in fact used by a third party to infringe copyright or a related right. This causal connection supplies the basic justification for imposing a blocking obligation, since unless the defendant’s services are actually being used to carry out an infringement it would be difficult to describe them as ‘best placed’ to bring that infringement to an end. The required causal nexus has been explained in a number of different ways in the European and English authorities.

14.59  Practice and procedure. Where a blocking order is made under section 97A, the practice of the English courts is to require the inclusion of a recital to the effect that the Court is satisfied on the evidence before it that the operators or users of the relevant target websites use the services of the relevant defendant to infringe the copyrights of the claimant in the United Kingdom.55

14.60  Network-layer intermediaries. In LSG-Gesellschaft zur Wahrnehmung von Leistungsschutzrechten GmbH v Tele2 Telecommunication GmbH,56 the claimant was a collecting society that sought an order to disclose the identities of subscribers who were suspected of transmitting its members’ copyright works on peer-to-peer file-sharing networks. The national court referred a question about whether the defendant ISP was an ‘intermediary’ whose services were used within the meaning of article 8(3), given that it did not supply any services to its subscribers apart from internet access. The European Court held that it was:

Access providers which merely provide users with Internet access, without offering other services such as email, FTP or file-sharing services or exercising any control, whether de iure or de facto, over the services which users make use of, must be regarded as ‘intermediaries’ within the meaning of Article 8(3) of Directive 2001/29.57

14.61 The reason for this was that access providers such as Tele2, ‘in granting access to the Internet, make it possible for such unauthorised material to be transmitted between a subscriber to that service and a third party’.58 Accordingly, the ISP’s services were used to infringe because subscribers were transmitting infringing sound recordings across its network. The focus was on the use of the ISP’s network to transmit the infringements.

14.62  Carriage in a network. More recently, in UPC Telekabel Wien GmbH v Constantin Film Verleih GmbH, the Court of Justice held that ‘the term “intermediary” used in Article 8(3) ...covers any person who carries a third party’s infringement of a protected work or other subject-matter in a network’.59 Again, the Court emphasised the need for actual transmission (or carriage) of the claimant’s work on the intermediary’s network. What matters is that the ISP:

is an inevitable actor in any transmission of an infringement over the internet between one of its customers and a third party, since, in granting access to the network, it makes that transmission possible, [so] it must be held that an internet service provider...which allows its customers to access protected subject-matter made available to the public on the internet by a third party is an intermediary whose services are used to infringe a copyright or related right within the meaning of Article 8(3) of Directive 2001/29.60

14.63  Contract with the infringer unnecessary. The Court went on to clarify that, for services to be used, there need not be any contractual link between the ISP and the person who infringes copyright or a related right.61 It appears that this observation was directed to the situation where a person, although not personally a subscriber of an ISP, uses the computer or connection of a subscriber in order to infringe. At least one decision has construed this principle to extend more widely, so as to include third party website operators who have no connection at all to the intermediary other than that a subscriber may access the website using the services of the intermediary.62

14.64 In UPC Telekabel, the Court of Justice explained that there is nothing in article 8(3) that requires a ‘specific relationship’ between the primary infringer and the intermediary.63 It is sufficient that the defendant is an intermediary whose services are being used by the primary infringer to infringe. The Court therefore rejected any implication that a more proximate degree of involvement is required. The rationale appears to have been that to require more would reduce the level of protection given to right-holders.

14.65  Proof of actual infringement. For similar reasons, in UPC Telekabel the Court of Justice did not require that copyright owners ‘must show that some of the customers of [the ISP] actually access, on the website in issue, the protected subject-matter’.64 Thus, a claimant who seeks a blocking injunction for copyright need not adduce proof that a subscriber has in fact infringed copyright by downloading a particular copyright work using the services of the defendant. This is because, first, the act of making a work available to the public does not require evidence of actual access by end users in order for there to be a primary infringement and, second, the preventative effect intended by article 8(3) presupposes that copyright owners may obtain an injunction without having to prove that customers ‘actually access the protected subject-matter made available to the public’ using the intermediary’s services.65 Nevertheless, it is suggested that the prophylactic aim of article 8(3) will not be engaged until it at least appears prima facie likely that infringements have taken place or will take place in the future.

14.66  Marketplace intermediaries. Guidance may also be taken from L’Oréal SA v eBay International AG,66 where the Court of Justice was asked to consider whether an injunction under article 11 of the Enforcement Directive was available to prevent future trade mark infringements which took place using the services of eBay. In that case, eBay was an intermediary by virtue of hosting and transmitting listings in its online marketplace for goods which infringed the claimant’s trade marks. Its services were used because it provided the platform for uploading the infringing offers for sale, caused those offers to be made, and transmitted those offers across its network, accepted payment for them, and provided search and listing tools to sellers.

14.67  Prevention of future infringements. The Court concluded that article 11 requires member states to ensure that national courts can order ‘the operator of an online marketplace to take measures which contribute, not only to bringing to an end infringements...by users of that marketplace, but also to preventing further infringements of that kind’.67 It was implicit in this conclusion that eBay was an intermediary whose services were being used to infringe. This arose from the fact that, before the infringing offers could be made to consumers, eBay’s network needed to be used.

14.68  Required mental state. The wording of section 97A(1) is that the claimant must prove ‘actual knowledge of another person using [the defendant’s] service to infringe copyright’. While this appears to set a high threshold requiring subjective knowledge of a specific infringement by a particular person, the statutory language is broad and it has been so interpreted. Curiously, both article 8(3) of the Information Society Directive and article 11 of the Enforcement Directive appear to lack any mental element, so the knowledge requirement added to section 97A(1) must reflect an additional ‘condition’ or ‘modality’ laid down by national law pursuant to the legislative freedom conferred by recital (59) of the Information Society Directive and recital (23) of the Enforcement Directive.

14.69  Mandatory considerations. In assessing whether the required standard of knowledge is met, the Court is obliged to consider all relevant circumstances, including whether the respondent received a notice of infringement containing the relevant particulars. Section 97A(2) provides:

In determining whether a service provider has actual knowledge for the purpose of this section, a court shall take into account all matters which appear to it in the particular circumstances to be relevant and, amongst other things, shall have regard to—

(a)

whether a service provider has received a notice through a means of contact made available in accordance with regulation 6(1)(c) of the [2002 Regulations]; and

(b)

the extent to which any notice includes—

(i)

the full name and address of the sender of the notice;

(ii)

details of the infringement in question.

14.70  Failure to notify. A notice complying with section 97A(2) is not required for actual knowledge. Whether or not such a notice has been received by the service provider is merely one factor among many that will be considered by the Court in determining whether the threshold condition is satisfied.68 Other relevant factors may include whether a notice in another form has been received, whether the service provider is aware of a target website and the particular contents alleged to infringe the claimant’s rights, and whether notices of copyright infringement have been forwarded to subscribers in respect of activity occurring in relation to a target website under the Digital Economy Act 2010.

14.71  Notifications of subscribers’ infringements. More recent cases have accepted that actual knowledge can arise from notifications sent to a defendant ISP detailing instances in which infringing communications of the claimant’s works appear to be emanating from IP addresses belonging to the ISPs’ subscribers. This is so even if those transmissions are not shown to have been initiated by means of the website that is the subject of the blocking application.69

14.72  Earlier judgments. In other cases, knowledge has been inferred from the ISP’s awareness of an earlier decision in which the operators of a targeted website have been held to infringe copyright,70 or simply from the evidence served upon the ISP in support of the blocking application.71 As a result, the standard of knowledge that is required more closely resembles constructive knowledge: a respondent service provider is fixed with notice upon being sent a sufficiently detailed notice of one or more infringements which have taken place using their services, and being given a reasonable period within which to investigate the allegation.

14.73  Actual knowledge of the material facts. Although this approach may be criticised as excessively broad, it is not unprecedented: it reflects a similar standard of knowledge to that applied in cases of secondary infringement under the Copyright Act 1956 and earlier legislation. Those provisions72 are cast in terms requiring of the defendant ‘to his knowledge’ that an article would infringe ‘that’ copyright, yet they have been interpreted far more liberally to include knowledge of the material facts. Thus, it was said that it was sufficient for the claimant ‘to show that there were within the knowledge of the defendant such facts as would give rise to a reasonable suspicion that copyright existed, and was being infringed’.73

14.74  History of the knowledge requirement. The history of the requirement of actual knowledge provides some guidance as to the breadth of the knowledge requirement under section 97A. In 1880, Jessel MR wrote that:

Persons in the position of the Defendants, that is, of agents for an American publisher, must be taken to know that Americans are in the habit of printing exporting piratical works, and they must therefore know that they import books from America at the risk of their containing what is piratical, and of thus committing an unlawful act...74

In other words, at least until the Copyright Act 1911 (Imp), the risk of infringement lay with the importer, who owed a duty to make inquiries that was tantamount to indemnifying copyright owners against the risk of infringements caused by inadvertent importation. This unsympathetic attitude reached its zenith in Hanfstaengl v W H Smith & Sons, where a newsagent sold copies of a magazine in which was printed on a single page an infringement of the claimant’s copyright.75 The newsagent was liable to pay damages despite being completely unaware of the infringement and promptly withdrawing the magazine from sale upon receiving notice of the complaint. This reflects a particularly zealous conception of copyright as a tort of strict liability.

14.75  Purpose of the knowledge requirement. To guard against such absurdities, a requirement of actual knowledge was introduced for the secondary copyright infringement torts. However, the requirement of knowledge has generally been treated loosely. For example, in one case the Supreme Court of New South Wales held that ‘knowledge’ is tantamount to ‘notice of the facts such as would suggest to a reasonable man that a breach of the copyright laws was being committed’.76 If this view is correct, it suggests that the knowledge requirement in section 97A is at least as wide.

14.76  Facts. The first English decision to apply section 97A was Newzbin1,77 a case discussed in the context of copyright authorisation liability and joint tortfeasorship in chapter 6. There the defendant was a website operator who controlled the platform newzbin.com, an index of Usenet binaries which included a very high proportion of infringing material.

14.77  Scope of injunctive relief. Having found that both the operators and users of Newzbin.com infringed copyright, the key issue was what should be the scope of the injunction. The claimant copyright owners sought an injunction under section 97A in very broad terms, namely to restrain the defendant from indexing on its platform any material that infringed copyright. It was common ground that the defendant was a service provider and that its services were being used to infringe copyright. However, the defendant argued that it did not have actual knowledge of all such infringements carried out using its services, chiefly because it had never been served with a notice complying with section 97A(2). Kitchin J rejected this argument. Receipt of a notice in the prescribed form is not a precondition of actual knowledge, but merely one factor to be considered.78

14.78 However, Kitchin J confined the injunction to files which infringed the claimants’ film copyrights: to enjoin Newzbin from linking to any infringing binary content would be ‘very uncertain’ and encompass works in respect of which the claimants possessed no rights.79 The Court nevertheless took an imprecise, pragmatic approach to establishing actual knowledge; the existence of infringements ‘at large’ appears to have been enough, even without knowledge of specific acts or users. Even these limitations have been largely abandoned in later cases.

14.79  Blocking by an ISP. In Twentieth Century Fox Film Corporation v British Telecommunications plc a blocking injunction was made against the respondent ISP (‘BT’) to prevent its services from being used to infringe the applicants’ copyrights via the Newzbin website and its phoenix-like successor, Newzbin2.80 Building on the findings made by Kitchin J in Newzbin1, Arnold J held that the Court had jurisdiction to grant the order, because (1) the operators of the website (in combination with BT’s subscribers) used BT’s services to infringe the applicants’ copyrights; and (2) BT had the necessary knowledge.

14.80  Threshold of knowledge. The Court reinterpreted the requirement of actual knowledge broadly, to require only that the service provider actually knew that at least one person was using its service to infringe copyright. Unlike other forms of secondary liability, section 97A does not require knowledge of ‘a specific infringement of a specific copyright work by a specific individual’.81 This is a lower threshold than that which applies to authorisation liability and joint tortfeasorship, but reflects (1) the broader statutory language of section 97A; (2) the necessity of a broad interpretation to make the remedy achieve its stated purpose; and (3) the fact that it is concerned with injunctive relief rather than the imposition of monetary liability.

14.81 This approach is nevertheless a highly permissive interpretation. It is relatively easy for a service provider to discover (or be alerted to) a single infringement carried out using one its subscribers’ IP addresses; the test pays no regard to the volume of infringement or the proportion of non-infringing use. On the current approach, section 97A does not even require the service provider to know the identity of the person who is engaging in the infringing activity; for all the Court knows, that person could even be a licensee or representative of the applicant (though the reality is that probably he or she is not). Similarly, the applicant need not establish which copyright work is being infringed and, by extension, need not specifically prove that it has any rights to that work or that copyright even subsists in it.

14.82 In Newzbin2, Arnold J observed that actual knowledge can be conferred by notice of facts arising from the receipt of a ‘sufficiently detailed notice and a reasonable opportunity to investigate the position’.82 This appears to be so whether or not the ISP did actually investigate the position. In other words, knowledge extends to being put on notice of facts that would suggest to a reasonable person that an infringement of copyright was taking place. This appears to amount to a conclusion that the phrase ‘actual knowledge’ encompasses constructive knowledge. In some respects, Newzbin2 was a special case, since BT accepted that it had actual knowledge of all the facts as found by Kitchin J in Newzbin1, which already ruled on the liability of Newzbin and its members in considerable detail. However, subsequent cases have confirmed that the requirement of knowledge will ordinarily be satisfied by merely serving the respondent with the claim form and evidence in support of a blocking application.83

14.83  Form of order. The terms of the Newzbin2 injunction required BT to implement Cleanfeed, which was a blocking system available to BT that uses a hybrid combination of IP address re-routing and URL blocking.84 It contemplates the prevention of future infringements of the same kind—though despite the comments of the Court of Justice in L’Oréal, it is clear that this prevention extends beyond infringements committed by the same tortfeasor in respect of the same copyrights.85 BT was obliged to block access for all downstream subscribers who received Cleanfeed-enabled services86—in other words, customers could not opt out simply because they did not want filtered access to the internet.

14.84  Circumvention and extension of blocking. By the time the final injunction had been granted, Newzbin2’s operators had already released proxy circumvention software designed to allow BT’s subscribers to continue accessing Newzbin2. In an attempt to defeat the operation of this software and guard against future circumvention, the injunction provided for a mechanism by which additional IP addresses and URLs could be added to the blocking measures in correspondence. Such locations must have the ‘sole or predominant purpose’ of enabling or facilitating access to Newzbin2. The applicants bore responsibility for ensuring that the notified locations met this condition.87 Upon notification, BT would have 10 working days to implement the updated block.

14.85  Restrictions on technical suspension. As the first example of a blocking order being issued against an English ISP, it might have been expected that the terms of the order included an allowance for unforeseen technical errors introduced by blocking. However, the Court refused to allow BT unilaterally to disable Cleanfeed blocking if it reasonably considered that network stability or the preservation of other, more important IWF blocking functionality would be compromised.88 Instead, the order required BT to obtain the claimants’ (or, in urgent cases, the Court’s) consent before shutting down its blocking system. This inflexibility is open to the criticism that it compromises network security and stability, and risks limiting the ability of ISPs to make enhancements to their blocking systems and network infrastructure without engaging in a delicate process of negotiation with parties who may have limited concern for the interests of ISPs and their customers.

14.86  Extension to other ISPs. Orders in almost identical terms were made against other British ISPs, including Sky89 and TalkTalk.90 They required each ISP to ‘block or attempt to block access’ to specified Newzbin2 URLs and IP addresses, along with any others notified by the claimants which had the ‘sole or predominant purpose’ of enabling or facilitating access to Newzbin2. Both orders were highly specific as to the blocking technology to be employed, which reflected the filtering technology available to each ISP. That approach provides certainty about what the ISPs must do to comply.

14.87  Primary liability. In Dramatico the claimant record companies sought a blocking injunction against the six main English ISPs to prevent their subscribers from accessing the notorious BitTorrent tracker, The Pirate Bay. In a preliminary judgment, Arnold J held that the users and operators of that website infringe copyright, which satisfied the first element of section 97A.91 The website operators infringed because they authorised, procured, and acted pursuant to a common design with users of The Pirate Bay, to infringe copyright. Users infringed by communicating the claimants’ works to the public by electronic transmission within the meaning of section 20(2)(b) of the 1988 Act.92

14.88  Knowledge and proportionality. In a subsequent decision, Arnold J held that the ISPs had acquired actual knowledge from previous notifications sent to them, as well as the claimants’ evidence and the first judgment.93 Although the Court could not simply ‘rubber stamp’ the order agreed between the parties, blocking was said to be proportionate for similar reasons to Newzbin2. If anything, Arnold J concluded, the scale and proportion of infringing activity made this an even stronger case for blocking than Newzbin2. Because The Pirate Bay was hosted on a dedicated server that did not share an IP address with other websites, it was appropriate to order the ISPs to adopt IP address blocking.94 A similar update mechanism was specified to accommodate changes to the website’s URL and IP address.

14.89  Facts. In EMI Records Ltd v British Sky Broadcasting Ltd the Court issued injunctions under section 97A against the same six ISPs to block access to three BitTorrent trackers.95 The trackers functioned similarly to The Pirate Bay and were plainly liable as primary and secondary infringers. By all accounts, they were very popular websites which indexed a ‘vast’ selection of infringing content: KAT was one of the top 135 most visited websites in the world; H33T and Fenopy had several hundred thousand British visitors each month.

14.90  Primary liability of users. Arnold J concluded that users of the websites infringed copyright by copying (downloading) works without the licence of the copyright owner, and in many instances by communicating the works to the public in the United Kingdom by uploading those works, which thereby made those works available by electronic transmission. As the judge explained:

the answer to the question...whether the act of communication to the public for the purposes of Article 3(1) of the Information Society Directive occurs where the communication originates or where it is received or both, is that the act may occur both where the communication originates and where it is received. Where a communication to the public which originates outside the UK is received inside the UK, the act will be treated as occurring within the UK if the communication is targeted at the public in the UK.96

14.91  Targeting. In relation to the target websites, communications to the public would take place in the United Kingdom at least where the uploader is located in the United Kingdom. The judge regarded it as ‘rather more doubtful’ whether a communication to the public would occur in the United Kingdom where only the downloader was located in the United Kingdom, since it was not clear on the evidence that the three websites were targeted at the United Kingdom. However, it was unnecessary for the judge to decide the point.97 Had it been necessary to do so, it is suggested that some communications (eg those relating to English artists or television programmes) may well have been targeted to users including those in the United Kingdom, which would be sufficient to confer jurisdiction.

14.92  Infringements by website operators. For similar reasons to Newzbin2 and Dramatico, the operators of the target websites infringed copyright by communicating the relevant works to the public, authorising users’ infringements, and as joint tortfeasors.98 The same authorisation factors were relevant: each of the target websites enabled users to search for ‘torrent’ files using a variety of tools, took steps to index and arrange the means of infringement (the BitTorrent files and magnet links), had the objective and intention of causing infringement, which was the inevitable consequence of providing the torrent files, and made very substantial revenues from doing so. There was also evidence that the website operators had moderated indexed files to remove certain illegal and objectionable content (such as child abuse material), which demonstrated that they had control over the content and the ability to remove material when notified.

14.93  Earlier enforcement attempts. Importantly, the claimants had attempted to follow the website operators’ notice-and-takedown policies. The evidence was that, six months after sending notifications to the websites, 76 per cent of notified material remained accessible via KAT, while 83 per cent remained accessible on H33T (which also charged $50 per URL notified) and 25 per cent remained accessible on Fenopy.99 As to Fenopy, this suggests that even compliance with a majority of takedown requests will not be sufficient to avoid a finding of infringement liability. It also provides a powerful reason in support of blocking where other, more direct means of enforcement have clearly failed.

14.94  Use of the ISPs’ services. The Court held that both the users and operators of the target websites used the ISPs’ services to infringe the claimants’ copyrights. The claimants had adduced evidence of the IP addresses that had made available sound recordings belonging to the claimants on BitTorrent, with some 4.3m instances of infringement attributable to the ISPs’ subscribers. Although not all of these instances would have related to torrent files obtained specifically through the target websites, the Court was satisfied that ‘they confirm substantial use by users of the [ISPs’] services for these kinds of activities’.100

14.95  Knowledge. Actual knowledge was satisfied by means of weekly notifications sent by the claimants’ monitoring agents to the ISPs, which set out details of the numbers of subscribers using their services to infringe the claimants’ members’ copyrights. Coupled with the evidence in support of the application, Arnold J concluded that the ISPs had actual knowledge that the users and operators of the target websites used their services to infringe.101

14.96  Facts. In Football Association Premier League Ltd v British Sky Broadcasting Ltd,102 Arnold J made blocking orders against the six main British ISPs requiring them to block access to a website known as FirstRow, which was a portal providing access to streamed broadcasts of sporting events. By visiting the website, users were able to access live coverage of sporting events, including Premier League matches to which the claimant possessed the exclusive right to licence broadcasting and in which it owned the copyright. At the time of the application, FirstRow was among the top 300 websites in the United Kingdom.

14.97 FirstRow did not itself transmit any of the live streams, but instead relied upon third party user-generated content services to do so. Trusted users provided unauthorised streams to the third party websites, which were then indexed and aggregated by FirstRow and presented to end users. The evidence was that the claimant had been unable to identify the operators of FirstRow and that the website was hosted by a Swedish hosting company that was ‘a well-known haven for pirate sites’.103 The claimant’s agents had also sent ‘a number of’ notices to FirstRow but these had not been responded to.

14.98  Jurisdictional requirements. The Court held that the threshold requirements for jurisdiction were clearly satisfied. The defendant ISPs were service providers. The operators of FirstRow infringed copyright by communicating the claimant’s broadcasts via the internet. This was despite the actual transmissions emanating from the third party user-generated content services, because:

FirstRow aggregates together a large number of streams from a variety of streamers, indexes them for the convenience of the user and provides a simple link for the user to click on in order to access a specific stream. It is true that the technical effect of clicking on the link is to direct the stream from the [user-generated content] site to the user’s computer, but even so the stream is presented in a frame provided by FirstRow. In all the circumstances, I consider that FirstRow is responsible for the communication.104

14.99 In any case, FirstRow was liable as a joint tortfeasor with the third parties who uploaded the live streams to the user-generated content services. The communications to the public took place in the United Kingdom because the FirstRow website was targeted at the public there. Relevant factors were the language of the website (English), the inclusion of advertising for English companies and products, the inclusion of content popular with English audiences such as the Premier League, the popularity of the website in the United Kingdom, and the relatively high proportion of traffic that originated from the United Kingdom.105

14.100  Use of the ISPs’ services. For the same reasons as in Newzbin2, Dramatico, and EMI, the users and operators of FirstRow used the ISPs’ services to infringe the claimant’s copyrights. Actual knowledge was conferred by reason of detailed letters before action sent to the ISPs, which enclosed the evidence in support of the application.106 Accordingly, the Court had jurisdiction to make the orders sought, and—upon finding them to be proportionate—did so in similar terms to Dramatico [No 2] and Newzbin2 [No 2].

14.101  Facts. In Paramount Home Entertainment International Ltd v British Sky Broadcasting Ltd, Arnold J ordered the six main English ISPs to block access to two websites which provided access to streaming versions of cinematograph films and television programmes.107 Like FirstRow, the website operators provided a ‘comprehensively categorised, referenced, moderated and searchable’ index of material which made it easy for users to access high quality versions their chosen copyright works, which were hosted by third parties.108

14.102  Jurisdictional requirements. For similar reasons to Dramatico [No 2] and FirstRow, the Court held that the jurisdictional requirements under section 97A were satisfied. Like FirstRow, the website operators at issue ‘make it much easier for members of the public to find what they want’, and therefore make the content available to the public even though it is ultimately hosted by third parties.109 In any case, the website operators were jointly liable with their users, who themselves communicated the works to the public by uploading material to a host site. The website operators also authorised those infringements.110 Those infringements were carried out using the ISPs’ services for the same reasons as were given in Newzbin2, Dramatico [No 2], and FirstRow. The ISPs had actual knowledge by virtue of emails sent to the ISPs by the claimants attaching evidence of infringement, and by service of the application.

14.103  Facts and policy. In a second blocking application made by Paramount, Paramount Home Entertainment International Ltd v British Sky Broadcasting Ltd [No 2], Henderson J made blocking orders in relation to four target websites which communicated the claimants’ films and television programmes to the public.111 Henderson J explained the rationale of section 97A in the following terms:

A particular reason why section 97A needed to be enacted in order to give effect to Article 8(3) is that under English law an injunction cannot normally be granted against a person against whom no substantive cause of action exists. Authority to do so, therefore, was specifically provided in section 97A.112

14.104  Primary liability and targeting. Like previous applications, the website operators in question were ‘intervening in a highly material way to make the copyright works available to a new audience’.113 Such communications were targeted at users in the United Kingdom, having regard to the popularity of the websites there, the use of local advertisements, and the use of English as the default language. Although it may be doubted whether this will normally be sufficient evidence of targeting, it is clear that the target websites in question directed a large part of their focus and content to a United Kingdom audience. As before, actual knowledge was satisfied by means of emails sent to the ISPs giving notice of the application, and service of the application itself.114

14.105  Proportionality. For much the same reasons as Arnold J, Henderson J concluded that the orders were proportionate. Both the website operators and their users must know that their activities are unlawful and illegitimate. Users would not be deprived of access to the material through lawful channels. The ISPs were capable of giving effect to the orders; the cost of compliance was ‘relatively modest’ and ‘will not cause them any particular difficulty’. Accordingly, the ‘plain’ result of the balancing exercise was that the claimants’ rights to protection of their copyright should prevail over the rights of the website operators, users, and ISPs.115

14.106  Facts and decision. In 1967 Ltd v British Sky Broadcasting Ltd, Arnold J made blocking orders against the (by this time) five main English ISPs requiring them to block access to 21 target websites which were all BitTorrent trackers similar to The Pirate Bay, KAT, H33T, and Fenopy.116 For similar reasons, all the websites used the services of the defendant ISPs to infringe copyright. Of importance was that the target websites ‘all have a large number of UK users’ and contain musical albums that were popular in the United Kingdom; their acts of communication to the public were therefore targeted at the public in the United Kingdom.117 Moreover, they were liable for authorising users’ infringements and as joint tortfeasors. Insofar as United Kingdom users accessed and downloaded content, thereby copying it, or uploaded content into the BitTorrent ‘swarm’, thereby communicating it to the public, they were also using the ISPs’ services to infringe.118 Accordingly, the Court had jurisdiction to make the orders and it was proportionate to do so.

14.107  Facts. In Twentieth Century Fox Film Corp v Sky UK Ltd, Birss J made blocking orders in respect of 9 websites which provided access to the claimants’ films and television programmes.119 The target websites fell into three sub-categories: streaming sites (analogous to those previously considered in Paramount [No 1]); BitTorrent indices (analogous to those considered in Dramatico); and a new category: ‘Popcorn Time type sites’. In broad terms, these sites involved a downloadable application which enabled access to copyright-protected content which could be streamed or downloaded using BitTorrent. The application was obtained from a website which did not itself make available any content. The application worked by aggregating torrent files indexed by third party hosts, where the list of aggregated sources was obtained from a further ‘source’ website. Once content was selected by the user to be played or downloaded, it was transmitted using the BitTorrent protocol by the usual process of assembling its constituent parts from third parties in the ‘swarm’. It was clear that Popcorn Time was not used to watch lawful material, but was used almost entirely to watch copyright-infringing content.

14.108  Primary liability. The analysis of primary infringement for the Popcorn Time-type sites was complicated by the fact that the website providing the downloadable application did not itself commit any act of communication to the public (the Popcorn Time application was not an infringing work). Similarly, the operators of ‘source’ websites could not be said to communicate works to the public; all they provided was a data file with a list of feeds. From the perspective of a user of the Popcorn Time application, the ‘source’ websites were invisible. Although it might have been argued that the application suppliers had authorised infringements by users, the claimants did not make such an argument and instead relied upon authorisation by the hosts of the websites. The difficulty with that submission is that there was no evidence about their relationship with the application suppliers, so the allegation failed.120 However, a similar allegation based on joint tortfeasance succeeded: the website operators intentionally procured and induced users to access the hosted source and torrent material and thereby caused the infringements. They also engaged in a common design with the operators of the host websites to communicate works to the public.121

14.109  Use of the ISPs’ services. In these circumstances, it was clear that the respondent ISPs ‘had an essential role in the infringements’ because the ISPs’ networks were used to transmit the Popcorn Time application, the source lists, and the transmitted copyright works.122 As such, the jurisdictional preconditions for a blocking order were satisfied. On proportionality, this was a ‘clear case’: the claimants were seeking both to prevent access to the Popcorn Time application and to inhibit the usage of copies that had already been downloaded. Blocking orders directed at both the application suppliers and the source websites were likely to achieve those objectives and it was proportionate to make the order.123

14.110  Practice and procedure. The court also commented on the appropriateness of dealing with an application for a blocking injunction on paper. The starting point is that it is a matter for the court whether or not to accede to such a request. Birss J explained that, in a blocking claim, such an application would effectively dispose of the action, yet would not fall within any of the ‘limited exceptions’ to the general rule that trials are to be held in public.124 Three factors will often militate in favour of paper disposal in these cases: first, where the factual circumstances are the same as similar applications which have been considered in reasoned public judgments; second, where the order is unopposed by the respondents; and third, where no third party opposes the order either.125 However, in light of the complexities raised by the Popcorn Time-type sites, and the fact that those issues had not previously been decided, the court directed that the application proceed to a hearing.126

14.111  Facts and decision. In this case, Mann J ordered the five largest British ISPs to block or attempt to block access to eight websites through which infringing copies of books and other literary works were available.127 The application was brought by the Publishers Association, which claims that the target websites collectively purported to host ten million book titles,128 80 to 90 per cent of which were infringing. Copyright owners had previously requested the de-indexing of over 1.75 million URLs from Google search results and issued almost one million takedown requests.

14.112  Relevant dimensions. Once the four threshold conditions are satisfied, the Court has jurisdiction to make a blocking order under section 97A. Before it will do so, however, it must be satisfied that the order is proportionate, both as between the claimants and defendants, and as between the claimants and third parties such as affected website operators and subscribers.

14.113  The overall inquiry. The general approach to proportionality is considered in detail in chapter 13. The same broad inquiry is followed in website blocking claims. However, as Arnold J noted in EMI, proportionality is inherently a fact-specific issue:

the proportionality of a blocking order is bound to be a context-sensitive question. As I hope my judgments in 20C Fox v BT, Dramatico v Sky (No 2) and the present case demonstrate, this Court does not make such orders without thorough consideration of whether it is appropriate to make an order in the light of the specific facts of each case.129

14.114  Relevant factors. A number of factors will commonly be relevant to the proportionality analysis in blocking cases. As between claimants and defendants, the cost and availability of technical means to comply with an order will be of primary importance, as will the magnitude of the corresponding benefits to the claimants. As between claimants and third parties, the impact on lawful content and legitimate trade, the justification for interfering in third parties’ freedom of expression and privacy, and the effectiveness of the order at preventing future infringements of the same kind, will ordinarily be relevant.

14.115 In all of the cases considered in section 3.6, the Court had no hesitation in making the orders sought.130 Proportionality was usually dealt with in a more or less cursory fashion, which reflected the fact that the target websites in those cases were notorious havens for infringement and lacked substantial non-infringing uses. Based on those cases, following factors may be relevant to the assessment of proportionality.

14.116  Assent to the terms of the order. It has been considered relevant that, as between a claimant who seeks a blocking order and the ISPs against whom it is sought, the ISPs do not oppose the making of the order and have agreed the terms of the draft order.131 This is obviously a relevant factor to be taken into account, since an order whose terms are hotly disputed is less likely to be proportionate than one which stands agreed as between the parties. However, it should not be forgotten that ISPs often do not oppose orders because they fear a risk of adverse liability for costs should they appear to make submissions. Further, this does not necessarily mean that an order is proportionate as regards the rights of third parties.

14.117  Where costs are modest. The costs of complying with a blocking order are a relevant factor in considering the proportionality of the order as between claimant and defendants.132 In all blocking cases to date, the Court has described the costs to ISPs of implementing blocking orders as ‘modest and proportionate’.133 This appears to be based on an estimate given by BT in Newzbin2 that blocking would cost approximately £5,000 to implement and £100 per subsequent notification of a new URL or IP address to maintain.

14.118  Aggregate costs. In Cartier, Arnold J explained that the costs associated with blocking measures, and in particular the costs of implementing them, will be a particularly important consideration when evaluating the proportionality of relief. In that case, the marginal operating cost of implementing a single blocking order was ‘modest’, largely because the ISPs ‘already have the requisite technology at their disposal’.134 However, as Arnold J explained:

that is not the whole story. From the ISPs’ perspective, what really matters is not the cost of implementing a single order, but the cumulative cost of implementing all website blocking orders. Counsel for [the claimant] submitted that [the] application should be judged on its own merits, and hence solely on the basis of the costs consequences to the ISPs of the orders sought by this application. I do not accept that submission, which in my view does not reflect the reality of the situation. The impact of one order cannot be judged in isolation from the impact of other orders that have been or are likely to be granted.135

It is therefore signficant that costs cannot be judged in isolation but must be assessed in the wider context of past and future blocking orders. In this regard, although it is difficult to predict the future, the courts generally proceed on the footing that claimants will ‘concentrate their resources on seeking relief in respect of the more egregious infringers’ so that there will not be a ‘flood of applications’.136 If correct, this approach will need to be kept under review.

14.119  Overall costs of doing business. The approach ultimately accepted by the Court in Cartier suggests that the proportionality of implementation costs imposed on ISPs by blocking orders is to be assessed relative to the overall operating costs of the intermediary. If this approach is upheld on appeal, it will mean that even relatively large cumulative costs are unlikely to be disproportionate in the context of a multi-billion-pound enterprise.

14.120  Applications involving multiple websites. In 1967, Arnold J clarified that the number of websites sought to be targeted by a blocking order (in that case 21), or the cumulative number of blocked websites, was not of itself disproportionate having regard to the impact on ISPs. As the judge explained:

I am not persuaded that the number of websites which have been ordered to be blocked by virtue of section 97A, or are likely to be ordered to be blocked in the future, imposes disproportionate implementation costs on the Defendants.137

In light of this, there appears to be no reason in principle why claimants cannot seek to target a large number of infringing websites in a single application. They will presumably wish to do so in order to take advantage of economies of scale when making applications, with the objective of minimising their enforcement costs.

14.121  Absence of other effective remedy. In FirstRow and EMI, the Court was clearly sympathetic to the difficulty faced by the claimants in identifying and pursuing the operators of infringing websites. The evidence in those cases was that a variety of other enforcement methods had been pursued—including notice-and-takedown in respect of specific content, letters to the website operators, and letters to the websites’ hosting companies—but that these had all failed to produce results. It was therefore said that blocking orders were necessary to protect copyrights of the claimants from large-scale infringement. The absence of any other effective remedy supported the proportionality of the orders in those cases.138

14.122  Secondary benefits of blocking. In FirstRow, the Court considered that blocking orders were desirable to protect the ‘sporting objectives’ of the claimants’ broadcasting restrictions, which were designed to encourage attendance at live football matches and support the sport of football, by prohibiting live broadcasts of such matches on Saturday afternoons. The Court relied upon Viagogo to reason that it was legitimate to take this factor into account, since it was in the public interest to uphold those restrictions.139

14.123  Relevant considerations. As between claimants and third parties, it is of critical importance to determine the extent, if any, to which blocking orders are likely to interfere in the transmission of lawful material. This is usually affected by two considerations: first, the type of blocking being sought (whether IP address, URL, DNS, or DPI blocking); and second, the safeguards contained in the blocking order.

14.124  Type of blocking. In Newzbin2, Dramatico, and EMI, no difficulty arose because the target websites in those cases were all hosted on dedicated servers and on their own domain names. This meant that the risk of overblocking was relatively limited (subject to malicious DNS redirection by a website operator). In FirstRow, the claimant’s evidence was that the target website was hosted on a dedicated IP address; IP blocking was therefore ordered in respect of the primary domain name, and two-stage URL filtering for any mirror that was hosted on a shared IP address. However, after the order was made, the website moved to a different IP address, which was notified to the ISPs for IP blocking. It later eventuated that the new IP address was shared with a number of other lawful websites, including the Radio Times. Similarly, when implementing the block of the YIFY-Torrents website, access was inadvertently blocked to the popular image-sharing platform Imgur.140

14.125  Difficulty of predicting errors. These incidents highlight one of the inherent difficulties associated with determining the effects of blocking orders: it is notoriously hard to predict how blocking will affect third party websites, especially in circumstances where: (1) the evidence available at a hearing is often imperfect; (2) the respondents to a blocking application are frequently unrepresented and neither oppose the application nor appear; (3) the actual effects of blocking may be unknown until after an order has been implemented; and (4) actions taken by website operators or third parties may affect lawful material long after an order has been made. For example, a lawful site may begin sharing an IP address with a blocked site, a host may consolidate multiple servers together, or a website operator may change hosts or use a shared server.

14.126  Failure to pursue alternative measures. Also relevant to the issue of proportionality is the availability of alternative measures which are less onerous than website blocking. In most cases, there will be a variety of enforcement measures which can be deployed in aid of the claimant’s rights. However, the existing practice is to require the defendant to establish that the comparative efficacy and cost of those measures is such that blocking would be disproportionate. Cartier suggests that this is a very high burden: even if the claimant has not attempted alternative measures which are straightforward, and has stated that it has no intention to do so, blocking is not necessarily disproportionate if the alternatives are not ‘realistic’ or unlikely to achieve more than short-term disruption to the infringing activity.141 In short, what must be established is that there is an equally effective but materially less burdensome measure than blocking.

14.127  Conventional practice. The practice of the English courts is now to insist upon a number of safeguards being included in a blocking order. These may be summarised as follows:

(a)

Publication of notice to affected users. Orders now require ISPs against whom a blocking order has been made to take reasonable steps to notify their subscribers of the reason why access to an internet location has been denied. This important provision aims to improve the transparency of blocking by notifying members of the public of the order that has been made, the identity of the parties who obtained the order, and the possibility of applying to the Court to discharge or vary it. It is also said that this may improve the deterrent value of blocking.

(b)

Liberty to apply. The defendants, their affected subscribers, and affected third parties (such as the operators of target websites or other lawful websites) have a general permission to make submissions to the Court in the event that there is any material change in circumstances, or if it appears that their fundamental rights and freedoms are being unduly impacted.

(c)

Sunset clause. Blocking will automatically cease a fixed period (eg 2 years) after the date of the order, subject to any further order of the Court. This sensible limitation reflects the relatively short half-life of many websites associated with infringing activity, and the desirability that blocking ‘not endure longer than necessary’.142

14.128 These safeguards have been progressively introduced and, at least since Cartier (a case discussed in section 4.1), seem likely to be followed in future blocking applications. Although Cartier was a case concerning a blocking order for trade mark infringement, Arnold J has clarified in 1967 that the same safeguards should be incorporated into future copyright blocking injunctions.143

14.129  Technical and commercial effectiveness. In determining whether blocking injunctions are effective to prevent copyright infringement, two considerations are relevant. First, as Consumer Focus has argued,144 there is the technical question of how far access to the targeted resource can be interdicted. Second, wider effects on the growth of legal markets must be considered.

14.130  Effect on traffic volumes. As to technical effectiveness, the record is mixed in copyright cases, and it is difficult to isolate the effect of one national blocking order on overall traffic levels. It is generally acknowledged that blocking is imperfect but likely to have some material effect upon access to the websites targeted by an order. The Court explained of blocking orders in FirstRow that:

While they are unlikely to be completely efficacious, since some users will be able to circumvent the technical measures which the orders require the [ISPs] to adopt, it is likely that they will be reasonably effective.145

14.131 The technical effect on access can be appreciable. According to Alexa traffic statistics, overall access to Newzbin and Newzbin2 has fallen by 49 per cent since BT implemented blocking, and the website’s global traffic rank has fallen by 900 per cent.146 The local effect is likely to be even more pronounced, as these are global, rather than UK-specific statistics. Where Newzbin was once a predominantly Anglophone service, by 2012 UK visitors had fallen to just 8.5 per cent of the website’s traffic. Importantly, traffic to other infringement portals does not appear to have materially increased over the same period.147

14.132 In 1967, Arnold J noted the claimants’ evidence provided further support to the effectiveness of blocking orders in copyright cases:

using comScore data, the Claimants calculate that, on average, the number of UK visitors to Bittorrent websites which have been the subject of blocking orders has declined by 87%. No doubt some of these users are using circumvention measures which are not reflected in the comScore data, but for the reasons given elsewhere it seems clear that not all users do this.148

14.133  Trends in other member states. A similar trend can be observed in Denmark, where unique visitors to TPB reportedly declined from 140,000 to under 10,000 per day following implementation of domestic blocking orders.149 In Finland, thepiratebay.org was the twenty-seventh most popular website before a blocking order, and now does not appear at all on the list of the top 10,000 websites. However, thepiratebay.se, an exact mirror, soon rose to fifty-first position, which suggests that many users located the mirror and continued to use it. In the Netherlands, thepiratebay.se remained the thirty-ninth most popular website despite partial blocking, while in Sweden (where blocking is very limited) that website ranked seventeenth.150 Conversely, the Austrian traffic rank of kino.to slid from around 800 to under 100,000 following its blocking and seizure in Austria.151

14.134  Empirical research. One Dutch academic study concluded that blocking access to The Pirate Bay in the Netherlands had no lasting net impact on the overall number of downloaders from unlicensed sources, either because new users started downloading or because existing blocked users exhibited a relapse effect, whereby they circumvented blocking or found alternative sources for blocked material.152 However, empirical data relied upon by the claimants in the Cartier case produced a ‘striking contrast’ between United Kingdom traffic and global traffic to websites blocked under section 97A orders.153

14.135  Emerging themes. It is difficult to draw any consistent conclusions from these sources of data: collection methods vary widely, do not always track all URL or domain name changes, and do not always take account of circumvention using mirror sites and VPNs. These trends may be partly explicable by the fact that where only some ISPs block a popular file-sharing website, only a certain percentage of consumers are motivated to switch ISPs or circumvent the block, whereas the majority will simply switch to an alternative source. However, it appears that blocking can sometimes be associated with substantial reductions in traffic to websites targeted by blocking orders.

14.136  Commercial effectiveness. As to the second consideration, the positive effects of blocking and de-indexing in catalysing the adoption of legal services should not be underestimated. Blocking measures increase the cost of infringement by adding to search costs (finding a mirror or alternative website), configuration costs (eg having to download and install proxy or anonymisation software), and in some cases service costs (eg paying for a VPN provider or new ISP). While consumers will evaluate these added costs with different price sensitivities, it can be expected that overall levels of infringement will fall. Eventually, the cost of infringement will rise to a point where—when the probabilities of enforcement leading to personal liability and future inconvenience are factored in—it becomes more efficient for a consumer to acquire content from a lawful source. Thus, as Arnold J reasoned in Newzbin2:

the cost differential between using Newzbin2 and using a lawful service...will narrow still further. This is particularly true for less active users. The smaller the cost differential, the more likely it is that at least some users will be prepared to pay a little extra to obtain material from a legitimate service.154

14.137 Moreover, even if most users are willing to invest some additional time and effort to evade a website blocking injunction, many will simply not possess the technical skill required to do so on an ongoing basis. Even the simplest circumvention techniques (such as Tor or a web-based proxy) still require a degree of knowledge that many casual infringers lack. In any case, it is difficult to predict how many users would actually take steps to circumvent a block, as Parker J noted in BT:

In theory, some may cease or substantially curtail their unlawful activities...; others may simply seek other means to continue their unlawful activities, using whatever technical means are open. The final outcome is uncertain because it is notoriously difficult accurately to predict human behaviour...155

14.138  Circumvention by website operators. One point that these arguments fail to consider is the degree to which website operators could themselves render blocking injunctions ineffective by (1) altering their URL; (2) arming their users with circumvention tools like The Pirate Bay did; or (3) shifting to new protocols or adopting new technologies (such as HTTPS encryption) which defeat conventional blocking systems or make blocking more difficult. Operator circumvention is a potentially more serious concern than user circumvention, since it will often result in access being restored to infringing material without end users having to do anything at all.

14.139 Notification mechanisms of the kind used in Newzbin2, Dramatico, and EMI can ward off the first kind of operator circumvention, since blocking orders can be updated to accommodate new URLs, IP addresses, and domain names. The second method raises more significant practical difficulties, particularly when the circumvention tools or instructions are distributed by an independent third party who is not otherwise a tortfeasor. While there may be an argument that such a party is itself an intermediary amendable to a blocking injunction, or joins with users and operators of the service in a common design to infringe copyright, this is far from certain. Ultimately, methods based on user circumvention involve the same assumptions about human behaviour which Parker J declined to validate in BT. While the potential exists for infringement services to migrate to more sophisticated protocols, this does not make remedies against existing services any less effective.156

14.140  Use of warning messages. Also relevant to proportionality is the extent to which blocking orders dissuade users and website operators from infringing copyright. This is especially so where a warning page is displayed in place of the blocked site, to explain the reason why blocking has occurred.157 To maximise the effectiveness of blocking remedies, provision should be made for displaying hyperlinks to legitimate sources of material in place of the infringing substitute a user was attempting to access. However, courts should exercise some supervisory role to ensure that content on redirection pages is framed in appropriate terms.

14.141  The need for regular updates. Blocking remedies should not be treated with a ‘set and forget’ mentality. The methods used to implement an order must be re-evaluated periodically as circumvention rises and accuracy falls. Even if they were once effective, they will only remain so if viable countermeasures are eventually applied. As Clayton points out:

the effectiveness of any blocking system, and the true cost of ensuring it continues to provide accurate results, cannot be properly assessed until it comes under serious assault.158

Blocking thus represents an ongoing process of maintenance, circumvention detection, countermeasures, and monitoring whose effectiveness will change over time. A failure to notify updates will normally result in the effectiveness of the order reducing.

14.142  Interim blocking injunctions. An effective blocking remedy may sometimes require relief to be available rapidly and on an interim basis. The most obvious example where effective interim relief is necessary is to prevent the unauthorised streaming of sporting events and other live content such as performances, or access to pre-release films, software, and games. If an unauthorised stream is operative, much of the economic damage may be done to the copyright owner within a matter of hours. Given that it can be impossible to remove such materials at their source within this timeframe, blocking may have a role to play in reducing instances of streaming infringement. Existing statutory remedies involve a prolonged two-part trial and require proof of actual infringement, which makes them impractical as preventative measures. Substantial improvements to blocking and de-indexing technology are also required before interim orders will become effective; it can currently take hours or even days to be implemented. However, these remedies may be the only effective recourse against wrongdoing of this kind.

14.143  Relationship to other forms of relief. Finally, it is important to remember that blocking is just one remedy in an armoury of many. While none will be completely effective at occluding access to internet material, each can form a useful and effective component of a wider suite of remedies designed to reduce access to infringing materials. When deployed together, their efficacy may be far greater than the sum of their individual effects. Complementary remedies are vital for effective relief.159

14.144 In the short term, blocking is likely to cause infringers to shift to unblocked platforms, ISPs, or circumvention tools. As it becomes more difficult to access unblocked platforms and blocking technologies increase in accuracy and power, we can expect that the cost and difficulty of engaging in infringing activity will increase. Overall, common sense dictates that casual infringers will eventually shift to cheaper alternatives, such as reasonably priced legitimate sources. The remainder—so-called ‘hard core infringers’—are likely to concentrate in small communities using darknets, VPNs, and anonymising tools. While it is probably impossible to prevent this kind of activity, it is conceivable that the scale of tortious activity will be substantially curtailed.

14.145  Statutory basis. Although now repealed, section 17 of the Digital Economy Act 2010 included a procedure—reserved until declared by the Secretary of State160—by which a party may apply for an injunction against any service provider to prevent its service from being used to gain access to an internet location which:

(i)

hosts or is likely to make available a substantial quantity of infringing material; or

(ii)

is being used to facilitate access to such a location.161

14.146  Definition of ‘service provider’. ‘Service provider’ was given the same definition as under section 97A,162 which would include ISPs. ‘Internet location’ was left undefined, and potentially included anything that is accessible using an internet network, such as a website, URL, TLD, IP address, subnet range, or individual file. Precisely what constitutes a ‘substantial quantity’ remains unclear, but would have been the subject of future regulation. Like sections 97A and 191JA, section 17 was limited to the blocking of internet locations which facilitate copyright infringement.

14.147  Comparison to section 97A. Section 17 went well beyond the remedies required by EU law. It was a ‘purely domestic provision’ with a broader purpose and effect.163 It was broader than section 97A in various respects, including the omission of a knowledge requirement (other than notice of the application),164 and its applicability to websites whose only connection to tortious activity was that they were ‘likely to be used to facilitate access’ to another location at which infringing material may be found. Further although section 17 did not oblige ISPs to take any specific preventative steps, its practical effect might have been to encourage voluntary blocking of material notified by copyright owners.165

14.148  Future of section 17. On the basis of Ofcom’s report,166 the government indicated that it would not take further steps to implement the provisions of section 17.167 Subsequently, section 56 of the Deregulation Act 2015 repealed section 17 and 18 of the Digital Economy Act.

14.149  Overview. It is notable that website blocking remedies have been sought almost exclusively in the context of copyright infringement, at the expense of other forms of intellectual property infringement and other kinds of wrongdoing. This section discusses the wider availability of blocking measures for non-copyright wrongs.

14.150  The Cartier decision. In Cartier, the High Court held that it had jurisdiction to grant a website blocking injunction against an ISP in a trade mark case, and that there was a principled basis upon which the Court may exercise that jurisdiction. That conclusion stemmed from the Court’s approach to section 37(1) of the Senior Courts Act 1981, which it held reflected an unlimited jurisdiction (in the sense of a power) to grant injunctions, and from a Marleasing interpretation of that provision to reflect the scope of article 11 of the Enforcement Directive. It is notable that in that case the claimants did not have, and did not assert, any cause of action against the respondent ISPs. Instead, the Court relied upon an analogy with the equitable protective duty which provides the basis for Norwich Pharmacal disclosure:

If a man has in his possession or control goods the dissemination of which, whether in the way of trade or, possibly, merely by way of gifts will infringe another’s patent or trade mark, he becomes, as soon as he is aware of this fact, subject to a duty, an equitable duty, not to allow those goods to pass out of his possession or control at any rate in circumstances in which the proprietor of the patent or mark might be injured by infringement ensuing. The man having the goods in his possession or control must not aid the infringement by letting the goods get into the hands of those who may use them or deal with them in a way which will invade the proprietor’s rights. Even though by doing so he might not himself infringe the patent or trade mark, he would be in dereliction of his duty to the proprietor. This duty is one which will, if necessary, be enforced in equity by way of injunction.168

14.151  Doctrinal principle. Arnold J took from Upmann v Elkan the proposition that an ISP owes a duty not to aid infringements of an intellectual property right by letting internet material get into the hands of their subscribers. They discharge this duty by taking ‘proportionate measures to prevent or reduce such infringements’, though they are not themselves liable for them.169 This was a sufficient basis in domestic law for the court to grant non-statutory blocking remedies for infringements of trade mark rights, though there is no reason in principle why the same reasoning could not also justify blocking remedies in respect of other classes of wrongdoing.

14.152  Basis under EU law. Although unnecessary in light of the judge’s primary conclusions in Cartier, a further basis for trade mark blocking orders would have been article 11 of the Enforcement Directive, which requires member states to ensure that national courts can order an intermediary to adopt measures which contribute to bringing to an end current and future infringements of intellectual property rights. Those injunctions must be effective, proportionate, dissuasive, and must not create barriers to legitimate trade.170 In light of that requirement, and the Marleasing principle, Arnold J held that the English courts are required to construe section 37(1) of the 1981 Act to empower the Court to grant an injunction against an intermediary in those circumstances.171 The ISPs have appealed against this conclusion.

14.153  Elements of relief. If the approach taken in Cartier is upheld on appeal, the threshold conditions for a trade mark blocking injunction will be equivalent to those enacted under section 97A:

(a)

First, the respondent must be an ‘intermediary’ within the meaning of article 11 of the Enforcement Directive. For most purposes, this will lead to the same result as the phrase ‘service provider’ used in section 97A.

(b)

Second, the users or the operators (or both) of the target website in question must be infringing the applicant’s trade mark.

(c)

Third, such infringement must be carried out using the services of the respondent.

(d)

Fourth, the respondent must have actual knowledge of that fact. This follows from the need to avoid imposing a general monitoring duty. In practice, this requirement is interpreted in the same manner as section 97A.172

14.154  Application to counterfeit websites. As a result of the approach taken in Cartier at first instance, it now appears that blocking injunctions will in principle be available to trade mark proprietors wherever a website selling counterfeit goods, or otherwise infringing a trade mark, is accessible from within the United Kingdom and targeted at consumers there. However, relief would still need to be proportionate in the circumstances, which involves consideration of the factors identified in section 3.7.

14.155  Application to unlicensed medicines. Another possible application of trade mark blocking injunctions may be to prevent the sale of counterfeit medicines over the internet. This phenomenon poses considerable challenges for trade mark owners, consumers, and national enforcement authorities. Unlicensed internet pharmacies are now widespread, many of which sell counterfeit or unlawful prescription medicines. Although the European Commission has indicated that it does not consider website blocking to be ‘an appropriate strategy’ for counterfeit products,173 the possibility now exists of seeking injunctions founded on national or EU trade mark rights.

14.156  Foreign publications. A claimant who successfully maintains an allegation of defamation in respect of an internet publication will normally wish to have that material removed. Although this can be done relatively easily where the identity of the publisher is known or the host of the material is domiciled in the United Kingdom, in cases of anonymous publication it may be far from trivial to achieve174—especially if the platform on which material is hosted proves uncooperative or cannot be contacted outside the jurisdiction. In this situation, the claimant may wish to obtain a blocking order against a domestic internet intermediary to attempt to reduce the local dissemination of the material.

14.157  Jurisdiction. It is not presently established whether or not the courts of England and Wales have a basis upon which to exercise jurisdiction to grant a blocking order in respect of defamatory internet material. However, the expansive approach to section 37(1) of the Senior Courts Act 1981 taken in Cartier suggests that power exists to make such an order, even where the respondent is not itself a wrongdoer. If this approach is upheld on appeal, then it would be arguable that a blocking remedy would be available for defamation where an ISP (or other service provider) has actual knowledge that its service is being used to facilitate the publication of the defamatory material. Although the ratio of that decision was formally limited to infringements of intellectual property rights, there is no obvious reason why the same analogy with the equitable protective jurisdiction could not be applied to other classes of wrongdoing which are facilitated by an internet intermediary.

14.158  Other jurisdictional limits. A further difficulty in the context of defamation may lie in showing that there has been substantial publication of foreign-hosted material in the United Kingdom which is sufficient to satisfy the threshold requirements for a claim in defamation. Additionally, insofar as section 10 of the Defamation Act 2013 applies to claims for a blocking injunction made against an intermediary, the claimant may be required to show that it would not be reasonably practicable to commence action against the primary author, editor, or publisher of the material before a court will have subject matter jurisdiction to hear the claim.

14.159  Effectiveness of blocking. Even if a blocking remedy is available as a matter of jurisdiction, it should be carefully considered whether blocking of defamatory material would be effective in a given case. It is often easier to mirror and redistribute textual information—which may be encapsulated in an idea, allegation, or rumour—than binary content such as films and music. Any order could be trivially circumvented by anyone who had previously accessed the material, simply by copying and pasting that material onto a social network, document repository, or comment box on an unblocked webpage, or by summarising its import in any medium. It would be very difficult indeed to halt the spread of a defamatory rumour once it has received widespread internet publication.

14.160  Futility of blocking. Second, as Giggs (previously known as CTB) v News Group Newspapers Ltd illustrates, the public imagination is disproportionately captured by attempts to restrict access to information, whether tortious or not. The words with which Tugendhat J began his judgment are instructive:

There can be few people in England and Wales who have not heard of this litigation. The initials CTB have been chanted at football matches when Mr Giggs has been playing for Manchester United.175

14.161  The ‘Streisand effect’. It is a curious irony of many internet remedies that the greater a claimant’s attempt to suppress information, the more likely it is to propagate or ‘go viral’.176 The phenomena of internet memes and social networks have only exacerbated this tendency. Indeed, in the interval of time between an order being made and a webpage becoming inaccessible, millions of people might access and distribute its contents. It can therefore be expected that attempts to block access to defamatory material by injunction will sometimes be counterproductive. Although the same logic applies to copyright-infringing materials, the effort and technical resources required to share text-based information are much lower.

14.162 Nevertheless, blocking may at the very least substantially reduce the future damage that would otherwise be suffered by a claimant who has been defamed by an anonymous tortfeasor. Indeed, blocking may even be more effective than in copyright cases since, curiosity aside, third parties may have less intrinsic desire to find material of which they are unaware.

14.163  The need for a cautious approach. There are independent reasons why it may be unwise to grant blocking remedies in defamation cases without careful scrutiny—not least the potential for abuse, chilling effects on freedom of expression, prior restraints on publication, and the difficulty of identifying related but non-identical publications. There is also the problem that many publications are made on websites which include a range of non-tortious material. Even a single page will include both defamatory and lawful statements, or justified and unjustified imputations. For this reason, it is suggested that any blocking remedies for defamatory material on shared websites ought to be strictly confined to material that has been the subject of a final judicial determination.

14.164  Statutory basis. There is an explicit statutory basis under United Kingdom and European law for blocking remedies against internet intermediaries who process personal data unlawfully, including where data are incomplete or inaccurate. Although this remedy has not yet been ordered against an ISP, there is every reason to suppose that blocking orders are available under sections 14(1) and 14(4) of the Data Protection Act 1998 against ISPs who are data controllers and who process personal data in the specified circumstances. The right to deletion of personal data is discussed further in chapter 10.177

14.165  Non-statutory remedies. Circumstances may also be imagined in which a non-statutory blocking remedy is required to uphold a claimant’s rights under article 8 of the European Convention. For example, a misuse of private information may result in it being published on the internet without the consent of the data subject, and the only effective remedy may be to prevent interested third parties from accessing that information through their ISP or a search engine.

14.166  Effectiveness of blocking. Potential claimants should bear in mind the same warning made in relation to defamation claims; namely, that blocking would only provide part of the redress that a claimant seeks. Private material would still be accessible using the services of other ISPs or search engines, or in other jurisdictions, which means that some amount of information leakage back into the forum is inevitable. In cases involving confidential information or privacy, this risk is even more acute. It is sometimes said that a secret is like an ice cube178—inherently perishable and worthless once melted—and partially blocking access to material it is akin to leaving the freezer door open: it is only a matter of time before the remedy is overtaken by outside forces.

14.167  Restrictions on jurors. Website blocking is sometimes mentioned in the context of maintaining civil order, particularly as means of preventing jurors from accessing prejudicial materials during criminal trials.179 Although it is unclear exactly how such a scheme would operate, it appears to contemplate a form of virtual sequestration, which would require jurors’ ISPs and search engines to block access to a list of prohibited search keywords,webpages, and files for the duration of a trial. Whether this could ever be effective would depend on how comprehensive the blocklist is and whether it was possible for intermediaries to block access on a per-user basis. Juror education and proper directions are likely to be more effective and less intrusive than monitoring and blocking domestic connections.

Notes
2

See Cartier International AG v British Sky Broadcasting Ltd [2014] EWHC 3354 (Ch) (‘Cartier’). An appeal to the Court of Appeal brought by the respondent ISPs is due to be heard in April 2016.

3

Enforcement Directive art 9(1)(a) provides for equivalent interim remedies to prevent imminent infringements of an intellectual property right.

4

See Cisco Systems Inc, ‘How NAT Works’ (IP Addressing Services, 29 March 2011) <http://cisco.com/image/gif/paws/6450/nat-cisco.pdf>.

5

IP addresses normally consist of four period-separated octets ranging from 0–255. Eg at the time of writing, the website for the Supreme Court of the United Kingdom had the IP address 137.117.231.217. Most web servers are assigned static IP addresses, meaning that they generally do not change (though they can change if a host moves within a network or changes upstream ISPs). Most internet users have dynamic IP addresses, which change from session to session. See generally Douglas Comer, Internetworking with TCP/IP—Volume 1: Principles, Protocols and Architectures (5th ed, 2006) chs 3, 4, 7.

6

See Cisco Systems Inc, ‘Use a Static Route to the Null0 Interface for Loop Prevention’ (IP Routed Protocols, 14 October 2004) <http://cisco.com/application/pdf/paws/14956/route_to_null_interface.pdf>.

7

See, eg, Pablo Ayuso, ‘The Netfilter.org Project’ (Netfilter, 2010) <http://www.netfilter.org/>.

8

Richard Clayton, Anonymity and Traceability in Cyberspace (2005) 121–2.

9

Benjamin Edelman, ‘Web Sites Sharing IP Addresses: Prevalence and Significance’ (Berkman Center for Internet & Society, 12 September 2003) <http://cyber.law.harvard.edu.libproxy.ucl.ac.uk/archived_content/people/edelman/ip-sharing/>.

10

Jillian York, ‘Argentine ISPs Use Bazooka to Kill Fly’ (Electronic Frontier Foundation, 19 August 2011) <http://www.eff.org/deeplinks/2011/08/argentina-isps-ip-overblocking>.

11

The main technique is known as a reverse IP address lookup, which relies upon third parties to aggregate searchable records of websites’ IP addresses. However, such records can be incomplete or out of date.

12

Ofcom, ‘Site Blocking’ to Reduce Online Copyright Infringement (May 2011) 28.

13

Ofcom, n 12, 29.

14

See John Klensin, ‘Role of the Domain Name System (DNS)’ (2003) RFC 3467 Internet Engineering Task Force 2. IPv6 introduces added complications and storage requirements.

15

Paul Mockapetris, ‘Domain Names—Concepts and Facilities’ (1987) RFC 1034 Internet Engineering Task Force 4, 18.

16

See, eg, Jeremy Kirk, ‘Anonymous Threatens to DDoS Root Internet Servers’ (PC Advisor, 20 February 2012) <http://pcadvisor.co.uk/news/security/3338618/anonymous-threatens-ddos-root-internet-servers/>. But see Kim Davies, ‘There Are Not 13 Root Servers’ (ICANN Blog, 15 November 2007) <http://blog.icann.org/2007/11/there-are-not-13-root-servers/>.

17

See Ethan Zuckerman, ‘Intermediary Censorship’ in Access Controlled: The Shaping of Power, Rights, and Rule in Cyberspace (2010) 71, 73–4; OpenNet, West Censoring East: The Use of Western Technologies by Middle East Censors (2011) 3.

18

See Richard Clayton, ‘Failures in a Hybrid Content Blocking System’ (2005) 3.

19

Rich Miller, ‘DNS Poisoning Scam Raises Wariness of “Pharming”’ (Netcraft, 7 March 2005) <http://news.netcraft.com/archives/2005/03/07/dns_poisoning_scam_raises_wariness_of_pharming.html>. See also Robert McMillan, ‘DNS Attack Could Signal Phishing 2.0’ (PC World, 11 December 2007) <http://pcworld.com/article/140465/>.

20

See Giuseppe Ateniese and Stefan Mangard, ‘A New Approach to DNS Security (DNSSEC)’ (Paper presented at the Conference on Computer and Communications Security, 5–8 November 2001, Philadelphia) <http://www.cs.jhu.edu.libproxy.ucl.ac.uk/~ateniese/papers/dnssec.pdf>.

21

Educause, ‘7 Things You Should Know about DNSSEC’ (January 2010) <http://net.educause.edu/ir/library/pdf/EST1001.pdf>.

22

Steve Crocker et al, Security and Other Technical Concerns Raised by the DNS Filtering Requirements in the PROTECT IP Bill (May 2011) 5–6.

23

Ofcom, n 12, 33–5.

24

Victoria Espinel, Aneesh Chopra, and Howard Schmidt, ‘Combating Online Piracy while Protecting an Open and Innovative Internet’ (The White House Blog, 14 January 2012) <http://www.whitehouse.gov/blog/2012/01/14/obama-administration-responds-we-people-petitions-sopa-and-online-piracy>

25

Many security vendors now offer appliances which offer this functionality by default: see, eg, Elitecore Technologies Pvt Ltd, ‘Complete Visibility & Control over HTTP & HTTPS’ (Web Filtering, 2010) <http://www.cyberoam.com/webfiltering.html>.

26

See Richard Clayton, ‘The IWF Blocking List: Recent UK Experiences’ (Paper presented at INEX, Dublin, 30 June 2009) 6–8.

27

Open Rights Group, ‘IWF Censors Wikipedia, Chaos Ensues’ (8 December 2008) <https://www.openrightsgroup.org/blog/2008/iwf-censors-wikipedia-chaos-ensues>.

28

Ofcom, n 12, 43.

29

See Nate Anderson, ‘New Zealand relies on BGP router protocol to filter the “Net”’ (Ars Technica, 12 March 2010) <http://arstechnica.com/tech-policy/news/2010/03/new-zealand-relies-on-bgp-router-protocol-to-filter-the-net.ars>.

30

Internet Watch Foundation, ‘15th Anniversary’ (26 October 2011) <http://www.iwf.org.uk/about-iwf/iwf-15th-anniversary>.

31

Becky Hogge, ‘IWF Censors Wikipedia, Chaos Ensues’ (Open Rights Group, 8 December 2008) <http://www.openrightsgroup.org/blog/2008/iwf-censors-wikipedia-chaos-ensues>.

32

Clayton, n 18, 9–12.

33

Enex Pty Ltd, Internet Service Provider Content Filtering Pilot (2009) 21–3.

34

Cory Doctorow, ‘Censorship is Inseparable from Surveillance’ (The Guardian, 2 March 2012) <http://guardian.co.uk/technology/2012/mar/02/censorship-inseperable-from-surveillance>.

35

Ofcom, n 12, 40–1.

36

See NetClean Technologies Sweden AB, ‘NetClean Whitebox: Child Sexual Abuse Web Site Filtering System’ (2010) <http://www.watchdoginternational.net/images/stories/ncwb2.pdf>.

37

See, eg, Websense Inc, ‘Web Security Gateways’ (2012) <http://websense.com/content/web-security-gateways-features-and-benefits.aspx>; Wikipedia Contributors, ‘Uses of Proxy Servers’ (22 February 2012) <http://en.wikipedia.org/wiki/Proxy_server#Uses_of_proxy_servers>.

38

See, eg, The Tor Project Inc, ‘Tor Browser Bundle’ (2012) <https://torproject.org/projects/torbrowser.html.en>.

39

See United States Navy, ‘Onion Routing’ (Center for High Assurance Computer Systems, 12 October 2006) <http://www.onion-router.net/>.

40

See Stack Exchange Inc, ‘How Does a Country Block/Censor an Encrypted Website (HTTPS)?’ <http://security.stackexchange.com/questions/64568/how-does-a-country-block-censor-an-encrypted-website-https>.

41

See, eg, Jon Brodkin, ‘Iran Reportedly Blocking Encrypted Internet Traffic’ (Ars Technica, 10 February 2012) <http://arstechnica.com/tech-policy/2012/02/iran-reportedly-blocking-encrypted-internet-traffic/>; Jacob Hoffman-Andrews, ‘ISPs Removing their Customers’ Email Encryption’ (Electronic Frontiers Foundation Deeplinks Blog, 11 November 2014) <https://www.eff.org/deeplinks/2014/11/starttls-downgrade-attacks>.

42

Hamed Saber and Nima Fatema, ‘Access Flickr! 1.11’ (Firefox Add-Ons, 10 March 2009) <https://addons.mozilla.org/en-us/firefox/addon/access-flickr/>.

43

TorrentFreak, ‘Newzbin2 Release Encrypted Client to Defeat Website Blocking’ (14 September 2011) <http://torrentfreak.com/newzbin2-release-encrypted-client-to-defeat-website-blocking-110914/>.

44

TorrentFreak, ‘Newzbin2 Team up with The Pirate Bay to Defeat Site Blocking’ (5 October 2011) <http://torrentfreak.com/newzbin2-team-up-with-pirate-bay-to-defeat-site-blocking-111005/>.

45

Twentieth Century Fox Film Corporation v British Telecommunications plc [No 1] [2011] EWHC 2714 (Ch), [35] (Arnold J) (‘Newzbin2 [No 2]’).

46

1988 Act s 97A(3). See chapter 12, section 2.

47

Twentieth Century Fox Film Corporation v British Telecommunications plc [No 1] [2011] EWHC 1981 (Ch), [146] (Arnold J) (‘Newzbin2’).

48

Information Society Directive arts 5(1)(a), 8(3).

49

Information Society Directive recital (59) (emphasis added).

50

UPC Telekabel, [34]–[35].

51

Tele2, [43]–[46].

52

It is arguable, by analogy with cases concerning Norwich Pharmacal disclosure and the ordinary requirements for interim relief, that an interim blocking order would be available under section 97A if the applicant could show a seriously arguable case of primary infringement , irreparable harm, and that the balance of convenience favoured relief.

53

UPC Telekabel, [31].

54

See Intellectual Property (Enforcement, etc) Regulations 2006 (SI 2006/1028) sch 3, para 6.

55

Paramount Home Entertainment International Ltd v British Sky Broadcasting Ltd [2013] EWHC 3479 (Ch), [43] (Arnold J).

56

Case C-557/07, EU:C:2009:107 [2009] ECR I-1227 (‘Tele2’).

57

Tele2, [46].

58

Tele2, [44].

59

Case C-314/12, ECLI:EU:C:2014:192 (‘UPC Telekabel’).

60

UPC Telekabel, [32].

61

UPC Telekabel, [34].

62

See Cartier, [155] (Arnold J), discussed in section 4.1.

63

UPC Telekabel, [35].

64

UPC Telekabel, [36].

65

UPC Telekabel, [38]–[39].

66

Case C-324/09 [2011] ECR I-06011 (‘L’Oréal’). Article 11 is analogous to art 8(3), since it also requires that the service provider be one ‘whose services are used’ to infringe.

67

L’Oréal, [144].

68

See Newzbin1, [134] (Kitchin J), discussed at para 14.77.

69

See EMI, [87] (Arnold J), discussed at para 14.94.

70

See Newzbin2, [149] (Arnold J), discussed at para 14.82; Dramatico [No 2], [7] (Arnold J), discussed at para 14.88.

71

EMI, [89] (Arnold J), discussed at para 14.95.

72

See, eg, Copyright Act 1956 (UK) ss 5(2), (3), (4), 16(2), (3), (4) (requiring actual knowledge that an article is or would be an infringement).

73

Meccano Limited v Anthony Hordern & Sons Ltd (1918) SR (NSW) 606, 611 (Harvey J).

74

Cooper v Whittingham (1880) 15 Ch D 501, 505 (Jessel MR) (emphasis in original). Doubtless, the same could not be said today.

75

[1905] 1 Ch 519.

76

Albert v Hoffnung & Co (1921) SR (NSW) 79, 81 (Harvey J). See also Barson Computers (NZ) Ltd v John Gilbert & Co Ltd [1985] FSR 489, 508–9 (Pritchard J).

77

Twentieth Century Fox Film Corp v Newzbin Ltd [2010] FSR 21 (‘Newzbin1’). See chapter 6, paras 6.50 to 6.62.

78

Newzbin1, [134] (Kitchin J).

79

Newzbin1, [135] (Kitchin J).

80

Newzbin2, [11], [204] (Arnold J). The form of order was determined in [2011] EWHC 2714 (Ch), [56] (Arnold J).

81

Newzbin2, [148], [157] (Arnold J).

82

Newzbin2, [149] (Arnold J).

83

See EMI Records Ltd v British Sky Broadcasting Ltd [2013] EWHC 379 (Ch), [89] (Arnold J) (‘EMI’).

84

Newzbin2 [No 2], [6] (Arnold J).

85

Cf L’Oréal, [141].

86

Newzbin2 [No 2], [9] (Arnold J).

87

Newzbin2 [No 2], [12] (Arnold J).

88

Newzbin2 [No 2], [17] (Arnold J).

89

Twentieth Century Fox Film Corporation v British Sky Broadcasting Ltd (Unreported, 12 December 2011, Vos J).

90

Twentieth Century Fox Film Corporation v TalkTalk Telecom Group plc (Unreported, 9 February 2012, Arnold J).

91

Dramatico Entertainment Ltd v British Sky Broadcasting Ltd [2012] EWHC 268 (Ch); [2012] 3 CMLR 14 (‘Dramatico [No 1]’).

92

Dramatico [No 1], [69]–[71] (users), [81], [83] (website operators).

93

Dramatico Entertainment Ltd v British Sky Broadcasting Ltd [No 2] [2012] EWHC 1152 (Ch); [2012] 3 CMLR 15, [7] (‘Dramatico [No 2]’). See chapter 6, paras 6.63 to 6.66 for discussion of the first judgment.

94

Dramatico [No 2], [13] (Arnold J).

95

EMI Records Ltd v British Sky Broadcasting Ltd [2013] EWHC 379 (Ch); [2013] ECDR 8 (‘EMI’).

96

EMI, [38] (Arnold J).

97

EMI, [41] (Arnold J).

98

EMI, [51] (communication to the public), [70] (authorisation), [74] (joint tortfeasorship) (Arnold J).

99

EMI, [66] (Arnold J).

100

EMI, [87] (Arnold J).

101

EMI, [89] (Arnold J).

102

[2013] EWHC 2058 (Ch), [2013] ECDR 14 (‘FirstRow’).