Ranking: Personalisation

Analogous to the ‘pre-history of search engines’ described in Chapter 1, where the ‘office’

collected the addresses of citizens, in contemporary digital society, the IP (Internet Protocol) address for communication is a unique identifier or numeric label assigned to any type of device that is connected to a network. ⁹⁸ The IP address is thereby comparable to the physical address of a neighbourhood, such as the name of the country, city, street, house number and floor/

apartment. The IP address usually consists of groups of numbers (4), with a network part and a host part, where one can find the ISP, the country, the region and state, the city and the location (longitude and latitude). It is read from right to left, gradually increasing precision as it reaches the final numbers to pinpoint the location of the device (and thereby the user) but it does not reveal a person’s name, the exact address, phone number or email address. The IP is the third or

98 As shown in Chapter 1 with the history of the postal services and later the telegraph and telephone cables, the present-day internet builds upon these infrastructures, through private and public means.

‘network’ layer that determines the path of the data packet and address, transmitting data through channels, which are occupied during transmission and can be used for other traffic afterwards.⁹⁹ (Figure 32)

To return to my Google search habit, Chun offers the reader a key insight about the

‘rehabituation’ of individuals through the changing nature of the IP address:

When the Internet was first conceived, IP addresses, even when fixed, were not viewed as permanently tethered to a computer, let alone a user. With the advent of changes to IP addressing, and more importantly the emergence of cross-platform logins, cookies, and other means of tracking through ‘unique identifiers,’ it has become easier to tie users to their actions. This traceability has entailed the massive rehabituation of individuals into authenticated users through the expansion (2016:57).

When users type in keywords, Google Search connects multiple parties through data trails comprised of ‘signals’ or markers that identify users, such as the IP address.¹⁰⁰ The IP address is the key signal that affected my results from my ‘experiment in living’. Acting as a tracking device (as well as GPS) and serving as a location marker, the IP address determines the device and identity of the user online. The IP address also facilitates data transfer, with search engines creating a database on each user containing their respective queries and search histories, which enables the construction of ‘the personalised subject’.

At the end of Chapter 3, Brin and Page first mention personalisation in their seminal text with their ‘trusted user’ who would ‘provide feedback and improve the quality of the search experience’ (1998). I explained how the more people used Google search, the larger its

proprietary ‘Database of Intentions’ became and that this ‘information represents the real-time history of post-Web culture’, as a

massive click stream database of desires, needs, wants, and preferences that can be discovered, subpoenaed, archived, tracked and exploited for all sorts of ends (Battelle 2005:7).

One of these ‘ends’ is ‘personalisation’, which Google publicly announced on December 4, 2009. Capturing the searching subject’s IP address as well as maintaining a log of previous queries, personalisation adapts them into real-time search results, even if one is not signed into a Google account. With personalisation, anticipatory searches are based on previous search

histories and users remain within what Eli Pariser describes as the ‘filter bubble’ ––where they receive positions, opinions and news that they already know and support (2011). Pariser’s critique is that this 21^st century zeitgeist creates a sense of deprivation, leading to the ‘distortion effect’, one of the challenges posed by personalised filters.

99 In 1974 Vint Cerf and Bob Kahn applied another layer on top, the TCP (Transmission Control Protocol), which enables the sharing of transmission resources that use packet switching (connectionless datagrams) among the nodes and is connection-oriented. It is the DNS (Domain Name System) protocol, located in part of the Application layer (7), that translates the (human readable) domain names into a numerical network-specific address (IP) that enables it to map it to the Data Link layer (2) with the help of the ARP (Address Resolution Protocol), a communication protocol.

100 ‘The standard search engines (Google, Bing, Yahoo, etc.) track and record everything you search for. Your typical search engine records the following information any time you use it: Your IP address, User agent, Unique identifier (stored in browser cookies) and Search terms’ (Taylor 2018).

Like a lens, the filter bubble invisibly transforms the world we experience by controlling what we see and don’t see. It interferes with the interplay between our mental processes and our external environment (ibid:82-83).

In The Filter Bubble Pariser also shows how different people, on different computers, in diverse locations in the world, receive different search results using Google Search (2011). This is echoed by Halavais who elucidates that personalisation is ‘an ongoing response to users that focuses on their individual needs’ (2009:51), which include not only their search queries and history but context.¹⁰¹ I received locative data, a known signal based on country and language that triggers a ‘response’ to Google’s personalisation algorithms. My results for Re:search - Terms of Art show that although certain URLs were the same in the dataset (Appendix E), the ranking was different. Therefore, I propose that the main difference between Google Search and Tor (Disconnect search) is that Tor obscures the IP address without pinpointing location.

Disconnect states that it does ‘detect non-personally identifiable geo-location information to optimize our services, but [unlike Google] we definitely don’t collect your precise geo-location or associate geo-location information with a particular user’ (Disconnect Search 2016).

Figure 39: Keyword postmedia. Google results page 2.

With the keywords containing the prefix ‘post’ (‘postdigital’, ‘postmedia’, ‘postinternet’ and

‘posthumanism’), I received Google results reflecting my search histories and location thereby showing user personalisation. Google Search results for ‘postinternet’ were often relegated to the Netherlands, where I live most of the timeeven though I was searching and collecting the data in Denmark.¹⁰² Besides ‘postinternet’ returning Danish postal results and information regarding governmental e-post and internet that shows locative data was being collected, I obtained a few results for mail delivery with the keyword ‘postmedia’, as well as URLs referring to Danish media outlets. Besides geolocation, I postulate that the ‘semantic’

interpretation of the keyword determined the divergent ranking of my results. My ‘keywords’

were not ‘trending’ in the commercial sense or terms of financial speculation, nor would they be included in the high bidding wars of AdWords, as I explain in Appendix E. Instead, because I used terms from contemporary art beginning with the word ‘post’, the algorithmic interpretation

101 In exchange for their data, users receive free search and ostensibly ‘tailored’ advertising, turning themselves into commodities for advertisers if they don’t delete ‘cookies’ or installing adblocking plug-ins that would inhibit it.

This personalisation is then a currency, with data correlated through algorithmic technologies and acquired by marketers, or third parties (Ridgway 2015).

102 Postinternet reflects the way that the internet as a tool can produce art, but is not specifically situated in it (as Net Art would be) and its influence of the internet on society and culture, both online and offline.

of the word ‘post’ varies (postal services, ‘post’ as a common naming convention for

newspapers, as a widely used contraction of ‘posting’, as in blogging or commentary, and as a term for histories and intellectual movements). Appendix F provides more examples.(Figure 39) As shown above and described in the previous section, language and its interpretation become important criteria along with the monetisation of certain keywords that reflects users’ thoughts as queries, giving these keywords the power of a certain ‘semantic governmentality’.

Not only is it the case that every word in every language now has its price that fluctuates according to the laws of the market, but additionally, both search results and the

corresponding advertisements shown are now optimised according to their potential market value based on pre-emptively calculated individual ‘user relevancy’ (Feuz et al.

2011).

With personalisation, Google claims that every search and individual user is unique, therefore it seeks to deliver customised results for each person that satisfies their interest. As demonstrated above, if personalisation does exist is very difficult to capture because it is hard to carry out an experiment and measure the results, as my ‘experiment in living’ attempted to do. Going beyond these claims, one particular study was able to show the effects and degrees of personalisation on the one hand and on the other, ‘deindividualised’ results.