The SEO effect:

The optimised, the under-optimised and the un-optimised web

Team Members

Almada, Pablo; Caroleo, Laura; De Amicis, Camilla; Giorgi, Giulia; Kristensen, Lisa Merete; Sathish, Christa and Teigeler, Lena

Facilitator: Prof. Richard Rogers

Contents

1. Introduction

Search engine critique has long revolved around privileging mechanisms and how to identify them. Does Google privilege the surface web, the well-linked, the optimised and/or its own properties? One of the means to become privileged – search engine optimisation – was once considered something of a dark art, or at least one that could be undertaken through ‘black hat’ techniques that if caught by Google would result in a website being throttled or even removed from engine returns. Since that time, Google changed how it approached search engine optimisation by encouraging websites to be designed in a way that is Google-friendly (Rogers, 2018). This website ‘googlization’ – optimisation for Google – is the subject of analysis of this project. Put differently, we are interested in how optimisation drives search engine placement rather than, say, quality content or even quality inlinks.

At the beginning of this project we decided to split the group into two subgroups. The first subgroup is: “Optimised for Hate? The presence of hate group websites in Google (and Bing) results”. The second subgroup is: “Optimised Politics: striving to be seen”.

In the first case study “Optimised for Hate? The presence of hate group websites in Google (and Bing) results”, we focused on American groups and organisations identified as hateful by the non-profit Southern Law Poverty Center, and analysed to which extend they are optimised for Google and which positions their websites take in Google results. Moreover, we selected “watcher” groups that also appeared in the results and that provide educational information on these hate groups. We analysed their optimisation level and their Google ranking when querying hategroups’ names. For those seeking hate groups, how accessible are their websites? Are there particular hateful ideologies more present than others in the top ten returns in Google (as well as Bing)? Do watcher organisations appear in the results when searching for those hate groups? That is, are there alternative narratives which appear?

The second (sub)project “Optimised Politics: striving to be seen” investigates the extent to which political actors have employed SEO to boost their visibility on search engine results, using Italy as a case study. Specifically, we are interested in examining how political actors employ SEO techniques to promote their visibility online and, contextually, to foster the circulation of their ideas and agendas around hot topics. Our empirical investigation focuses on five social and political issues, which are known for being particularly controversial and divisive in the Italian landscape: abortion, euthanasia, work, LGBTQ+, and migration.

3. Research Questions

Project 1: Optimised for Hate?
  1. Comparative optimisation of hatewatch vs. hate groups: To what extent are the websites of hate watchdog groups as well as those on SLPC’s hatewatch lists optimised?

  2. Optimisation as watchfulness: Which watchdog groups are (most) associated with queries of groups in the United States as identified by SLPC’s Hatewatch project?

  1. Optimised ideologies: Which ideologies (according to SLPC’s Hatewatch project) are the highest ranking when querying the group names on SLPC’s hatewatch lists?

  1. Finding hate and content moderation: When querying the names of groups on SLPC’s hatewatch list, are the hate groups at the top of the returns? Are they findable in the top 10 or 20 results? Are the hate groups websites in the results? Which position are they in?

  1. Google vs. Bing: How do the rankings of watchdog- and hate group sites compare in Google versus Bing? Which would be the preferred engine of hate-seekers (so to speak)?

Project 2: Optimised Politics
  1. To what extent do political websites employ SEO?

  2. Are right-wing websites more optimised than left-wing websites?

  3. Who implements the most ‘expensive’ modules?

4. Initial datasets

Project 1: Optimised for Hate?

We started with a list of hate groups identified by the non-profit Southern Law Poverty Center. The list contains 299 hate groups that are assigned to 15 different ideological categories: Anti-Immigrant, Anti-LGBT, Anti-Muslim, Anti-Semitism, Christian Identity, General Hate, Hate Music, Ku Klux Klan, Male Supremacy, Neo, Confederate, Neo Nazi, Neo Volkisch, Racist Skinhead, Radical Traditional Catholicism and White Nationalism.

To gather information on their ranking on Google and Bing and their level of search engine optimisation, we used the ‘The SEO-effect’ software developed by Sebastian Sünkler, University of Applied Sciences Hamburg.

Project 2: Optimised Politics

We have selected five particularly divisive social issues in the Italian context: abortion, migration, LGBTQ+, euthanasia and labour. For each topic, we extracted a list of keywords (one- and multi-word expressions). To do this, we selected the main Italian parties in terms of presence in the Italian Parliament (%) and the polls: Partito Democratico (centre-left) and Fratelli d'Italia (right-wing). We have examined their official websites - in particular, their political manifesto and news section - to understand how these actors communicate on these topics.

We ran the software using the query list as input, and the output was a .csv file containing 1,028 websites on the five topics.

5. Methodology

To describe the level of optimisation of the organisations’ websites, we ran their names in a tool developed by Sebastian Sünkler and colleagues at HAW Hamburg that analyses the HTML of a website for SEO signals, outputting levels of optimisation per website. The tool takes 21 indicators of optimisation into account, e.g. traces of SEO and analytics tools in the HTML comments, information on website level, like robots.txt or sitemap files, or the loading time of a website. Websites are classified in four classes according to these factors: definitely optimised, probably optimised, probably not optimised and definitely not optimised (Lewandowski, D. et al. 2021 , p.15).

Project 1: Optimised for Hate?

We queried the names of 299 organisations listed in the SPLC’s hate map and collected the 20 first results of Google and Bing for each query, their position in both search engines and their level of optimisation. From the search results, we identified 12 different ‘watcher’ sites that warn about the hate groups and provide educational information on them. These are: adl.org, Counterextremism.com, equityfwd.org, immigrationresearch.org., influencewatch.irg, islamophiabnetwork.com, plotagainstimmigrants.com, powerbase.info, pro-lies.org, sourcewatch.org, splcenter.org and trackingterrorism.org.

We calculated the average ranking and the average optimisation level of the groups’ official websites per ideology, as well as the average ranking and average optimisation level of the watcher groups appearing in query results. By analysing how many watchers appear in the Top 10 search results on average, we further determined a level of “watchfulness” per ideology.

Moreover, we turned to the optimisation level of each watcher and analysed, which of them appear the most in our search results. Finally, we compared the rankings of hate watcher sites and hate group sites in Google and Bing results. Which would be the preferred engine of hate-seekers?

Project 2: Optimised Politics

We have selected 5 social issues which we know are especially divisive in the Italian context. They are: abortion, migration, LGBTQ+, euthanasia, and work. For each topic, we wanted to extract a list of keywords (one-word and/or multi-word expressions). To do so, we selected the top Italian parties with respect to presence in the Italian parliament (%) and in polls: Partito Democratico (center-left) and Fratelli d’Italia (right). We then looked at their official sites - especially their political manifesto and the news section - to understand how these actors communicate on these topics. To simulate a search carried out in Italy, we used a VPN set in Italy; furthermore, we employed a clean browser and the hide mode - to reduce potential biases due to users’ profiling.

We extracted a list of polarised expressions (left/right) and included one neutral term for each topic.

Topic

Polarization

Query







ABORTION

neutral

Aborto

left

Diritti umani donne

left

Difesa del diritto all’aborto

left

Diritto di abortire

right

Pro Vita

right

Pro Life

left

Pro Aborto







EUTHANASIA

neutral

Eutanasia

left

Diritto all’eutanasia

left

Diritto di scegliere fine vita

left

Liberi fino alla fine

right

Sacralità della Vita

right

Tutela della Vita

right

Omicidio Consenziente








WORK

neutral

Lavoro

left

Aiutiamo i lavoratori

left

Più diritti e più uguaglianza

left

Caro Vita

right

Oppressione fiscale

right

DIfesa Made in Italy

right

Metadone di Stato
















LGBTQ+

neutral

LGBTQ+

left

L’amore non ha sesso

left

L’amore non ha genere

left

Sì DDL Zan

left

Diritti persone omosessuali

left

Legge unioni civili

left

Tutelare famiglie arcobaleno

left

Diritto famiglie omogenitoriali

right

No DDL Zan

right

Stop Gender

right

Famiglia Naturale

right

Utero in Affitto

right

Utero In Affitto Reato

right

No al matrimonio tra persone dello stesso sesso

right

Reato Universale












MIGRATION

neutral

IMMIGRAZIONE

left

Sì ius soli

left

Sì ius scholae

left

Porti aperti

left

Richiedenti asilo

left

No barriere immigrazione

left

Italia più umana e sicura

right

No ius soli

right

Stop agli Sbarchi

right

Prima gli italiani

right

No islamizzazione

right

No Invasione

right

Cittadinanza Facile

Table 1: List of the keywords per topic.

Afterwards, we ran the software using the query list as input and extracted, for each query entry, the first 20-30 website links returned by Google Search. The output is a .csv file including the following metadata for each queries keyword: the position of the website in the ranking, the probability of optimisation (ranging from “most probably not optimised” and “most probably optimised”), and a list of reasons (theoretically) responsible for the optimisation.

We then moved to data analysis. For the analysis of the data, we have considered the entirety of the results, which counts 1.028 websites across the 5 topics. In the following Table, we detail the number of results per topic: the differences in the results depend on the different numbers of keywords per topic and (probably) in the fluctuation in the layout of the Google Search page.

Topic

Frequency

Abortion

140

Euthanasia

154

LGBTQ+

325

Migration

274

Work

135

Total

1028

Table 2: Number of results per topic

Looking at the actors, we manually divide the political actors (official sites and social media of leaders and parties) from the non-political actors (mostly media outlets, associations and institutions); for the topics abortion and euthanasia, we have noticed that the main actors are NGOs. We found a striking predominance of non-political actors (97,96%) as opposed to political actors (2%). Among the political actors, around 33% are left-wing parties/actors and 64% right-wing.

Consciously aware of the possibility that sites classified as non-political may in fact have political links or directions, this classification was chosen precisely because relying on a declared subdivision provides an accurate view of the way the Italian political landscape communicates.

6. Findings

Project 1: Optimised for Hate?

1. Comparative optimisation of hatewatch vs. hate groups. To what extent are the websites of hate watchdog groups as well as those on SLPC’s hatewatch lists optimised?

The majority of watcher sites are optimised but do not show signs of professional SEO services or use of engineering resources.

Watcher

Optimisation level

adl.org

probably_optimized

counterextremism.com

probably_optimized

equityfwd.org

probably_optimized

immigrationresearch.org

probably_not_optimized

influencewatch.org

probably_optimized

islamophobianetwork.com

probably_optimized

plotagainstimmigrants.com

probably_optimized

powerbase.info

probably_not_optimized

pro-lies.org

probably_optimized

sourcewatch.org

probably_not_optimized

splcenter.org

probably_optimized

trackingterrorism.org

probably_optimized

Figure 1: Optimisation of hatewatcher websites

To calculate an average optimisation level of hate groups, we translated the categorisation level assigned by the tool into numbers: 'most_probably_optimized':4, 'probably_optimized’=3, 'probably_not_optimized':2, ‘most probably not optimized’=1 and 'Not in search results':0. The ideologies that are most optimised are Anti-LGBT, Anti-Semitism and Neo-Confederate. When excluding those sites that cannot be found in Google results, Anti-Muslim, General Hate, White Nationalism and Racist Skindhead groups also appear to be highly optimised. The most optimised websites are those of “Euro Folk Radio”, a Christian Identity group, and “Radio Jihad/Global Patriot Radio”, an Anti-Muslim group according to SLPC. Most group websites are probably optimised (138 in total). More than a third of the group websites don’t show up in search results when querying for the groups’ names. 45 group websites are probably not optimised, none of them is most probably not optimised and only two are most probably optimised. The majority of the sites does not use professional SEO services or engineering resources. The status as being most probably optimised results from the classification system of the SEO effect tool that automatically classifies news websites as being most probably optimised. The least optimised are groups categorised as Neo-Volkisch and Neo-Nazi. and Ku Klux Klan. As only a relatively small number of Racist Skinhead and Ku Klux Klan groups’ websites appear in the Google results, we cannot make a reliable claim on their average optimisation level.

Ideology

percentage of websites not in search results

average optimisation

optimisation without sites not in search results

Male supremacy

0%

3

3

Anti-LGBT

10%

2,446808511

2,738095238

Anti-semitism

13%

2,375

2,714285714

Neo-confederate

17%

2,333333333

2,8

Anti-immigrant

22%

2,117647059

2,769230769

Christian identity

11%

2,111111111

2,714285714

Anti-muslim

31%

2

2,857142857

General hate

34%

1,901960784

2,852941176

Radical Traditional Catholicism

38%

1,625

2,6

Hate music

36%

1,6

2,666666667

White nationalism

57%

1,240740741

2,791666667

Ku Klux Klan

64%

0,846153846

2,75

Neo-nazi

70%

0,740740741

2,5

Neo-volkisch

71%

0,714285714

2,5

Racist Skinhead

78%

0,666666667

3

Figure 2: Ideologies and their optimisation level, including and excluding groups whose websites don’t appear in the top 20 search results.

2. Optimisation as watchfulness. Which watcher groups are (most) associated with queries of groups in the United States as identified by SLPC’s Hatewatch project?

Despite the optimisation the watchers rarely appear in the top 10 results for the queries of the hate watch list. When not considering SPLC (because we research their list), ADL appears in the most search results at 20%. Other watchdogs rarely appear. The highest number of watchers is presented in the top 10 results of Neo-Volkisch, Neo-Confederate and Racist Skinhead groups. Interestingly, these groups also are the least represented in the top 20 Google search results.

3. Optimised ideologies. Which ideologies (according to SLPC’s Hatewatch project) are the highest ranking when querying the group names on SLPC’s hatewatch lists? How well are they ranked in Google when querying for them?

The ideologies that rank the highest are Male Supremacy, Anti-Immigrant and Neo-Volkisch. The ideologies that are most optimised are Anti-LGBT, Anti-Semitist and Neo-Confederat. More than a third of group sites do not appear in Google search results. The least appearing ideologies, by the number of websites pertaining to that ideology, are Racist Skinheads (77% not appearing in Google search results), Neo-Volkisch (71% not appearing in Google search results), Neo-Nazis (70% not appearing in Google search results), and Ku Klux Klan (64% not appearing in Google search results). Note that these groups seem to operate on discussion fora instead, e.g. ‘Stormfront’.

Ideology

Average Ranking

Male supremacy

1

Anti-LGBT

2,285714286

Anti-semitism

3,714285714

Neo-confederate

3,8

Anti-immigrant

2

Christian identity

3,625

Anti-muslim

3,142857143

General hate

4,323529412

Radical Traditional Catholicism

8,4

Hate music

2,166666667

White nationalism

4,041666667

Ku Klux Klan

6,5

Neo-nazi

4

Neo-volkisch

1,5

Racist Skinhead

3,5

Figure 3: The average ranking of hate groups per ideology (excluding groups that don’t show up in search results).

Figure 4: The average Google ranking and SEO level per ideology. The size of the circles represents the number of groups per ideology. “Watchfulness” describes the average number of watchers appearing in the top ten Google results - the darker the blue, the higher the average number of watchers.

4. Finding hate and content moderation. When querying the names of groups on SLPC’s hatewatch list, are the hate groups at the top of the returns? Are they findable in the top 10 or 20 results? Are the hate groups websites in the results? Which positions are they in?

More than one third of hate groups’ websites don’t appear in Google search results. Some ideologies are less discoverable than others: At least 70% of the groups classified as Racist Skinhead, Neo-Volkisch and Neo-Nazi don’t appear in Google search results. Those Neo-Volkisch groups appearing in search results rank well compared to other ideologies.

Almost all group websites classified as Anti-LGBT, Christian Identity and Anti-Semitist are represented in Google search results. SPLC only classifies the group “A Voice for Men” as striving for Male Supremacy. When querying their name, the group’s website appears in the first position in Google search results and it’s classified as “most probably optimized”.

Radical Traditional Catholicism and Ku Klux Klan group websites appear at the lowest positions, while more than half of the Ku Klux Klan groups and 38% of the Radical Traditional Catholicism groups don’t show up at all.

Ideology

Average Position in Google Search Results

Percentage of Websites not in Google Search Results

Racist Skinhead

3,5

78%

Neo-volkisch

1,5

71%

Neo-nazi

4

70%

Ku Klux Klan

6,5

64%

White nationalism

4,041666667

57%

Radical Traditional Catholicism

8,4

38%

Hate music

2,166666667

36%

General hate

4,323529412

34%

Anti-muslim

3,142857143

31%

Anti-immigrant

2

22%

Neo-confederate

3,8

17%

Anti-semitism

3,714285714

13%

Christian identity

3,625

11%

Anti-LGBT

2,285714286

10%

Male supremacy

1

0%

Figure 5: Average position in Google search results of groups per Ideology and percentage of groups that don’t appear in the search results.

5. Search engine comparison (Google vs. Bing): Which search engine returns more hate? How do the rankings of hate watchdog and hate group sites compare in Google versus Bing? Which would be the preferred engine of hate-seekers?

When comparing the different results presented by Bing and Google, it becomes clear that hate group sites are rather discoverable on Bing than on Google. At the same time, Google presents a higher number of watchers when searching for a hate group’s name.

COUNT of

search_engine

Website Category

Bing.com_Example

Bing.com_Example Total

Google.com_Example

Google.com_Example Total

Grand Total

ideology

hatesite

uncategorized

watchdog

hatesite

uncategorized

watchdog

Anti-immigrant

35

83

12

130

19

126

25

170

300

Anti-LGBT

92

281

17

390

44

403

23

470

860

Anti-muslim

63

177

10

250

24

243

23

290

540

Anti-semitism

11

54

5

70

6

66

8

80

150

Christian identity

30

48

2

80

7

78

5

90

170

General hate

64

361

15

440

36

435

29

500

940

Hate music

16

60

4

80

6

102

2

110

190

Ku Klux Klan

5

87

8

100

3

106

21

130

230

Male supremacy

7

3

10

1

8

1

10

20

Neo-confederate

21

30

9

60

5

43

12

60

120

Neo-nazi

12

163

15

190

8

227

35

270

460

Neo-volkisch

6

51

3

60

2

57

11

70

130

No answer

5

5

10

2

8

10

20

Racist Skinhead

2

62

16

80

2

67

21

90

170

Radical Traditional Catholicism

12

46

2

60

6

67

7

80

140

White nationalism

85

377

28

490

20

464

46

530

1020

#N/A

9

1

10

16

2

18

28

Grand Total

466

1897

147

2510

191

2516

271

2978

5488

Figure 6: Comparison of search results in Google and Bing per ideology.

Project 2: Optimised Politics

1. The presence of political and non political websites in political themes communication
  • The resulting proportion of presence of political and non political websites

From the overall amount of websites resulting from the query research based on dividing current social themes in Italy it appears a strong unbalance in the presence of political vs non political actors. In fact the amount of non political websites present in the first 20-30 google results for the queries is much larger than the political ones. The visual model chosen to represent the resulting constellation of sites is the network. In the visualisation each dot in the network corresponds to a website resulting from the analysis, the sum of all the dots therefore corresponds to the total number of websites. In particular, sites classified as non-political are represented in light grey, and political sites in orange. Even at a glance, it is clearly understandable how much more non-political websites in Italy deal with research queries on the web.

  • Optimisation levels for political and non political websites

The network also makes it possible to visualise the relationships between the optimisation levels of political and non-political websites. Each cluster of dots corresponds to a different optimisation level, there are 4 levels and they correspond to: -most probably not optimised -probably not optimised -probably optimised -most probably optimised. The cluster to which the highest number of website results belong is the 'probably optimised' cluster, which shows that both non-political and political sites are probably well optimised and there is a general tendency and focus on optimising sites for Google. The second, albeit smaller cluster is the 'probably not optimised' one, showing that this portion of websites is probably not fully implemented. An interesting finding, however, can be found in the two still smaller clusters: in the group of “most probably optimised” websites, there is a significantly higher percentage of political websites than in the other clusters, which shows that although political websites appear in a very small percentage overall, a large proportion of them are optimised to an almost excellent degree and that politicians who decide to communicate online via direct channels pay attention to the subject of optimisation. On the other hand, in the cluster concerning the sites 'most probably not optimised', there are no political sites at all, which, in accordance with the previous findings, shows a certain awareness on the part of politicians for optimisation for Google.

Another detectable aspect concerns the relationship between websites, clusters and position in the google appearance ranking. In this case, the diameter of the dots is directly proportional to the ranking position: the higher the position of appearance in the search engine, the larger the diameter of the dot. It is interesting how some 'most probably not optimised' non-political sites appear in high ranking positions.

2. Optimisation levels for each topic query

The relationship between the level of optimisation of websites and the Google search topic involved was then analysed. The arc diagram shown here illustrates how optimised the resulting sites are for each individual topic (resulting from several associated queries). Each semicircle on the left corresponds to a topic and on the right corresponds to the optimisation level. The size of each semicircle is directly proportional to the number of resulting websites. Whilst the theme of abortion corresponds to 'most probably not optimised' websites, other themes such as LGBTQIA+ are characterised by websites with very good optimisation probability. Some bars below each topic semicircle then provide further reading, showing whether the resulting political sites with respect to the search for certain queries associated with that topic are more right-wing or left-wing. It is interesting to note, for example, how for the theme of euthanasia only non-political sites appear on google, for the theme of abortion only right-wing political sites and for the theme of work the situation is rather balanced between right and left.

3. Optimisation left vs right wing

The visualisation above shows in this case the comparison between the levels of optimisation of right wing websites and left wing ones. Firstly, it is noticeable again, how there are no political websites “most probably not optimized”. Secondly in the first shape, belonging to the left wing websites, there is a huge predominance of the lightest colour tone which represents the “probably non optimized websites”, whereas on the right side websites appear mostly divided in good or high level optimisation. It is possible to assume that, globally, right wing websites are more optimised than left wing websites.

4. Effort Level for optimisation . Left vs Right

As previously explained, for the websites each level of optimisation has been deducted by the software based on some specific reasons. Each reason consists in a characteristic of the website and has been associated with a different level of effort. Some characteristics to optimise a website can require more advanced skills or a higher monetary investment, they have been classified in Low effort, Medium effort and High effort. The graph shows the tendency for right wing websites and for left wing ones to spend less or more effort in optimisation strategies. The results show a predominant percentage of appearance of right wing websites and a tendency for right wing websites to spend more effort in optimisation strategies.

7. Discussion

Project 1: Optimised for Hate?

When searching for the names of hate groups listed by the SPLC, Google in many cases doesn’t return the websites of these groups as search results. At the same time, websites of organisations are presented that provide information on hate groups and explicitly classify them as dangerous. Both, websites of hate groups and watcher organisations use SEO, but don’t make use of professional services or engineering resources. Optimisation doesn’t necessarily correspond with a high position in Google search results. Anti-Immigrant groups perform well in Google search results, while not being especially well optimised. Those Neo-Volkisch groups appearing in the search results can be found in high positions, while they have a tendency towards probably not being optimised. Ku Klux Klan websites have a tendency to appear on lower positions, while not being less optimised than other groups appearing on higher positions.

More than a third of all group websites cannot be found on Google. This doesn’t necessarily mean that Google strongly intervenes in search results, but can also result from the fact that groups don’t have websites. Interestingly, some ideologies have a stronger tendency to not appear in search results than others: The majority of groups with a long racist tradition, like Racist Skinheads, Neo-Volkisch, Neo-Nazi and Ku Klux Klan groups cannot be found via Google. Those Neo-Volkisch and Neo-Nazi sites that are listed in Google search results are less optimised than most hate group websites in our sample, while the appearing Ku Klux Klan websites have an average optimisation level and the listed Racist Skinhead websites are well optimised compared to other ideologies. Almost all groups assigned to Anti-LGBT, Christian Identity and Anti-Semitism can be found in search results, while not presenting a particularly high optimisation level.

When comparing Google and Bing search results, it becomes clear that websites of hate groups are more visible on Bing. Watchers, at the same time, are more likely to be found in Google results. This could be interpreted as an effect of content moderation. However, it doesn’t necessarily mean that Google has actively decided to display watchers instead of certain hate groups, but can also result from random effects or from Google generally attributing a great “authority” to the watcher’s sites (as it e.g. does to wikipedia).

Still, we could observe some tendencies: groups with a tradition of being publicly problematised (like Ku Klux Klan groups) are harder to find than less traditional hate groups and groups that might not openly communicate hate in a straightforward way, e.g. by advocating for traditional family values, while in fact attacking LGBT+ rights.

The examples of the two groups classified as the most optimised in SPLC’s list show that some of the indicators of the SEO effect tool might not always produce reliable optimisation levels. The two radio station websites were classified as ‘most probably optimized’, only because they are news sites, which are likely to be optimised. It’s an effect of the tool’s training data. Not all new sites are well optimised for Google, especially lay pseudo-news sites installed for spreading ideologies might not have professional or technical support.

Project 2: Optimised Politics

This (sub)project investigates how political actors have used search engine optimisation (SEO) to increase their visibility in search engine results. This study addresses this issue from the actors' perspective using website optimisation, taking Italy as a case study. In particular, we are interested in examining how political actors use this service to promote their online visibility and, at the same time, to increase the circulation of their ideas and agendas around hot topics.

We provide an initial answer to our question(s) by examining the optimisation level of Google results produced by queries related to five social and political issues: abortion, euthanasia, labour, LGBTQ+ and migration. We found a clear prevalence of non-political actors (97.96%) over political ones (2%). Of the political actors, about 33% are left-wing parties/actors and 64% right-wing. Our analysis reveals that political websites are more optimised than non-political websites (e.g. media sites, institutions and NGOs).

Among non-political websites, we found a predominance of NGOs for abortion and euthanasia issues, e.g. the 'Luca Coscioni Association', which has played a leading role in the legal and political battle in favour of euthanasia. Furthermore, the websites of right-wing politicians and parties are more optimised than those of left-wing political actors. Finally, looking at the level of commitment, right-wing websites have implemented more 'expensive' modules than left-wing ones in terms of skilled human resources and maintenance costs. Despite its limitations, we believe that our analysis contributes to the study of optimisation services implemented by political actors, presenting initial results and an innovative approach.

8. Conclusions

Project 1: Optimised for Hate?

Search engines take numerous variables into account when answering user queries and ranking query results. To further chances of being visible in search results, search engine optimisation (SEO), which has the purpose to identify and abide by ‘best practice’ of internet search, has become a common endeavour across websites (Lewandowski, D., Sünkler, S., and Yagci, N., 2021). In this project we asked whether people interested in hate groups and hateful ideologies (as identified by the SPLC) are likely to encounter websites of hate groups in search results as well as whether watchdog websites warning against the groups would show up when searching for a specific group.

To do this, we took a list of hate groups identified by the non-profit Southern Law Poverty Center. To gather insights on their ranking and level of SEO (if any), each of them were queried in Google (and Bing) and subsequently crawled for signals of search engine optimisation. For this, we used the ‘The SEO-effect’ software developed by Sebastian Sünkler, University of Applied Sciences Hamburg (Lewandowski, D., Sünkler, S., Schultheiß, S. and Häußler, H., 2022).

Using the data from the US watchdogs, we found optimised results, but in most cases no professional optimisation. However, some ideologies, such as anti-LGBT, Christian Identity or Neo-Confederate, are more optimised than others. This result suggests a new configuration of hate ideologies that differs from the historically more prevalent ideologies of hate in SPLC’s 2000 hate group mapping, where neo-confederate, Ku Klux Klan, neo-nazi and white supremacy were by far the most prevalent (SPLC Hate Map, n.d.). This new configuration applies SEO and can be found more easily via Google. Interestingly, this research suggests that those groups that are prevalent for a long time are operating ‘off the grid’ (e.g. on Stormfront) with no website and perhaps exclusion from search results.

Finally, comparing Google and Bing, this last one is more fruitful in returning hate websites, while Google returns more watchdogs. This difference may be significant in identifying the profile of SEO users and their use by hate groups and their supporters.

Project 2: Optimised Politics

We investigated how Italian political actors used search engine optimisation to increase their visibility in search engine results. We found that political websites are more optimised than non-political websites; of these, about 33% are left-wing parties/actors, and 64% are right-wing.

Looking at the effort level, right-wing websites implemented more 'expensive' modules than left-wing ones regarding skilled human resources and maintenance costs. Furthermore, on abortion and euthanasia, we noticed that there is a predominance of NGOs, e.g. the 'Luca Coscioni Association', which has played a leading role in the legal and political battle in favour of euthanasia.

Despite its limitations, we believe that our analysis contributes to the study of optimisation services implemented by political actors, presenting initial results and an innovative approach.

9. References

Cui, M., & Hu, S. (2011, September). Search engine optimisation research for website promotion. In 2011 International Conference of Information Technology, Computer Engineering and Management Sciences (Vol. 4, pp. 100-103). IEEE.

Lewandowski, D., Sünkler, S., and Yagci, N. (2021). The influence of search engine optimisation on Google's results: A multi-dimensional approach for detecting SEO. In 13th ACM Web Science Conference 2021 (WebSci '21). Association for Computing Machinery, New York, NY, USA, 12–20. https://doi.org/10.1145/3447535.3462479

Lewandowski, D., Sünkler, S., Schultheiß, S. and Häußler, H. (2022). GitHub repository, seoeffekt_EDU. Available at: https://github.com/searchstudies/seoeffekt_edu/. Accessed July 27th 2022

Rogers, R. (2018). Aestheticizing Google critique: A 20-year retrospective. Big Data & Society, 5(1), 2053951718768626.

Search Studies research group (2021). The effect of search engine optimisation on the search results of web search engines (SEO Effect). Available at: https://searchstudies.org/research/seo-effekt/. Accessed July 27th 2022

SLPC Hate Map, n.d. Hate Map. Available at: https://www.splcenter.org/hate-map. Accessed July 27th 2022

SPLC Southern Powerty Law Center (n.d.). Official website. Available at: https://www.splcenter.org/. Accessed July 27th 2022

Zilincan, J. (2015, September). Search engine optimisation. In CBU International Conference Proceedings (Vol. 3, pp. 506-510).

Topic attachments
I Attachment Action Size Date Who Comment
Areagraph_03_Tavola disegno 1.jpgjpg Areagraph_03_Tavola disegno 1.jpg manage 302 K 21 Oct 2019 - 13:36 EmilieDeKeulenaar  
Atlantis_WikiTimeline_Tavola disegno 1.jpgjpg Atlantis_WikiTimeline_Tavola disegno 1.jpg manage 86 K 21 Oct 2019 - 13:28 EmilieDeKeulenaar  
Crusade_WikiTimeline-02.jpgjpg Crusade_WikiTimeline-02.jpg manage 70 K 21 Oct 2019 - 13:27 EmilieDeKeulenaar  
Screenshot 2019-07-22 at 15.22.51.pngpng Screenshot 2019-07-22 at 15.22.51.png manage 429 K 21 Oct 2019 - 13:20 EmilieDeKeulenaar  
Screenshot 2019-07-22 at 16.42.17.pngpng Screenshot 2019-07-22 at 16.42.17.png manage 527 K 21 Oct 2019 - 13:37 EmilieDeKeulenaar  
Screenshot 2019-07-23 at 12.25.46.pngpng Screenshot 2019-07-23 at 12.25.46.png manage 60 K 21 Oct 2019 - 13:24 EmilieDeKeulenaar  
Screenshot 2019-07-23 at 16.10.01.pngpng Screenshot 2019-07-23 at 16.10.01.png manage 327 K 21 Oct 2019 - 13:31 EmilieDeKeulenaar  
WW2_WikiTimeline-03.jpgjpg WW2_WikiTimeline-03.jpg manage 66 K 21 Oct 2019 - 13:28 EmilieDeKeulenaar  
cluster 2.pngpng cluster 2.png manage 1 MB 21 Oct 2019 - 13:44 EmilieDeKeulenaar  
image-wall-e3b55f6d8e296e95f13bd18fc943dd55.pngpng image-wall-e3b55f6d8e296e95f13bd18fc943dd55.png manage 934 K 21 Oct 2019 - 13:33 EmilieDeKeulenaar  
pasted image 0.pngpng pasted image 0.png manage 1 MB 21 Oct 2019 - 13:23 EmilieDeKeulenaar  
pasted image 2.pngpng pasted image 2.png manage 1 MB 21 Oct 2019 - 13:32 EmilieDeKeulenaar  
unnamed-2.pngpng unnamed-2.png manage 12 K 21 Oct 2019 - 13:34 EmilieDeKeulenaar  
unnamed-3.pngpng unnamed-3.png manage 11 K 21 Oct 2019 - 13:34 EmilieDeKeulenaar  
unnamed-4.pngpng unnamed-4.png manage 54 K 21 Oct 2019 - 13:37 EmilieDeKeulenaar  
Topic revision: r2 - 16 Aug 2022, LisaMereteKristensen
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback