Almost from its very beginning, the Web has been ambivalent. It has facilitated freedom for information, but this also included the freedom to spread misinformation. It has faciliated intelligent personalization, but at the cost of intrusion into our private lifes. It has included more people than any other system before, but at the risk of exploiting them. The Web is full of such ambivalences and the usage of artificial intelligences threatens to further amplify these ambivalences. To further the good and to contain the negative consequences, we need a research agenda studying and engineering the Web, as well as numerous activities by societies at large. In this talk, I will present and discuss a joint effort by an interdisciplinary team of Web Scientists to prepare and pursue such an agenda.

1. Steffen Staab 1Institute for Web Science and Technologies · University of Koblenz-Landau, Germany
Web and Internet Science Group · ECS · University of Southampton, UK &
Web Futures
Inclusive, Intelligent, Sustainable
This is really about Web Science
Steffen Staab
@ststaab
https://de.slideshare.net/steffenstaab

2. Steffen Staab 2
• Why wasn‘t the Semantic Web working?
(Staab et al 2019)
• Why did the Web work?
– Why did some Web encyclopedias flourish and others not?
– Why did some social networks grow and others stall?
– Why do open source projects grow and why do companies
put money into it?
• What could threaten the Web?
– Privacy (then!)
– Monopoly/Oligopoly (internet.org)
The origins of Web Science in 2004

3. Steffen Staab 3
What is Web Science?
(Staab, 2013)Co-constitution of Technology and Society

4. Steffen Staab 4
Less than the union, more than the intersection
Web Science Butterfly – How do you see the Web?
not loved by anyone and still the best depiction

5. Steffen Staab 5
Manifesto (Berendt et al) in preparation with most
significant contributions by
• Bettina Berendt, Fabien Gandon, Susan Halford,
Katharina Kinder-Kurlanda, Eirini Ntoutsi
Dagstuhl Seminar:
Web Science 10 years – closing the Loop
Dagstuhl picture
here

6. Steffen Staab 6
What is Web Science?
(Staab, 2013)Co-constitution of Technology and Society

7. Steffen Staab 7
Social determinism
Society defines technology and its effects.
Example:
EU politics defines what technology
„is able to do“ –
upload filters in copyright law
private? commercial?
Tower in Paris or Las Vegas?
Permanent nightly lightning vs light show?
France? Germany? USA?
Science and Technology Studies: Co-constitution
Von Benh LIEU SONG - own work, CC BY-SA 3.0,
https://commons.wikimedia.org/w/index.php?curid=6926930

8. Steffen Staab 8
Technological determinism
Technology alone determines
usage
Example about privacy intrusion
by technology:
"If you have something that you
don't want anyone to know, maybe
you shouldn't be doing it in the first
place.“ – ERIC SCHMIDT:
gawker.com
Science and Technology Studies: Co-constitution
By Hecker / MSC -
https://www.securityconference.de/mediathek/munich-
security-conference-2018/image/eric-schmidt/filter/image/,
CC BY 3.0 de,
https://commons.wikimedia.org/w/index.php?curid=69416818

9. Steffen Staab 10
Co-constitution
Technology and society determine each other
Example:
• Technology: SMS
• Usage:   hdgdl lol rofl
• Derived technologies: Twitter, Whatsapp
• Usage: #
• Derived technologies: Instagram, Snapchat
• Usage:
Politics,
Influencing as
business model
Science and Technology Studies: Co-constitution

10. Steffen Staab 11
The Web as a boundary object
• „data“
• „algorithms“
• „artificial intelligence“
• „gold standard“
• „communities“
• „survey“
• „data protection“
By Illustrator unknown - From Charles Maurice Stebbins & Mary H. Coolidge, Golden Treasury Readers: Primer, American Book Co.
(New York), p. 89., Public Domain, https://commons.wikimedia.org/w/index.php?curid=4581171

11. Steffen Staab 12
Ambivalences of the Web
Utopia vs Dystopia

12. Steffen Staab 13
• Germany:
„Freie Fahrt für freie Bürger“ – no speed limit
• USA:
„The right to carry weapons“ – self-defense
• UK Brexit:
„Take back control“ – no delegation of power
Society not good dealing with ambivalences
Technical solutions exist to improve social welfare –
but they are of no use, if people disagree

13. Steffen Staab 14
Information
Freedom
Information
Quality
Ambivalences of the Web
Personalization Privacy
Influence by
the Masses
Manipulation
of the
Masses
Inclusive
and Fair
Exploitative

14. Steffen Staab 15
Latest example
• „CNN Refuses to Run Trump Campaign's Biden Ad“
because of factual incorrectness
• Facebook accepts same video as part of its ad
business
• Fact checking as a way out?
– IFCN
• >100 fact checking
organizations
https://www.poynter.org/ifcn
• Facebook partners
• (Brandtzaeg and Følstad 2017)
Ambivalence:
Information Freedom vs Information Quality
Fact
checking
(Mis-)Trusting
the Fact
Checkers

15. Steffen Staab 16
RumourEval 2019 (Gorrell et al 2019)
Veracity: False
Stance: Comment
Stance: Comment
Stance: Deny
Stance: Deny
Subtask A:
• classify each comment as
support, deny, query, or
comment towards the
statement in the post
Subtask B:
• classify the statement
expressed in the thread’s
source post as true, false, or
unverified
Labelled collection of comment
threads from Twitter and Reddit

16. Steffen Staab 17
CLEARumor (Baris et al 2019)
• Pre-trained ELMo embeddings
• CNN-based model with auxiliary
input features
• Second place in Subtask B
(Veracity)
Idea: User‘s stance towards a claim
could be a clue for debunking the claim in
early stage.

17. Steffen Staab 18
• Data sets: size, balance
• Real-time
• Astroturfing campaigns
• No journalistic knowledge on fact checking:
just canned knowledge
⇒ Community oriented approach with FactCheckingNI
Failures of current misinfo detection

18. Steffen Staab 19
• No obvious solution
• No stopping condition
– Campaigns: Trolls / electronic armies / astroturfing
• Involving many stakeholders with diverging interest
• Needs technology, behavior, education, policy,
media/journalism
Misinformation is a Wicked Web Science Problem
An AI filter can be a
building stone, but not
the eventual solution
to deal with
ambivalence of
filtering
Mor Namaan 2019

19. Steffen Staab 20

20. Steffen Staab 21
Influence by the Masses: Online Participation
Democracy!
Everyone may participate!!
CD

21. Steffen Staab 22
Future of Online Participation?
https://de.wikipedia.org/wiki/Boris_Johnson#/media/Datei:Johnson_(48791303991)_(cropped).jpg
Deal or No Deal?

22. Steffen Staab 23
Exclusion:
• Not everyone has internet
• (functional) analphabets
Self censorship:
• legasthenic?
• Not reveal themselves
Gender discrimination
• Offline: gender quota for
presenters at Green Party
• Online: females trolled
more than males
Online Participation: Always great?

23. Steffen Staab 24
Exclusion:
• Not everyone has internet
• (functional) analphabets
Self censorship:
• legasthenic?
• Not reveal themselves
Gender discrimination
• Offline: gender quote for
presenters at Green Party
• Online: females trolled
more than males
Online Participation: Always great?
Renate Künast, Member of Bundestag, Green Party
Olaf Kosinsky - own work, CC BY-SA 3.0 de,
https://commons.wikimedia.org/w/index.php?curid=62640921
Attacked with
speech so abusive
that I cannot repeat it here.
Considered o.k. by court,
because she is politician

24. Steffen Staab 25
Exclusion:
• Not everyone has internet
• (functional) analphabets
Self censorship:
• legasthenic?
• Not reveal themselves
Gender discrimination
• Offline: gender quote for
presenters at Green Party
• Online: females trolled
more than males
Communication limited to
elites?
Online Participation: Always great?
Influence by
the Masses
Manipulation
of the
Masses

25. Steffen Staab 26
Typology of Online Participation
Main target
group in our
project
Not all
participation
is positive

26. Steffen Staab 27
• Mobilization
Party members become more active through online
• Replacement
Members replace traditional activity through online
• Reinforcement
Active members reinforce their participation relatively
• No Use
No change of behavior
Online participation model of Thuermer
Gefion Thuermer PhD thesis „The effect of
the introduction of online participation
processes in the Green Party Germany“

27. Steffen Staab 28
Survey Measurements
(in the social science sense)
Activity Tool Use
Activity
Increase
Mobilisation - + +
Reinforcement + + +
Replacement x + -
Non-Use x - x
Activity relates to institutional activity (various factors)
Tool Use - use of various tools;
Activity Increase compares actual against intended change of behavior
- indicates either no difference or negative correlation;
+ indicates a difference or positive correlation;
x indicates either (as it does not make a difference to the model).
(Thuermer 2018)

28. Steffen Staab 29
Several waves of data collections
(interviews & surveys)
(Thuermer 2019)

29. Steffen Staab 30
“I think the surveys were very accessible, so that everyone,
including old people, could participate. It stated clearly to
‚now click this link in the next line.’”
“Online participation is one way to improve inclusion. For
example, we have older people who are less mobile who
could participate through this route.”
„Our assemblies always happen at children’s bedtime. It sounds
trivial, but this highly specifically excludes parents. For polls,
discussions and so on, online participation would be really
great.”
“[Online Participation] is really good because it allows easy
access independent of people’s life and circumstances. For
example, shift workers who work at night and can then go
online and participate when they have the time.”
(Thuermer et al. WebSci 2018)

30. Steffen Staab 31
Heterogenity of effects
(Thuermer 2019)
variable
value
tools
direction of effect

31. Steffen Staab 32
• Online participation generates data
• Data impact reality:
– Ex.: Online EU survey on whether to keep time change
• Which data do you want to have/?
– Democracy requires representativity
• Representative for party members (more college degrees)
• Representative for voters (more elderly)
• Representative for the people für (who represents kids?)
Online participation makes data become an actor
in the sense of actor-network theory
My optimistic point of view:
Many questions about „Correctness of
data“ are put forward, because they can
be put forward now!
Research task:
De-biasing as task of
Computer Science, ...
and society

32. Steffen Staab 33
Another actor...

33. Steffen Staab 34
Data:
- Linux Kernel Mailing List
2014
- LK Github Repository
2014-2015
Questions:
• Which shares have hobbyists vs firm-sponsored
developers?
• Is there a trend over the years?
• Are firm-sponsored developers more effective?
What motivates companies to Open Source?
Inclusive
and Fair
Exploitative
(Homscheid et al. 2016)

34. Steffen Staab 35

35. Steffen Staab 36
Sponsorship an important moderator
• Structural capital: degree (in+out)
• Relational capital: tie strength / weighted degree
• Cognitive: involvement on different mailing lists
(Homscheid 2019)

36. Steffen Staab 37
Sustainability
A Network Science perspective on

37. Steffen Staab 38
Preferential Attachment over Time
Prediction by Bianconi-Barabasi
Impossible to catch up? (Sun et al 18)

38. Steffen Staab 39
...and in reality: Time invariance
system time = network size ≠ real time
Well possible having a popular paper / Web site,
even if you come later

39. Steffen Staab 40
• Preferential attachment with decay of relevance 𝑅(𝑡 − 𝑡𝑖)
– Fewer citations as the paper gets older
– Π𝑖 ~ 𝜂𝑖 𝑘𝑖 𝑅(𝑡 − 𝑡𝑖)
• Exponential growth of the network size
– Holds empirically for publications!
– 𝑠 = 𝑒 𝛼𝑡
• Power-law degree growth of nodes with regard to their
ages.
– Holds empirically (see previous slide)
Core idea
age
fitnessdegree

40. Steffen Staab 41
fitness
Network growth: exponential or constant
Growth
𝒈~𝒆 𝝈𝒕
𝜎 ∈ ℝ
When (exponential)
growth can not be
uphold, some
assumptions will
have to give in
Academic reward
Economics
Etc.

41. Steffen Staab 42
Experiment

42. Steffen Staab 43
India: Net neutrality extended
• No internet.org (by Facebook)
• Online-Shops must not sell their own stock
Inclusive and fair to achieve sustainability
• Also for the latecomers
Sustainability

43. Steffen Staab 44
What is the Future of the Web?

44. Steffen Staab 45
• Documents
• Data
• Services
• Things
• People
• Assemblies of items
Toward a Web of Everything
(Berendt et al. 2019)

45. Steffen Staab 46
Somewhat intelligent, maybe
• biased
• unreliable
• unavailable
• unsafe
• unsecure
• unaccountable
Intelligences hosted on the Web
(Berendt et al. 2019)
Failure of the socio-technical system

46. Steffen Staab 47
Hybrid societies of intelligence

47. Steffen Staab 48
• Web → AI
– ImageNet
• AI → Web
– Chatbots
– Virtual
Assistants
• Learning
from Web dialogues
• Safeguarding
Web and Artificial Intelligence

48. Steffen Staab 49
Conclusion
Longer

49. Steffen Staab 50
• Digital observation of everyday routines
• How data is created depends on
– technology, communities, ownership, markets, regulations,
rights, ...
• Just to describe these processes
and what they mean, we need
– Computer scientists, lawyers, political scientists,
sociologists, ...
Sociotechnical challenges: Datafication
Computer science tends to
underestimate the descriptive dimension
(Staab, Halford, Hall 2019)

50. Steffen Staab 51
• 3 billion people without internet
• no normative propositions
„The Web is `good‘ for everyone“
• Digital divide:
Internet benefitting the privileged –
rarely than the impoverished?
• Digital literacy: Understanding the Web as a system
– Everyone must know how a meme grows to avoid
misinformation!
– Not the case now!
Soziotechnical challenges: Digital Divide
(Staab, Halford, Hall 2019)

51. Steffen Staab 52
• Google, Facebook, Amazon,...
• 245,000 US$ to train XLNet
https://syncedreview.com/2019/06/27/the-staggering-cost-of-training-sota-ai-models/
• Who decides the bias?
– Representation of women on Wikipedia
• Edit-a-thons
• GNU for AI?
– (https://voice.mozilla.org/)
Who owns AI that steers the Web?

52. Steffen Staab 53
• Where are we now? What works well? And doesn‘t?
For whom, when and why?
• What are the possible futures for specific AI applications?
• What would have to happen to get us there?
• Diversifying the vision of the common good
– „good for someone else“
• Empowering participation in the future
• Bringing people back in - not as users or consumers, or
in terms of impact
but as part of the world we are building
Susan Halford (Exauguration speech 2019)
Democratizing futures: An utopy

53. Steffen Staab 54
Susan Halford (Exauguration speech 2019)

54. Steffen Staab 55
Ipek Baris, Lukas Schmelzeisen, Steffen Staab (2019). CLEARumor at SemEval-2019 Task 7:
ConvoLving ELMo Against Rumors.
Bettina Berendt, Fabien Gandon, Susan Halford, Jim Hendler, Katharina Kinder-Kurlanda, Eirini
Ntoutsi, Steffen Staab. 10 Years of Web Science — Dagstuhl Manifesto, Manifesto from
Dagstuhl Perspectives Workshop 18262, in preparation 2019.
Petter Bae Brandtzaeg and Asbjørn Følstad (2017). Trust and distrust in online fact-checking services.
Communications of the ACM, 60(9), 65-71. doi: 10.1145/3122803.
Gorrell et al (2019). RumourEval 2019: Determining Rumour Veracity and Support for Rumours.
Dirk Homscheid, Mario Schaarschmidt, Steffen Staab: Firm-Sponsored Developers in Open Source
Software Projects: a Social Capital Perspective. ECIS 2016: Research-in-Progress Paper 12
Dirk Homscheid. Firm-Sponsored Developers in Open Source Software Projects: A Social Capital
Perspective. PhD thesis submitted at Universität Koblenz-Landau, 2019.
Steffen Staab, Susan Halford, Wendy Hall: Web science in Europe: beyond boundaries. Commun. ACM
62(4): 74 (2019)
Jun Sun, Steffen Staab, Fariba Karimi, Decay of Relevance in Exponentially Growing Networks. In:
Proc. of ACM WebSci ‘18. ACM.
Jun Sun, Matus Medoy, Steffen Staab. Time-invariant degree growth in preferential attachment network
models. Submitted 2019.
Gefion Thuermer „The effect of the introduction of online participation processes in the Green Party
Germany“, PhD thesis, University of Southampton 2019.
Thuermer, G., Roth, S., O'Hara, K., & Staab, S. Everybody thinks online participation is great – for
somebody else: A qualitative and quantitative analysis of perceptions and expectations of online
participation in the Green Party Germany. In Proc. of ACM WebSci 2018, pp. 287-296.
References

55. Steffen Staab 56Institute for Web Science and Technologies · University of Koblenz-Landau, Germany
Web and Internet Science Group · ECS · University of Southampton, UK &
Thanks to my many, many collaborators!
Ipek Baris, Lukas Schmelzeisen,
Jun Sun, Dirk Homscheid, Mario
Schaarschmidt (U Koblenz)
Gefion Thuermer (U of Southampton)
and all the others!