Professional Documents
Culture Documents
Elizabeth Englezos
Introduction
Our world comprises of both material and immaterial parts – the body
within the material part, while data and ideas in the other. As humans
“being” we exist in a material form within the physical space. Within that
events that influence our identity, who we are and whom we become. As
material beings, humans cannot occupy the immaterial space in-person and
it is within this immaterial space that the digital space exists. As physical
1
The published version of this article is available at in the Journal of Information Ethics
(2020), Vol. 29, Iss. 2, pp 46-69.
digital space. Presence within the digital space requires the substitution of
entity. A digital substitute is constructed out of this data and it is via this
becomes the actor, leaving the individual one-step removed from the forum
between this substitute and other third parties occur and continue without
the further involvement the individual. The following section begins with a
material and physical person and its relevance to the digital space. Followed
as a means through which the material person can delimit external access to
their person and personal materials, the power imbalance between material
that material consent becomes irrelevant within the digital space. The
article concludes that digital space requires a specialized legal and ethical
If we are to get to the root cause of law’s failure to address the issues
digital and immaterial sign demands closer attention. The material person
something else (Chandler, 505). Without delving too deeply into semiotic
(Saussure, 2013 [98]). For Saussure then, the sign was the combination of
the idea of an object (rather than the object itself) and the word used to
some time clarifying Saussure’s intent before moving any further forward.
“sister” and the French sequence of sounds s-ö-r” (Saussure, 2013 [100]). It
combinations with each other and other letters, that allows combinations of
signs. Peirce adopted a triadic model for signification (see Fig 1). The
Representamen
Object Interpretant
For Peirce, the object referred to the object of signification, or that which is
interpretant is not the party that interprets the sign, but the impression
Take, for example, the representamen “ə”. Some will see this symbol and
an error.
informs our understanding of the object. For Peirce, the meaning of a sign
perspective not only explains why the law of consent and privacy have
In the digital space, personal data3 coalesces to form a data entity. The data
entity is a formless repository that contains all data related to the material
From here, algorithms select and draw from available and readable data in
material person – not as they exist in the material space but as product of
on. Once substituted, representamen can interact with third parties in place
material person.
of the Latin con (together) and sentire (feel or think) and indicates an
persons with control over their data and its use (Englezos 2019). Today,
bilateral, material persons may withdraw their consent at any time. In this
agreement.
government blogs, news sites and so on. “Material consent” refers to the
similar, however, this will not always be the case. “Outcome” is used to
This article builds on the concepts of Peirce’s triadic model to explain the
builds on figure 1 by showing the division between the digital and physical
Representamen
PHYSICAL SPACE
digit
al –
phys
ical
divid
e
Interpretant
Object
Interpretant
60%
30%
* representamen x
10%
interpretant x
algorithm x 1
DIGITAL SPACE
70%
30%
* representamen y
interpretant y
algorithm y 1
25%
50%
* representamen z
25%
algorithm z
interpretant z
1
interpretantz
2
object interpretanty
2
interpretantx
2
PHYSICAL SPACE
In this diagram we see the production of three representamen from one data
the same object (or, in this case, physical person) according to the
does not identify the data selected from the data entity. The algorithm
those provided by data brokers (Û) and sensor data (r)) while still giving
exclusively on the data available from data brokers (Û) and a smaller
selected personal data () and sensor data (r) while prioritizing government
11
object as they exist in physical space albeit based heavily (if not solely) on a
digital representamen.
However, each of these interpretants will have some impact on physical and
This article does not claim that all forms of signification are unwanted or
space through consent and privacy reforms. Consent fails because the
physical person gives consent to share their data, yet has no input or
The digital space is not easily defined. It seems infinite and ethereal and
physical space – yet data’s value comes from its application outside of that
12
are also aware that the replication or repetition of data occurs without loss
the physical person will not know when such replication or access takes
place.
grasp. Nonetheless, we know that humans and other material beings cannot
exist in the digital space. Nothing material can. Presence in the digital space
requires translation of the material into a digital code or format in line with
increasingly occur in digital space and exclude the human form. In short,
algorithmic means lacks the sophistication, nuance and detail of the physical
person, yet we know that important decisions that influence the future of
13
Before the first fracture of individual identity, data is stripped of context and
uploaded into digital space. Without context, the data loses nuance and
persons cannot know which data is selected or whether this data (or its
person and their incorporeal essence are not accurately captured by data
Alain Supiot refers to personality as “the generic concept in which body and
soul are held together (Supiot 2017).” However, translation into digital
space requires a separation of the two. Legal personality refers to our right
14
they see fit: as their ideal selves, aspirational selves, or possible selves (Belk
appearance.
Our data entity expands and expands as we upload more and more data.
has far more utility within the digital space. Algorithms are designed to
select the most relevant characteristics from the data entity according to a
signifier may depend more heavily on credit rating (Citron and Pasquale
2014), health (Peppet 2014), social media interactions and our address. The
Once de-linked from the physical person, the interpretant becomes the actor.
15
these interactions. Instead, the physical person becomes the object affected
by the actions of the interpretant as the subject. Subjectivity and bias within
for each data entity (Grafanaki 2017, Peacock 2014). While the increased
availability and variety of data sources may improve the potential quality of
the data entity, the accuracy of any interpretant remains dependent on the
algorithms used. The data broker industry generates as much as $200 billion
Sadowski notes that the term “mining” implies the data already exists. In
reality, the data and signifiers produced through data mining are not mined
16
person develops. Data brokers begin with raw data that is processed to
data is stored, accessed or used makes physical persons easy targets for
“data mining”. The data produced provide additional input into the physical
This article does not suggest that there are no benefits to algorithms or
day (Marr 2018). This data would be useless without and effective means to
attention.
17
group according to the data point (or points) used (such as gender, marital
also selects which data the algorithm will target. However, a preference for
means that a higher proportion of the preferred data is considered than data
have already touched on data selection as the first point of potential bias
within algorithmic models. The second point of potential bias occurs with
theories to properly appreciate why this subjective value has the potential to
18
occurrence of one event in correlation with another (Sarwar et al. n.d.). This
of a particular event (Dowe et al. 2013). This prior relies on one party’s
Any bias on the part of the determining party will be imported into the
19
mitigation will again depend on the algorithm’s design – and therefore the
for the material person in digital space. Algorithms rely on the presumption
that what we have done before we will do again. Such presumptions create
important issues in the digital space for two reasons. Most significantly, this
yet allows limited (if any) opportunity for review. Unlike real life where we
are free to break with tradition, algorithms also prescribe what we see and
20
therefore, rendered invisible to us. Thus, in digital space, we are more likely
yet these interpretants inform choices made on the physical person’s behalf
in the digital space. These digital choices have real world outcomes for the
new ethics to address this reality. When algorithms turn our search results,
clicks and interactions with technology “into a conversable code,” this code
is stored for later “interrogation and analysis” (Harper 2017). Raw data is
algorithms lack transparency (Pasquale 2015, Ziewitz 2016), yet may affect
21
“analyses each one and spotlights the top candidates.”5 Only those
persons can be excluded from contention without knowing they have been
short-list the available data into relevant and manageable subsets. Some
consider some data more probative than others and priorities that data set.
the potential for cumulative bias grows with each inclusion or exclusion.
threat lies not in predictions or advertisements that “miss the mark” but in
22
Data without context is also likely to conflate the temporal with the static. A
would not usually visit, but as this behavior increases, we see more of this
content – even though our feelings may be transitory. The effect of these
clicks becomes more fixed with each click or scroll that affirms the
permanent part of the material person’s data entity, regardless of the limited
In his 2011 work, “The Filer Bubble: What the Internet is hiding from you”,
Pariser notes three key aspects of the filter bubble: isolation, invisibility,
available for consideration based on their data entity – “data entity A.”
23
future.
If we imagine the digital space and its infinite webs of data and information
outer spheres, within them are topics or beliefs and poles of opinion (pro- or
A similar hierarchy exists within each topic. The more popular posts, clicks,
24
Group 1) will exist somewhere along the continuum between pro- and anti-
Pariser also notes that human beings tend to choose the “least
we engage with the least objectionable version. The quantity of clicks might
be the same as that of someone who genuinely agrees with the content.
quality engagement. The quantity of clicks will not represent the true nature
of the material person. Probability will infer that the group prefers Pro-X
the continuum towards Pro-X and see less and less anti Belief X (“Anti-
feedback loop that rewards behaviors that conform to the Group 1’s norms.
Our tendency to engage with the least objectionable can unknowingly trap
to content we are not actively seeking (Sulkowski and Picciolini 2018). Our
25
affects the way material persons are seen by other material persons within
Algorithms will look for parallels between our interpretants and link parties
that might not meet otherwise (Mantelero 2016). By grouping like with like,
2019, Harper 2017). These groups or “aggregates” lack the “social and
models (Tene and Polonetsky 2013). Minority opinions are less visible
because algorithms assume that group behaviors are static and unlikely to
change (Tene and Polonetsky 2013). The material person’s world view is,
may not accurately represent them. The result is not only an increase in
autonomous subject which selects the content they receive (Ibrahim 2017).
26
this aspect which a new data and information ethics must address.
Figure 3 shows the link between signification, its outcomes in both digital
become and our perception by others. Laws built on the legal person as
something singular and material cannot address this translation. Law that
the next section, this article explains why the consent is irrelevant to the
What is consent
The previous section showed that data about the object, algorithms, the
outcomes in the physical space. The examples showed that these outcomes
have concrete consequences that may not always benefit the material
person. Next we must consider whether this occurs with the consent of the
material person.
27
between parties. An absence of consent can ground legal actions under tort
or criminal law (Brosnan and Flynn 2017). Essentially, consent allows the
physical person to delimit what others can do to, for and against them.
Consent is an ideal that developed from its early origins around the 12th
valid consent is given freely7 and occurs once the material person has
adequate notice, knowledge and choice as to how (and whether) they will be
affected. However, only the signifier has the power to act within the digital
consent each signifier would need adequate notice, knowledge and choice.
28
comes from a long line of supporters of political party X. They are Pro-X in
their belief, but following a news report on radio, they are interested to
staunch Pro-X voter but begins to receive more posts on this particular
no middle ground and that PP-A can either be Pro-X or Anti-X. This
inquiry. Each search along these lines pushes them further into the outer
margins of Group 1 and further along the Belief X continuum from Pro-X
PP-A’s posts within that group become less visible to other Pro-X signifiers,
but increasingly visible to Anti-X. PP-A then receives more and more Anti-
29
within those groups will receive less acknowledgement from Pro-X group
members but increasing support from Anti-X members. The example, while
more Pro-X content. Occasionally, those with more extreme Pro-X beliefs
than those of PP-A will share content that engages PP-A. As with our
previous example, this engagement – this time with Pro-X sentiment – will
push PP-A along the belief X continuum further towards the Pro-X end. As
this shift occurs, the Opposition view that initially drew MP-A’s attention,
At its most basic level, the interpretant has become the actor on which
digital decisions are made. The interpretant does not review the Terms and
Conditions (“T&Cs”), nor does it click “I agree”. The physical person gives
consent in the physical space. Consent presumes that the physical person’s
consent is relevant to subsequent acts within the digital space. However, the
30
The question then becomes whether the physical person has adequate
legal entry into a relationship that allows the buyer to take their property,
and in exchange, the buyer accepts entry into a legal relationship that allows
the seller to keep money, data or property that otherwise belongs to them.
Both parties concede their right to claim back their property without legal
31
review and understand the T&Cs. Given that the signifier has no
opportunity to review the T&Cs, we shall proceed on the basis that consent
presumes that physical persons have consented to the use of their data in
to the access of personal data and subsequent data use in exchange for a free
service or discount.
Free social media platforms and digital services are not “free” (Hoofnagle
and Whittington 2014; Hull 2015). Physical persons pay through non-
pecuniary performance at a data and privacy cost. The initial draft of the
certain aspects concerning contracts for the supply of digital content (“EU
32
incentivized the exchange of digital content and services for data (EU DCD
The Directive as adopted earlier this year includes contracts where the
services, Art. 3, 5(f)). Perhaps the reader recognizes these terms from
privacy notices they have previously skimmed? They are not uncommon
however, that material persons are unaware of the true cost of these
33
privacy creates a boundary between the values of the physical person and
their community. Ideally, the physical person only agrees to share their data
if willing to concede to a loss in privacy over that data. The result is a kind
Privacy is also relative in that it exists only in reference to the person from
whom we wish to hide (Hand 2018). The constant connectivity of the digital
interact via a network of screens they exist as part of a circuit. They are
their social network. However, once these expressions have entered digital
space, the physical person cannot prevent their access by other unforeseen
parties. In order to give valid consent, physical person must understand the
34
The quality of these digital disclosures directly affects the quality of notice
the individual receives. However, the digital space poses a two-fold threat to
individual notice. First, studies estimate that less than 10% of users who
access End User License Agreements (“EULAs”) spend more than two
minutes reviewing their contents (Bakos et al. 2014). Given the length of
sufficient notice for the physical person to acquire the requisite knowledge
for valid consent. In fact, studies suggest that physical persons are not fully
produce longer disclosure documents filled with more and more detail (Ben-
standard form disclosures have become “so long and elaborate that
overwhelmed with disclosure documents and those who wish to read the
35
to be virtually impossible.
Our third requirement for valid consent is choice. This is also problematic in
the digital space given the consenting party may be unaware that consent
has taken place (Obar and Oeldorf-Hirsch 2018). The option to click “I
buying mood” and carries little gravity (Obar and Oeldorf-Hirsch 2018).
browse wraps such as “I agree” also exploit our bias towards agreement by
Google and Facebook. In its final report, published June 2019, the
36
one of the most significant and recent developments in the law of consent as
and, therefore, provides a new de facto global standard for data regulation
As we have seen, the digital space and physical space could not be more
material and ephemeral. The new inclusions and new rights contained
consent in and beyond the digital space because its focus remains on making
the consent of the physical world fit the digital space. As outlined above, no
matter what our approach – be it through the consent of the physical person
Conclusion
This article provides a new lens through which we can understand the
the physical person just as the actions of the physical person continue to
37
media platforms and other digital content providers will continue to shape
and inform our individual realities. In reality, we have limited control over
our data (if any) and limited control over who has access.
childish with our children. In the digital space, we have no such control
different personas (Belk 2013). The physical space and digital space have
perceive each other in the physical world as well as in digital space. Within
outside of the digital space. The object, the representamen and the
For now, the first step is to recognize that signification renders consent unfit
38
autonomy. The most successful approach may be one that recognizes the
automated decisions are made about you (GDPR Art 22). Oversight of
This can only work if there are limits on the creation of unique signifiers –
Declaration
1
Signified and signifier are the commonly used English translations of signifié and
signifiant from the first published edition of Cours de linguistique générale
2
adapted from Chandler, 836.
3
I have specifically adopted the term “personal data” in this article in keeping with the
European Union’s General Data Protection Regulation, Art 4 (“GDPR”).
4
Although GDPR OJ L 119/1 Art, 17 and other similar legislation allow the natural person
to request the erasure of their data, the data can never be entirely erased from the digital
space.
39
hire, for any size company, in every industry which states in relation to employers who post
job advertisements on Ziprecruiter.com “Once you post, our matching technology actively
scam millions of resumes and profiles on ZipRecruiter and 100+ job boards and invites
qualified applicants to apply. As applicants come into your dashboard, we analyse each one
and spotlight the top candidates to make sure you never miss a great match.” from 0:21.
6
Whether this right is respected and observed is a matter for separate discussion.
7
See, for example, Commercial Bank of Australia v Amadio (1983) 151 CLR 447, Louth v
Diprose (1992) 155 CLR 621 and Blomeley v Ryan (1956) 99 CLR 362 in contract law and
Reference List
Alaimo, C., & Kallinikos, J. (2017). Computing the everyday: Social media
(1992).
Amer, K., & Noujaim, J. (2019). The Great Hack. United States of
America: Netflix.
40
https://doi.org/10.1086/674424
Barron, L. (2011). Living with the Virtual: Baudrillard, Integral Reality, and
Press.
41
information, and law. New York University Law Review, 76(1), 23–
113.
https://machinelearningmastery.com/classification-versus-regression-
in-machine-learning/.
Condliff, M. K., Madigan, D., Lewis, D. D., & Posse, C. (1999). Bayesian
42
3(3), 398–415.
Zealand.
2015/COD0287 (COD).
processing of personal data and on the free movement of such data, and
43
https://doi.org/10.3145/epi.2018.mar.12
27(4), 803.
Harper, T. (2017). The big data public and its problems: Big data and the
19(9), 1424–1439.
learning-basics-with-the-k-nearest-neighbors-algorithm-6a6e71d01761
44
Hoofnagle, C. J., & Whittington, J. (2014). Free: Accounting for the costs of
the internet’s most popular price. UCLA Law Review, 61(3), 606–670.
Ibrahim, Y. (2017). Coalescing the mirror and the screen: consuming the
19(2), 131.
Lucy, N., Hartley 1948, J., Thwaites, T., Colebrook, C., Tofts, D., Briggs
45
https://www.forbes.com/sites/bernardmarr/2018/03/05/heres-why-data-
Nikolov, D., Lalmas, M., Flammini, A., & Menczer, F. (2019). Quantifying
Pariser, E. (2011). The filter bubble: what the Internet is hiding from you.
Pasquale, F. (2015). The Black Box Society: The Secret Algorithms that
Peacock, S. E. (2014). How web tracking changes user agency in the age of
Big Data: The used user. Big Data & Society, 1(2).
46
Press.
Sarwar, B. M., Karypis, G., Konstan, J. A., & Riedl, J. (n.d.). Item-based
Wiley.
Sulkowski, M. L., & Picciolini, C. (2018). The Path into and out of Violent
Psychologists.
47
Tene, O., & Polonetsky, J. (2013). Big data for all: privacy and user control
Wen, Y. F., Wong, H. M., Lin, R., Yin, G., & McGrath, C. (2015). Inter-
48