<!-- .slide: data-background-image="https://i.imgur.com/qcsF5qq.jpg"; data-background-size="100% auto"; -->
## <span class="censor">Data Subjects 2021 <br>// Session 08: The subject of surveillance capitalism
</span>
<img src="https://i.imgur.com/bacaj11.png" width="45%">
<span style="color:white">Pablo Velasco // Information Studies // pablov.me </span>
---
## Many surveillances...
* Sur-veillance: watching-over
* Disciplinary society (Foucault 1975)
* Panoptic surveillance (Elmer 2003)
* few observing many ("**unequal** gaze")
* **internalized** regulatory gaze
* Sousveillance (**inverted** surveillance) <!-- real time tracking of the police in london protests-->
* Panspectric veillance (De Landa 1991)
* enabled by **sensors and signals**, instead of human senses
----
<img src="https://i.imgur.com/G9JkPAv.png" width="60%">
<img src="https://i.imgur.com/EvPx1A4.png" width="35%">
<small>https://www.dr.dk/nyheder/indland/regeringen-ser-flere-overvaagningskameraer-som-vejen-til-mere-tryghed-kvinder</small>
---
## Dataveillance
* Digital Surveillance:
* **Bigger** scope of data: CCTV, RFID, biometics...
* "Liquid surveillance" (Lyon & Bauman 2013)
* fluid, uncontained, **pervasive** monitoring
* *Dataveillance* (van Dijck 2014)
* continous tracking of (meta)data for **unstated** purposes
"The **digital surveillance economy** is that combination of institutions, institutional relationships and processes which **enables corporations to exploit data arising from the monitoring of people's electronic behaviour** and on which consumer marketing corporations ara rapidly becoming dependent" (Clarke 2019, 62)
----
The organisers of the campaign are a *principal* who tries to pursue political goals by **incentivizing and instructing agents to create and share messages** congruent with the pricipal's goals. The purpose of **astroturfing** is to **reach and persuade** as many *regular users* as possible. (Keller et al 2020)
* another kind of human labor
* modification of behaviour as enterprise
----
* blurred distinction between state and corporations
* market oriented for the most part
* displacing the *real* users/consumers for users/consumer *profiles*
<span class="censor">Behavioural knowledge, manipulation, and prediction: microtargeting as **a combination of politics and consumer marketing**, and the capacity of manipulation of affective states (O'Neil 2017)</span>
---
## Digital Surveillance economy
<img src="https://i.imgur.com/Dm1kgwD.png" width="65%">
<small>Key processes of the digital surveillance economy* (Clarke 2019)</small>
----
## Digital persona
Digitalisation of individuals
<span class="censor">"a process almost entirely based on the machine-readable-data-assembly that is thought to relate to the applicant (Clarke 2019, 61)</span>
The consolidation process (of a digital persona)is performed through different identifiers:
* loginid
* stable IP-address
* **browser-fingerprint**
* GPS coordinates
(Clarke 2019, 65)
----
### Profiling from metadata
1. Go to Panopticlick: https://panopticlick.eff.org/
2. Click on "Show full results for fingerprinting" after the test is done
3. Is your browser fingerprint unique? What does this mean?
----
## Profiling from 'learned' models
<a href="https://mobile.twitter.com/JacquesTD/status/1308777336574291971" target="_BLANK"><img src="https://i.imgur.com/J7IT3Zo.png" width="50%"></a>
----
* [Twitters response](https://blog.twitter.com/official/en_us/topics/product/2020/transparency-image-cropping.html) and [Rationale (saliency)](https://blog.twitter.com/engineering/en_us/topics/infrastructure/2018/Smart-Auto-Cropping-of-Images.html)
**taken from**
Theis et al (2018)
**which uses**
Simonyan and Zisserman (2015)
**which is based on**
Deng's et al (2009) Imagenet,
**which inherits categories from**
Wordnet (Miller / Fellbaum 1995-1998)
**and uses images from internet queries**
----
<img src="https://i.imgur.com/FdFwmnj.png" width="70%">
<small>Deng et al 2009, 251</small>
<small>Another example of data used to create common models comes from the Enron emails dataset (recommended hearing for this: https://99percentinvisible.org/episode/youve-got-enron-mail/)</small>
----
<!-- .slide: data-background-image="https://i.imgur.com/q2p2x3z.jpg" data-background-size="100% auto" -->
<span style="color:black;font-size:2em">Recommendation/Prediction systems</span>
<span>[The Napoleon Dynamite problem](https://www.nytimes.com/2008/11/23/magazine/23Netflix-t.html?pagewanted=all)</span>
----
## Why do we keep using systems that track us?
* hedonic benefits
* functional benefits
* social benefits
* psychological benefits
---
## Surveillance Capitalism (SC) surplus
* "Machine intelligence" -> analysis, Machine Learning, prediction, recognition, modeling...
<span class="censor">Every action a user performs is considered a signal to ne analyzed and fed back into the system (Hal Varian)</span>
Google's reinvesment cycle turns from "improving the product" towards "prediction of user behavour" <!-- no silicon valley enterprise knows how to make money -->
* Instead of 'fordist' production, Google UPI (user profile information) extraction and analysis tailored to advertisment (*ends*)
* Unique mix (*means*): behavioural surplus, data science, material infrastructure, computational power, algorithmic systems, and automated platforms ("machine intelligence")
----
<span class="censor">"The term 'exploitation' is subject to interpretations both positive - as in the extraction of advantage from one's own assets - and negative - as in the unfair utilisation of something belonging to someone else" (Clarke 2019)</span>
<img src="https://i.imgur.com/rE092ZL.png" width="70%">
----
## SC not a technology, but a *logic*
<!--INHERENT-->
<a href="https://www.facebook.com/TechAtFacebook/videos/1146186389155473/" target="_BLANK"><img src="https://i.imgur.com/sgAOgFw.png" width="70%"></a>
"You are in **control** but the system is exactly **inferring** the right thing for you to control"
---
## Task: survey me
**Find information about me (Pablo)**
- by any means (legal)
- try to have clarity on the type of collection (overt, covert, public domain, etc) and actors involved (companies, instutions, platforms, etc)
- keep your findings private (as in, not in a public folder, for example). You will show them on Friday
**However**, you will also be looking for who is tracking you while you track me.
* For this, install the [Thunderbird-Lightbeam plugin](https://chrome.google.com/webstore/detail/lightbeamchrome/hjkajeglckopdkbggdiajobpilgccgnj?hl=en-GB) for Chrome. This tool will log trackers, and visualise them as a network. You can also download a json file for further analysis.
* This means that for this exercise, you will have to use Chrome and deactivate your ad-blockers, whenever you are surveying me.
<!--what is the role of consent?-->
<style>
.reveal{
font-family:mono;
font-size: 25px;
}
.reveal .censor{
background:black;
color:white;
}
.reveal .censorw{
background:white;
color:black;
}
.reveal section img {
border:none;
box-shadow: none;
}
.reveal section left{
width:50%;
}
</style>
{"metaMigratedAt":"2023-06-15T21:37:07.243Z","metaMigratedFrom":"YAML","title":"DSub21S08","breaks":true,"slideOptions":"{\"theme\":\"white\",\"transition\":\"slide\"}","contributors":"[{\"id\":\"088a33aa-785b-401b-a225-d782cd214529\",\"add\":13425,\"del\":6072}]"}