at the “facial machines” seminar in aarhus, denmark for two days. i’ll be tracking some of the presentations in this thread. 
denmark has had a face covering ban for only a few months. there are some practical exceptions: if it's really cold, if it's halloween, or if you are wearing the extremely badass traditional dress of people of fanø island.
off to a great start with some notes about physiognomy and the history of measuring faces. if this is the intro i’m psyched for the speakers 😳
just learned about "snapchat dysmorphia", when people start to feel more connected to their digitally filtered portrait than their reflection in the mirror (and sometimes get surgery to match) vox.com/science-and-he…
first up is @jilltxt who gives this example of a self portrait in a convex mirror from 1523 as an early example of representing the self as reflected in technology
@jilltxt introduces @katie_warfield who describes the outstretched arm of a selfie as "embracing the viewer" makingselfiesmakingself.com
@jilltxt introduces @alicetiara who describes social media as being the thing that trains society to accept quantification the same way that schools were designed to train effective factory workers yalebooks.yale.edu/book/978030020…
@jilltxt has an open-access book that covers these references and more, check it out palgrave.com/gp/book/978113…
next up is @adamhrv who introduces us to a chinese toilet paper dispenser with a face recognition system (to keep people from taking too much toilet paper)
adam shares the idea of doddington's zoo, a system for talking about when people are hard and easy to recognize, and those that are easy to imitate or those who can imitate others easily
as an example, in the context of the openface system github.com/cmusatyalab/op… jon is a sheep relative to anne and eva because he is "well-behaved", easy to segment from the others
adam is talking about how many possible images exist at small resolutions, and how it is possible to do high accuracy face recognition or activities recognition even at these low resolutions
adam has been working on mapping the connections between published research and the companies and governments funding them. first he played a video about surveillance in china washingtonpost.com/news/global-op…
then he digs into @SenseTime_AI sensetime.com the company behind the tech, and some of their published research arxiv.org/pdf/1704.03373…
adam shares an incredible project from @pengberlin where they blend photos of immigrants with photos of government officials to create fake passports pen.gg/campaign/mask-…
next up is @zachblas who starts with a discussion of his process of immigrating into the UK and having his biometrics collected, focusing on the word “bio-exempt” as a piece of “administrative poetry”
zach is building up some language, connecting to philosophies of surveillance from deleuze, foucault.. he starts with "surveillance and capture: two models of privacy" by philip agre tandfonline.com/doi/abs/10.108…
zach asks: what does foucault's panopticon look like on our own bodies? one answer is his "face cages" (2013-2016) work in collaboration with three other artists, where they 3d printed the triangulated points of any tracker. definite inspiration from torture devices.
zach presents research from 2012 where subjects were asked to determine sexual orientation from a brief glimpse of a photo. he says it's significant and representative in that it treats a face as something that can be extracted/separated from a person journals.plos.org/plosone/articl…
zach gives a shout out to research from @wewatchwatchers including the book "dark matters" which "locates the conditions of blackness as a key site through which surveillance is practiced, narrated, and resisted" dukeupress.edu/dark-matters
zach has a huge range of projects using masks, covering everything from border violence to "gay face", and references the way that the NYPD was arresting masked protestors in 2011 during occupy wall street (enforcing a law from 1845) gothamist.com/2011/09/19/nyp…
zach is wrapping up with édouard glissant's essay "for opacity", asks how to bring this concept into a world of automation? maybe: "a practice of anti-standardization", of anti-biometrics shifter-magazine.com/wp-content/upl…
it's incredible how deeply embedded zach is in the cultural and philosophical dialogues connected to his work. i wish i could situate my work even half as clearly as him. if you want more info, this talk looks similar/related:
next up is @lila_morrison who has been investigating the eigenface technique for facial recognition
@lila_morrison eigenfaces were the first successful technique for face recognition, and helped convince people that automated recognition was something worth funding. this image from at&t labs was her introduction to "how machines see us".
@lila_morrison lila jumps into the earliest research on automated face recognition from w.w. bledsoe (FOIA'd by my former student @JustinLange). bledsoe used double exposures to explain pixel-based similarity archive.org/details/firstf…
lila digs into the details of eigenface recognition: the euclidean distance from an example to a target "is a differential recognition". connects this idea to "differential interventions: images as operative tools" academia.edu/6153012/Differ…
lila says that statistics is "how the state sees" and connects the combination of statistics and vision that is present with face recognition to francis galton's composite portraits. both technically and aesthetically. galton.org/composite.htm
(you might know galton because he coined "eugenics" and inspired the nazis). wittgenstein also made a composite portrait of himself + his three sisters. he was less interested in the "center" of the image, and more interested in the periphery: the variety present in the unity.
that's all for today, i'll be back with more tomorrow!
i just finished my talk. next up is daniela agostinho, a postdoc at university of copenhagen. i loaned daniela my laptop, so these notes might be more sparse.
daniela argues that refusing surveillance is not an option because it is a prerequisite for many kinds of social interaction. and resisting surveillance is only possible within the predefined affordances of surveillance infrastructure (and are increasingly limited).
daniela quotes @zachblas to elaborate on this
some artists and activists create or argue for camouflage and masking in the face of surveillance and internet culture, but does this simply shift the responsibility (and blame for failure) from the system to the user?
daniela is showing “lineage for a multiple-monitor workstation” by sondra perry vimeo.com/131805970
daniela explains: even though sondra’s family is wearing a chromakey green mask, they will still be read the same way. masks don’t necessarily open up the potential for defining appearance to everyone equally. (i might not have the right language here)
another interesting comment in q&a: what are the differences and intersections between interpersonal vulnerability and vulnerability to algorithmic surveillance?
next up is @kristoffer_orum who is taking a more performative approach to his presentation. he started by explaining that he updated his OS last night and openframeworks broke. but he salvaged some pieces to share with us!
he suggests that he's terrible at keeping track of time, so he installed an app to get track of time. but the trial expired so he uses a keygen. and the keygen starts covering his screen with ads...
he's asking some questions about what it means to see faces separated at a distance. while he speaks, a face-swapped version of himself is interrogating mark zuckerberg on the screen behind him.
"i didn't know i was a shower cap [...] it seems like neural nets are a machine more misunderstanding the world just as much as for understanding it"
he has a camera pointed at the crowd doing expression recognition. every time it detects that we are "too bored" it plays 15 seconds of a trance build up with motivating slogans and fire in the background.
kristoffer is holding a microscope camera to himself and talking about how it manages to create a distance between himself and his presentation. under a microscope he doesn't read as "caucasian", but you do see sweat drops reflecting his anxiety about the presentation.
kristoffer is talking about his interest in noses as being a part of the face that is not typically used for identification. that it's the first thing to become distorted when someone creates a caricature. but it's also associated with the difference/the other.
kristoffer just wrapped up. before the next talk, i wanted to add: @jilltxt tweeted a few notes from my talk if you're interested in what i spoke about
next up is share lab labs.rs/en/ with @TheCreaturesLab and mr. andrej petrovski. like all good presentations they start with the panopticon and a discussion of tarkovsky's "stalker".
one of their first projects was making maps of how data travels through the web labs.rs/en/as/ to help understand the status of cyberwarfare in serbia
they argue that if you look at the way facebook profits off our behavior, there is no way for anything like "algorithmic transparency" to exist within that infrastructure labs.rs/en/quantified-…
they start everything we know about facebook (including all the public endpoints, all the patents) to try and understand what happens internally. this map looks like a data visualization, but don't be fooled: in fact it's a narrative.
they compare these kind of maps to ancient maps of the world where "america is undiscovered, and africa looks funny". they argue that even inside facebook no one know what this map actually looks like. and in some ways, it's changing every second.
now they are presenting their most recent work, "anatomy of an ai system" with @katecrawford where they build a narrative around & teardown all the tech & infrastructure around amazon echo anatomyof.ai
someone asks about the aesthetics of their maps: "well, people tend to believe more if it looks complex and black and white. if it's in color people think 'maybe this is just some nice data visualization from a studio in new york'"
this seminar is wild. someone is asking about the freudian interpretations of this recurring idea of "digging into the void" and "uncovering the hidden black box". everyone giggles a little.
that's all, @ShareDataLab just wrapped up. the organizers close out with a nice quote from foucault. thanks everyone for following along 🙏
