FIRST WORKSHOP ON AGONISTIC ALGORITHMS
===============================================================

Participants: Janet Vertesi, Carl DiSalvo, Nick Feamster, Seda Gurses
Date: 28./29. March 2016
Princeton


faking, manipulating, jamming other than activism, ...
challenging etc. being inherently good

i had not seen the thing that bots immitating "hispanics"


j: 
there is a scale and visibility issue
organizations have been doing this politically for a long time
faking the other persons followers
isn't this part of how democracy works
wouldn't this be part of agonism


c: 
the good citizen
variety of civic experience based on the simpsonsthis idea of the informed citizen is a fairly new construct
he is interested in the citizenship in the us
and citizenship has always been this fairly laxidaisical participation
the bowling alone: we are not doing these things together
actually, we never did these things together
i actually this is how democracy and diplomacy
there always seems to be some underlying
what way do you think that this is not politics as usual
the connection to advertising
since the 2nd half of the 20th century
the politician as a brand and using brand style strategies
reputation management as a major component of the political work involved
that is brand and marketing work

all the people i know that make bots
we have a number of bot makers out of our program
literature/media experience
they all do it for advertising
how do you drive traffic to articles on the various platforms

janet:
angele christin
chartbeat and slate
what happens when you have these metrics programs
the clicking and sharing become a metric of success or not
there is not an implicit notion that followers are bettter
they need to be made and they need to exist in particular context
the metricism, and the fetishization of metrics
to understand the kinds of games they are playing when they are botifying
you want things to be forwarded, but you don't want to flood
there must be different mechanisms
trying to understand why metrics matter and to whom?
[*** MAKE VISIBLE THE POLITICS OF METRICS]

why did he say that hillary and trump have 50 percent
and bernie  10 percent
meaning bernie is worse at it, less so than 
put some political assumptions

c:
the documentation of occupy sandy
the way they moved money
amazon wedding registry
you could make wedding lists
but it was for recovering response 
ad hoc stuff on top of existing systems
that conflates the message
you can't use these systems for bottom up response because they are top down systems
at the end, one of the statements
DHS says
what does it mean to be the state
if our citizens our doing such great job, maybe they don't need us
we can outsource it our citizens
this is called the resilient network
it complicates the usual arguments of you are either on this or that side

the bernie argument is interesting
and those distinctions are the ones that we are trying to trouble

j:
america has all of these frontier narratives
some dude goes to the frontier and survive against all odds
canada, we have wilderness narratives
you go out and die
nobody makes it: they are eaten by the people they are with
best canadian folk song: whaler ship
you are iced in, and the song is about how you are not going to come back in
there would never be a resilience thing: the government takes care of you
you don't leave without
we would call it a public failure
if people had to use amazon
[*** RESILIENCE AS A PROJECT TO RESPONSIBILIZE CITIZENS, ONLINE USERS?]

c:
we had this amazing discussion
that would not be the natural conclusion
frank was like you are living the end of times
that is the absence of staTE
and that is true
i do think that that is one of the different discussions that one has to have about politics in the us
even the state is weak

j:
i love how the weakness of the state becomes the rational for the continuing weakness off the state

c:
the weakness of the state makes us innovative
that is the smart city discourse

there is a political theory discussion here, too
why is agonism more popular in europe
and why do they look so horrible in the us
in europe there are forums for public debate
but there are still rules around that
we have foregone all rules in political discourse

in the current election cycle
you have discussions about the size of trump's hands and genitals being a part fo political discussion in a debate
trump is acting as a troll
he is an amazing troll
and in the past week, he has bated people, ted cruz about these affairs
what happens

affect theory and politics
contemporary us politics: the problem is not that it is driven by affect but by emotion
instead of passion, people were becoming hateful
the problem is that affect somehow exists without a narrative
literary reference: a poetic response to something
emotion: is something you write a narrative around it
and the narrative around is around hatred
if you take that as your starting point
whether they are bots, iot things that print out hate objects

j:
the politics of spectacle is really different
there is a lot of that going on
and none of it is about actual issues
the political divides are not about the issues
but they become changed by spectacle
when you have this move towards the affect of anger and hatred and fear
you need to disentangle that from mouffe talking about it agonistic democracy
it is about the contestation of ideas
agonism is supposed to stop short of violence

c:
the myth thing
challenge with the agonistic discussion
it refuses to say that someone ever wins
the idea that somehow you win and it is over is a f low
there was a women: bonny honig
bonny honig - roe v. wade
the problem that people had that when roe v wade had passed the issue was settled
when years later people try to overturn
that is one of the fatal flaws
we had the fight, we passed the law, it is settled
that is not the case
it might have been settled for the moment
but it will always be open to some party coming and questioning it
the minute such things cannot be questioned is when you have an authoritarian regime


c:
series called occupied
the difference between us and scandinavian narrative
house of cards is brutal: it is mean, violent
in occupied: norway stops to produce oil because of its affect on climate change
the eu and russian gets into cahoots
russia occupies norway and forces them to open up their oil production
it is interesting to see the way politics get acted out
what does it mean for norwegians to resist when they don't want to go to law
it is the opposite of the us narrative
unlike the characters in house of cards
they are able to brutalize their way through
the norwegian way is to negotiate and he fails but that is the way to do things


j:
the natual conclusions is about glossing over what particular actions mean
the consensus based stuff
i studied a consensus team
it is really hard to do 
it has a lot of merit
commonwealth runs as a consensus organization
that is a counter balance to the un
which doesn't
it is not that consensus is really bad
you cannot assume consensus 



went over the slides

j:
dichotomies
flattens things
pluralism where things are contested
it seems like have multidirectional things is interesting
before jumping into the space
what happens if we stop thinking about good and bad guys
what happens if you think about it as a complicated situation
and you have to situate yourself and the actors in relation to each other
does this get to agonism?

c:
networks are good
even if you were taking the dichotomous view
let's assume these are our bots
each bot is driven by something else
bots don't spontaneously produce themselves
what if all those other htings that make the bots possible
are these coming from marketing or from whom
i have a feeling that these overlap (the people that make the bots and their networks of production)
the trump and hillary bot are all messed up in the middle
so series of institutions that are uniquely hillary and trump but in the middle are these institutions that are common



j:
this is helpful for the discussion with nick
how it can be operationalized differently then the diad: which is antagonism
that is the thing that nick is trying to get away from
he was saying that in the security community they are trying to get away from the bot that is antagonistic
looking at agonism looks at these multivalent relationship

where affect comes in:
in the assumption of us vs. them
that is one affective move

c:
one of the things when you take the agonism seriously
is that there is a shared commitment
that there is a disagreement and a shared commitment
that it is worthwhile to disagree about
part of the work of doing an analysis of an agonistic condition
is to understand what that shared commitment actually is to

it could be things like global free trade
for trump and hillary it is america
even though they disagree

seda:
very charged situations like belgium or turkey or black women


j:
agonism allows you to zoom out of antagonism
it may look like middle eastern immigrants against the whole world
but if you look at it, there are different political parties and neighborhoods
you don't take into account all teh things that they share
all the actors they share
some of it is historical
some may be coming from have nots
when you just make it seem like the terrorist or immigrants vs. belgian government
you loose the larger associations
you can really easily get tunnel visioned
the affect maybe creates it
so it is really easy if you are the target of a twitter bot attackers
but really it is 4chan and a whole series of infrastructures
that people may have a recourse to 
what is horrible about the horrible twitter attacks
is that it feels like this antagonistic situation


c:
is it to far gone or can you stil have agonism?
i think agonism, but how do you create a sitaution where the conflicting dialogue plays out 
and can create conversation
the microsoft bot revealed the ways in which
i think it was gamed
but you could say
imagine that there was a bot that was created with interactions with 4chan
you could have different bots going at it,
similar to the way you would have oreilly and jon stewart go at each other
two characters who did the work of summaring the perspective of others
ampliyfing and turning them into cartoons
as providing a moment that we could share
that is partially the interesting bits of work
how do we create these bots
that let us see
the drumpf bot
that is to me a worthwhile thing
it captures trump's rhetoric and plays it non-stop


j:
it is also the result of a media personality, 

c:
what is missing about that
is that all the candidates have equally cartoonish characters

j:
i was talking w danah about tay
what is interesting about it is that it makes several things visible
not the drumpf bot but if you had a 4chan bot
what makes 4chan powerful is that it is in a particular corner of the internet
most people do not have to lok at 4chan
but if 4chan comes after you
you don't know where they come from
the threats that you received from them are not done publicly
it amplifies the antagonism but does not let you see the agonism

c:
what is a value to nick as a computer scientist?
i can imagine, if you had a twitter bot that was trying to 
talk about climate change
coalesce arguments for or against climate change
that becomes an interesting contribution in a way in which cs and platforms can contribute to public discourse
you can think about that as an automated polling
we have had these discussions about computationally generated journalism
here is a news story about a little league game based on tweets
and you can do something like that for attitudes about climate change
and condensing those to wholes that can be examined
that is an interesting use
what are the ways that people disagree
who does this twitter bot fold in and reject, who follows and retweets it
then that network reveals itself, and i bet it reveals itself in ways that are strange


j:
in the antagonistic view
it would be bot A against bot B
what if you took out pieces of the infrastructure that makes A possible
what if you treated the different bots as such
and studied them even handedly


c:
if we believe that politics is made up of different associations that are in conflict with one another
that they also have interesting connections
pro-environment
we should save the environment: scientifically proven
and others who think it is gods gift to us
by releasing twitter bots to computation prspectieves
twitter bots bring people to them
could you reveal networks that would otherwise not be seen?
can the bots become a social science intervention of research tool that shows us the shape of political associations?


j:
two party system view
trudeau is like a bot!!!!
things on the internet meet in other spaces that right and left
all kinds of networks that are intersecting and sharing or not sharing diffrent kinds of things


c:
the other two things you can say about this is it could potentailly allow you to build networks that you would otherwise not be able to build
there would have tobe a network analysis part of this
to understand who all this is 
there is a paper by 
they made some twitterbots
that ended up revealing a whole network of people who were engaged in a conversation about energy production in northern england
some were for and against and they identified by who followed
maybe they could have done this by then by hand
and go through followers
alex woki???
gave a talk at 4s
building twitter bots helped them understand the network they are working in

j:
that is all the work that sam is going by trying to find these people
if you think of it as infrastructurally
if you are a cs person interesteed in security
what if you saw attackers as larger networks
and if you could protect against the source

c:
or if you are a political agent of any kind
activist or political agent
who may be people who are reaching out to me

can bots express or stregthen networks and associations

c:
i think the iot thing is more pressing and interesting
you see things at defcon or ccc
i don't think it is th same as doing something playful
that seems too much like a demo
the printing of the racist pamphlets
is the first threat of iot instantiated in real life


c:
seda asked the question of how playful we want to be
we were again at th qeustione of iot
phil has this book on iot
the majority of the book is about twitter bots
because there are not as many iot examples
can we look at what we learn now from bots
and when all the appliances becomes platforms
where each of the actors is a different kind of thing
printere, refrigerator, light bulb

j:
here is the interesting thing
isp, money, where you live
some of them share those background assumptions
so the lamps become this perspicous things: the visible objects
and the thing under is where all the politics happens

c:
there is a diagram that i want to show
there is a diagram on platform studies
the different levels at which the humanities look at systems
we can look that and say
do we want to say these are the levels that we want to look at
or across levels

j:
the actor netowrk theory people will say there are no levels
the politically charged person would say that that is independent what is technical or not

c:
most media studies happens at the level of reception
some of it has been about interface
but nobody has looked at function, code, and platform

j:
that is why tarleton's work is so innovative

s:
the computer scientists are at the bottom

c:
but they are not reading their objects cultural artifacts

j: layer 8, the human layer
techopedia
physical
data link
network 
transport 
session
presentation
application
and layer 8 is politics outside of the system that will come in and bug it up


s: nick would agree
he might be talking with people who think layer 8 is a distinct layer
we have all of these IoT objects
they are being supported by their interconnection
but we have to show that they are not distinct and part of the game


c:
they are hybrid systems where the human is very much in the loop
like getting trump to read mussolini
can we get someone to retweet mussolini
they made a bot that tweeted the tweets at trump 
it ran for a while and he finally retweeted one of them without know it is a mussolini quote
it was a human in the loop endeavor
there is a lot of human in the loop

seda:
algorithms cannot be all open
because people would game the system
but then kate has this example of the reddit algorithm
showing the intestines of algorithms

encore...


c:
a student is looking at data intermediaries
open data program
three parties involved
an advocate that wants to affect some change
a person that is able to find and get the data (the data intermediaries)
the data intermediaries are not committed, to the political action that is happening here, they are committed to openness of data
they will work to collect data so that somebody can use it adn they are not committed to the cause
she was looking at the pro-democracy movement in china
who was collecting data, who was making it actionable and who was using it
they were different groups
and they also asked her to burn their irb forms
they were concerned about the people she interviewed
they have to sign the forms, but afterwards they have to burn it

j:
don't people game the algorithms regardless of what they do or not
this is why i don't have an adblocker
because that is the only layer of transparency i have into what a system knows about me
to see if soething is traced or recorded
so people will play with what is given to them even if it is propriary


we have some ideas also about what each algorithms are and what they do
danah was saying people don't know how tay works
and they had ideas about what the obvious mistake would be


c:
with that example in particular
all of my reading on it is so biased
tarleton tweeted something
the four best things written on tay
i want to know what my brother things
and he doesn't know anything about this
it is one thing to sit in that room to understand bots
but if you just stumble about that

j:
my in laws are not on social mediea
the primary is not being fought in their newsfeeds: they don't have any
it is another battleground, with its own rulese
it is not a nw media landescape but another battle field

c:
it is nothing new at all but reputation management that would be handled through a press release
but it would be now through a press release and bots...

thinking about iot
or journalism
what does that look like and how does that get combined 

j:
in the gym you have all the different screens

c:
the orange zone gym
physical activity area
every screen is a members read outs
so you can get competitive with other people
i cannot imagine what this looks like

j:
metricized time, vs. where are you now in time
the orange zone: is about being in the metricized time
instead of being in your own time and knowing what your body needs
i want a watch that builds that for me
to have the moment: i need a workoute today, but it needs to be in a slow way
it is like the home horoscope

c:
home horoscope was a critical reflective project that was about data that was happening in the home
and would print out horoscope predictions for your based on that data

j:
we can also be more evocative
we have a plant in our house it is called the janet craig
it is an office plant that can go without water
we forget to water it when things are busy
the plant has the record of our relationship on it
it is the mood rings
this is when we moved
and it was neglected
we were eaway from xmas

c:
tom jenkins
iot work
his question is what happens when the iot home is not a nuclear family home
which is about ownership of space and objects
what does iot look like here?
a lot of the world does not live the way we live
where my home is only my home
it also breaks down in apartment
apartment complexes
i have a lot of objects that i share with other people
blenders
gardening tools
it would be very different if i had to think about
when my neighbor wants my blender
what does the neighbor know about me

c:
a friend of mine does advertisement work
they have an iot product
a pare of socks
it turns of netflix if you fall asleep
i can't believe that netflix paid you money to produce this as a marketing piece
they paid you to do critical design that is not critical
this is entirely the banal home of the future
that is the kind of ridiculousness
and now netflix knows my sleeping patterns



Nick is now with us!!!



j:
we discussed agonism and the guiding principles
we want to hear what you are doing and interested in


n:
timely that woolley's talk was today
that talk made sense to me
it certainly was familiar
the paper that i had sent to seda ahead of time
what he would call propaganda

backing up
after the talk
i think i understood a little bit
after reading the proposal that we had and hearing his talk
the two fit togethr a little bit more
i had thought about propaganda as a bad thing
this is spam and we need to stamp it up
after his talk, i was thinking
he was much more of the stamp this out kind of thing
my reaction after reading your work there is a more nuanced thing
and this becomes clear in the types of bots he was showing
that was timely
his way of studying this was more along the lines of
when i saw that there is a huge complement to how we study these problems
we were looking at the total twitter stream
ande figure out, 
we were not looking at content
he made a quick comment about retweets
and this not helpful and superficial
but from our analysis of network traffic
retweets are juicy
they are indicative of superficial activity
the way i look at these problems
we stop before we get to the content
we look at the graph at what rates and volumes
very complimentary to what he was saying
to me, if his was talking your language
my thought was, we could probably add a lot to this discussion
we are looking at the complement of things
compromised accounts
there is a whole field in cs 
that detects when an account gets compromised
those are the kinds of things
where the type of thing i work on and have experience in can come into play
twitter or forums
how do we tell if there is an account compromise
rates of posting
and other kinds of traffic patterns
are thsese people related
discovering behavior from traffic patterns is something i am interested in
i don't know how to think about it more
in a more nuanced way
i have been thinking about attack and mitigation
this project is really interesting because we have to come up with something else
i am not sure what it is
i am really interested in your theories
it is fascinating!!!
not going to stamp them out
so, how do we make a participatory dialogue
do we label them
all 10 people are saying the same thing
does it matter that they are human or not?
are day paid half a yuan, or are they sheep?


c:
do we label or coral them
do we understand what these clusters are and how they are interacting
in some cases you may want to amplify them
news media
jon stuart and oreilly have a productive act
amplify roles in a way that gets people engaged
they are clearly in conflict with one another
can you imagine a system like that that is not going to get rid of negative commentary
but think of the intermedieareis that can transform it

does this also tie into iot work?
do these bots that now live on twitter, end up living in our fridges and cars and what is that like as a space to engage?

n:
how do they correspond?

c:
we don't have good examples yet!
pax technica
but there is nothing in the book about iot
he thinks iot will be similar in the way that these objects are automated
another way to approach this project
do we maybe look at lessons learned from twitter
and apply them in iot

n:
where do these things translate to physical actuation
voter turnout
something relating to that


j:
we did some conceptual work
the first thing we wanted to do
the problem with that talk was there were good and bad guys
we are trying to figure out how you are talking about agonism in computational space
a pluralistic version of what the debate looks like
the relationship between 2 people can be antagonistic
and in reality it is more pluralistic
when you see the associations they are sharing that is where thee agonism is

n:
is there sth cs can bar with
to what extent 
presumably it is not healthy for democracies to be operating in dyads

c:
you get that filter bubble
and it is reinforcing how do you burst that and bring in more perspectives
there were two other things

J:
here is a dyad
hillary and trump bot
and that is all you see, because it looks like it is traceable
but they both come out of larger organizations
and they are embedded in a cultural context
and they are both acting because they are using advertisement stuff
there are things they share and don't 
from cs perspective
if one is your attacker and this is you
you are always thinkig of attacks as coming from your attacker but not the bigger picture

c:
all of this is to say this is no different between these and relationship management
the origins are in advertisement
the difference between two beer companies are the companies that produced them
are there things about bots that we can see from the outside

n:
it is interesting what you said about influencers
if this were a twitter graph
you may not have to follow the cirlced, but just the triangles and squares
and the attacks come from the square

c:
part of the question is what are the times you want to amplify and rally things
if you think of contemporary politics is not left and right
ecology and preservation and climate change
protect earth for scientific or religious reasons
can you understand the ways in which such associations are being built
it is much less a concern about trump and hillary
why these associations breaek apart and reform

j:
this is why we need a cs person in the room
one way to think about this is can we use the tools of cs what these actors are
where sam talked about tracing back to find the bot networks


n:
he mentioned sth briefly about tracing money
it was quick
cs has looked at porblems like that in the underground economy
where people looked at spammers and bought the stuff they were paddling and followed the money
in fact you are not going to get spam from 50 million people, instead it is three people


the closest that cs people talk about this is through botnets
in terms of looking at coordinated network attacks

c:
if a lot of twitter things are about capturing attention
the question is whether iot becomes about bandwidth
and energy usage
become an analogy and attention we usually give other media forms?
it is a measurable thing, it is like followers

n:
we can start to look at network traffic as a way of how users are interacting with the deviec
is it used or not
and how is it being used
that is monitorable


j:
when we were etalking about th entiteis
thy are all embededed in networks
here is my iot lamp
they are aell esharieng this social infrastructure
that is differentially distributed across them


n:
in the graph that you are talking about
you are identifying high degree types of nodes
based on say yes that is interesting

is this a device that gets a lot of your attention is maybe an interesting nexus point
twitter: what ever random thing pops up on my phone
iot: my echo starts reading the news out to me


chrome casts (audio version)
dropcam
ubi/echo: voice recognition
smart things


another way to approach: these are the ways we can participate
and what are the ways in which we may not want to participate
and what are the ways that people create a work around


c:
the challenge with iot
there is not good examples
so much it is 
the printers 
that is an example
but we don't have a lot of them

j:
we got better at defining agonism
but the connection to iot

n:
political discourse and online forums
for example
the printer gts m thinking
if i can accses public displays, cars
the other can be a monitoring aspect
if i want to forment political movements
i may want to know things about human movement
measuring attention is important


c:
so, we are part of the metro lab thing
ttri is going to deploy sensor pods
in atlanta
and the big thing they are talking about that they can monitor is use
it could be cars or people
how many people are in this space
and at what times
the sensing is fairly rudimentary
where is the attention being paid
and where should we pay attention to

i don't know if this is too orthogonal
depending on the type of data that is gathered can affects these dialogues
trump: you can imagine the discussions that informed by the data that is collected


c:
if you take the city is your site
they will put them onto intersection: power, wifi
intersecteions can dvocate on their own bhalf

j:
we were talking about affect
if you have a dyad
and focus on it
all of 4chan
goes after black women activists
it looks like a dyad
then an antagonism develops and an emotional response
if traffic really increases along a particular node
then this is where people concentrate their attention
that is the example with the boston potholes
because iphone is used only by a well off segment
and their part of the city get developer better


c:
in the city or online forums
one way to detect the voices that are not included
we know where the traffic is
in which case we know who is not there
here are a series of voices that have not been included
can we bring these other perspectives that are relevant

n:
so like recognizing when you are myopic
recognizing which particular perspectives
like climate
we can say all the major features in the climate debate
and look online and see which of these features are not in ther and bring thm in


j:
that is what i am trying to do with the critical nws bot
you have neglcted where your view is coming from

n:
wee did work on filtre bubblse
diffrnet gographic locations
and present it to the user
and show what was left out of your research

j: you did critical design 


n:
the irony was that because his talk was so good and bad
it created this knee jerk as to why he is making normative statements about these bots
citp has a bot that tweets about our blog
is that good or bad?
that is what made me think about advertisement

c:
that is a good thing
the talk missed the idea that you need the good and the bad to have democracy


j:
imagine you didn't think democracy was a good thing

c:
i have a friend who works at the atlantic that use bots to get attention to their site
i want sb to read my piece
the bot serves an important service

j:
this may be related to internet of things
we can also do surfacing
you think that these two things are acting as agents
if you wrote twitter bot
every time hillary or trump tweets
you show the associations that come along
say you have a map
it could be refugees in europe
or the muslim communities and brussels
or your toast or your fridge
if i think of this layer of bojects
and there are other actors
that are shared differently by this network
the understructure that alows them to even look like they are acting like different bots

say apply and samsung, they don't share things
so you have this other layer of objects and control
one of the things about iot, we are talking about these connecetiones but not this undrlayer

c:
can you trace that?
like your colleagues traced the money
they worked with a black or grey market
if i put a die in my blood to trace things
could you, put a marker on data and follow it through

n:
people have done different things
you can certainly, take the postal system
address is being shared
you can use your middle initial as a trace

j:
tweets and retweets

n:
we can look at fast retweets but not real time

j:
do you know lightbeam
it conceptualizes data and where the data moves

c:
it would be fascinating to take an issue
as expressed by a twittr ebeot and do a visualization of that
and how we understand politics in clusters that in first moment don't make sense
who is rallying around this network
and to see how associations change
maybe you have an organization that is partnering on one topic but n


seda: the  multiple layers of censorship
tech, bot, troll, legal

n:
the idea
is that censorship may not happen through blocking
one of them appears to be 
take down requests
in certain countries, south africe and us
the dmca
scientology has used this in the past
you threaten legal action
so basically looking at 
research question would be like
that crosses over into stuff like
that brendan and others look at
does the wording structure and origination of the take down request affect whether sb pays attention to it or implement it
you can go to isp 

s:
start with twitter and bots
and move to iot


c:
it will not be long before we see more examples of iot things


j:
when the home becomes a consumer product space we bring things from different companies
sb said all their stuff was apple
and it wasn't 
these companies are building these closed world models
there is no way your devices are working that way and nobody uses their devices that way

i visited 16 sites
86 third party sites

c:
you see the same iot argument
the att pitch in atlanta
you should also build things on our network
tell us the things that you want
but we don't want you to use another network
don't use the georgia tech network
use ours
they are looking into what sliceof the pie they can lock in

n:
trump hillary example
i want to figure out 
this appears to come from hillary but it is coming from some other place
i also think of the inverse of that
the complement
i see the square and triangles and who ever talking
they may something like
so and so is against free trade
and that is some issue i really care about
and i follow that particular triangle and not hillary
so i am interesting in figuring out the squares and triangles that represent hillary
even if she is not saying it, it is her

if you think about a candidate
it is not just that person
it is a colection of ideologies
they are pro choice, gun, etc.
in the end
you are kind of working for a collection of ideas
the fact that it comes from hillary's account matters less to me 
then hillary is bought by these people and this person said something
so thinking about it as inside out

j:
that is alligned what we know about politics
with koch brothers and their relationship with the teap party
big banks etc.
what i like about it is that we could surface ideological relationships
but if we could ground them in entities
people, institutions and other bots

c:
that is interestienge to m e
what ar the entities that make the networks of associations that make hillary
and who is she drawing from
being much more open about who the actors are
it is people and it could also be bots
and it is fair to consider them as actors
that is the interesting thing in contemporary discussions about political theory
that anything other than humans do politics

j:
they haven't read lagndon winner

c:
they haven't read anything
if you think of all the networks that make hillary
the agonism is by exploring all of that
they need to have certain things that are common
are technologies and bots part of that


s: 
we can make a game where which political group wins if you have a team of only bots, only humans, or mix...

j:
people can combine their own teams

you want to turn your students into pineapples
the goal of the game is to do that
it is brilliant
it makes the point


n:
q about agonism
can it apply in political discussoin
is it mainly related to politics?

c:
it is mostly about political discussion but has a wide take on what counts as political
there was the agon competition
and it was done as a way to keep everyone on their toes
the notion that
there was never a time when the competition doesn't take place
when we talk about agonism in us politics
we always want to solve something
we want to solve censorship and then it is done
no it is going tobe this constant state
whatever algorithm you create somebody is going to go around it

j:
roe v. wade
we fought hard
and people went home 
activists are surprised 


n:
agonistic tactic:
        is that a tactic to increase agonism?
        or a tactic that is used part of an agonistic process? like astroturfing
        or is it both?

c:
i think both
we need to break out of the dichotomy, the dyad poisition
and that is not enough to say here is a way to burst the bubble
and here is how the dialogue can be made more complex
the first one can be a necessary condition
it is more the second one those, what are series ofe tactics

j:
we might start by saying these are two ways of doing agonistic tactics
and to say there are computation and social science ways of doing it
and these would all qualify as tactics

n:
what are the most active
i can think of a lot of different places where agonism appears
it is all over the places
elections
policy debate
isps and privacy thing
you can super small scale
do you think that you can focus on one issue?
debtceiling
how this plays out in congress
how do you sort of think about that
what is a fruitful place to study right now, given that you can't do them all


c:
i am thinking about three generic scales
municipal local
federal and global issue

i don't know if the tactics would have to be different 
depending on the computational ways we go about this
i am wroking with a group that is combatting gentrification
in a way that may get engaged
one of the advantages of a topic that local
then you can bring some insight to bear on what is happening but it may not necessarily scale
in my ideal world
if there was time 
it would be interesting to look at those three scales
to take a case study with each
we should look at the national election
it is right there
with gentrification
i was able to go through social media and see a bunch of issues


n:
law enforcement, racial profiling
in princeton
maybe there are issues that have a local and national element


c:
it would be good to look at issues that are active
we should not take our time to find the issues
and think about how we can do things computationally
refug issue is more global

gentrification is an issue 


n:
gun laws 

c:
right to discriminate bill
you cannot be forced to have none discriminatory policies in your company
it is called religious freedom

governor of georgia vetoed that today
georgia now has the 3rd largest film industry and they threatened to pack 
the super bowl is supposed to be there
and nfl was like it is not happening

j: that is the triangles and squares

c:
the governor is dealing with pro business and religious republicans
and pro carry, every single president of every university said do not sign this
how will faculty think about responding?

j:
union?

c:
no unions...
gun control would be an interesting one, too
i want to either visualize it or create some bots

j:
hunting is an issue where left and right converge
lefties that go homesteading and hunting
natural birth rights
right and left converge although they are very diefferent commuenitis

c:
homeschooling
he has kids that are learning greek
but he also goes with homeschooling people who say evolution doesn't exist
and they are intermingling around this idea of rejecting the educational system
diametrically opposed in ideologies

j:
it is interesting to see that
we were learning something related to evluation and we were told some americans do not believe in evolution

n:
it relates
different groups can affect the debate
if you can figure out 
you have these nutty guys talking with these other people
now you have film makers
and it is now a different kind of discussion
and how do you make film makers visible
the people making the most noise
there is a dyad

s: vulnerable communities and mapping

c:
it would be interesting to bear knowledge that you have
both with baby economies
with gentrification
they are tracking mold
they want to use the data with respect to mold to advocate for the neighborhood
but they don't want to release the addresese, bcaus thy don't want to city to go after those houses and tear those down 
how do you map the power of a network and hide the individual?
you want to show that there are resources here but you don't want to pinpoint them...


n:
this is a technical discussion about capacity between large isps
net neutrality and all of that
this is an agonistic debate as well
isps are saying: how much cpacity exist between your and my network
no one wants to talk about individual links
but point fingers
so the isps have come to us, and said you can release data in these aggregate forms
we don't want to tell our business relationships or routing 
but we want to talk about our capacity
and how does it slice across region, and all isps in a particular region
it is not like tearing down houses
but you have people or organizations that are trying to contribute to a debate
but exposure of all the data
this is proprietary business data that gets xposd
thr is data that could tilt th discussion in one way or another but how do they release it
i have also seen how data is not released, like under certain auspicees
we don't want to release that
it is proprietary etc.
but it could also cause fingers to point in one way or another

j:
we think about this a lot in ethnography
and that has been really hard fo rme to talk about power relations
and people that are subjugated
and tries to do that in a way that is respectful

n:
this is where the iot crossover happen
does x effect energy usage
well, we have the iot device to measurement
the data in aggregate or something can be very informative about
does x,y,z result in energy reduction

n:
arvind may also be interested in talking about
how can we think about the data that we can collect
what can be learned from this kind of data
and how can data be released ethate can contribute to dialogue
example: i was paranoid working with isps with the release of this data
an easy way to flush it would be you are hiding x,y,z so you must be hiding something
so, how can we get some data released so taht we can protect ourselves against that kind of comment
if you are contriubting to a debate and show that youa re not biasede


j:
i do think that when we try to surface these elements of the network, we are making statemens about who we are protecting and who we are not
x is a politician
there is a way in which we talk about trump and hillary
and i know what it means to have someone in your family be a public political person
it is horrible what happens
so there is still a level of talking about human agents
something went viral, being made visible in that way is really horrible
if you read alice marwick stuff about twitter
it was born out of a specific social context obsessed with status
twitter was built by people who did not have to think twice about what they say in public
mostly white men
these people have never walked down the street thinking they may be assaulted 
there are communities that are expected to beahve in particular ways online
and for others it is not safe


c:
another thing we can do to look at those factors brought about mapping communities
are there strategies that happen in non-computational environments that we can carry over to the computational environment
what are those roles and systems in place
what are the ways in which these sorts of organizations
developed ways of hiding or resisting

n:
anonymous communication tools is maybe where there is a crossover on that
we did this anonymous communication channel
that would allow me to send a message to jannet such that you and c can't see it
we envisioned this as an activist tool
for operatives to exhange information and coordinate protest
or to leak information
one thing i had not thought of
sb. said this kindof thing may be useful in battered women shelter settings
deb was in cape town
these women they want to go in and say what happened to them 
but they need to cover that this conversation even eexiste
how about tor
no, the face to face is important
even if we are not directly talking
supposedly this is very important
coming back to what you said about how do we protect speakers
that is not an agonisic one
anon tools
we commonly think of these as people buying stuff on silk road
but if we are talking about posting in online forums
or having dialogues on twitter
any kind of online discourse
tor doesn't solve those problems
that is definitely an area,people in my area think about that
how do we design systems and communication protocols
with different design goals than tor
maybe there are differnt kinds of sttings
whr peeople need different things
and we can make systems for that

j:
what if we gave some people bot armies
without it being traceable to them
i don't have a recourse against the bot armies

c:
it is interesting to think of other nodes that youcan act upon so that you can have the affect you want to have
the gamergate situation
the best way to stop this is you find out
you look at who is doing the harassment
but you tweet that persons parents
9 out 10 it is a kid in the basement
you find the node that is above that node
jus so you know this is what your child is doing in the basement

n:
zeynep was talking about censorship in turkey
to clamp down on activism
you don't block whatsapp
you go to the parents and tell them they are dating such and such person

garry king and molly ...
censorship in china
it is more effective to let some communication to follow
if they were not formenting revolution
it was like the safety valve on their frustration
even about the goveernmnt
as long as you are not going to certain territory it is fine
and it clearly relates to agonism
how is the government controlling discourse

dan wallach
rice uni
they had a paper that looked at 
microcensorship
basically looking at deleted twets
and how china goes and deletes specific tweets
governments affecting agonism
who is talking what topics


DAY II:


presentations:
design strategy

identify tactics of political organizing and action
saul alinsky
sociologist by training
        ruckus society: pamphlets for greenpeace on how to do mediatic action
        beautiful rising: marketing approach to political discourse
        identifying tactics that are not computational
        a worthwhile starting point: look at the tactics proposed by these organizations and think of their digital counterparts
gentrification:
        properties become social media entities that are producing a voice
        you can look at change in terminology
                e.g., brunch ideas for easter weekend
        look over time how the description of particular activities change
                ??? are they geotagged? no, but tweets either have the name of the property or they are an address
                
step back:        
        do we do this?
        gun debate works better
        think of possible easy responses that are scripted
                homes as media entities
                personalized responses
                hashtag pla: #whatmatters?
                
                
national case:
        getting trump to tweet mussolini
        why don't all candidates have a bot?
                have a debate between all the bots!
                
common idea:
        flipping the story
        making the invisible visible
        davidson: inverting and showing something over and over again (but also shows the limit of a quick and clever move)
        
congress edits:
        anonymous edits to wikipedia form the congress
                this idea of the smoking gun
                how do we show that these processes are happening all the ime
                
        a bot that tweets to comcast about their internet speed
        it gets back to the discussion about IoT
        
        
taking these as starting points
        code what the tactics are
        and look for round 2: what is missing
        round 3: not just one clever person that knows what they are doing
        
        pragmatic: what can i do with a student over summer
        timeline...
                lofi prototypes
                june/july work with student to do some experiments
                get things we have learned in for the election cycle
                
nick:
        one of the followers of the comcast twitterer is a comcast person
        i also know somebody who do their social network stuff
        what is the role of misinformation in agonism?
                some random comcast user was not able to connect through tor
                it was a home network problem
                it wasn't comcast blocking tor
                but this guy blogged about it
                it was picked up
                people in comcast were like: no!
                it had been picked up in media, retweeted
                
                
daphne:
        get people to reflect on the news that they are consuming
        browser plugin
        when you are on a news article
        scrapes content: find article on the same topic from a different perspective
                NLP: topic model of the article
                        topic model of other articles
                        serve the one that is most dissimilar
                        
j:
        critical news bot
                news is filtered/bubbled/selectively attuned
                try to open up to different perspectives and situate your own
                jailbreak the patriarchy:
                        automatically change the pronouns in the page that you are looking at
                        different ways of seeing the same reporting
                        gym: multiple channels, same news, reported differently
                        
n:
        do you have a fixed set of resources that you pull from?
        
d:
        in the beginning, but we want to scale up from there
        google news for the most relevant articles
        and then hitting links and start from there
        
        
c:
        you can easily choose a handful of media outlets that are readily available
        it would be interesting to return back to the user
        highlighting those passages where you see the dissimilar sentiment
        is it as simple as a simple adjective
        is it an unruly protest or an effective protest?
        providing a sense to the reader where these distinctions are happening in the journalism itself...
        
j:
        same article with a couple of adjective changes
        nobody did any reporting here
        
s:
        what is the goal, self-reflection or a reflection on journalism
        
n:
        it is nice to watch time to time and see where people get their ideas
        
d:
        implicitly you are assuming that there is some center in a way
        
j:
        it is about where i am when i see all of these different outlets
        
        
n:
        fox news and huffingtonpost is now realigning again
        
c:
        history?
        
n:
        truthy project
        sam woolley alluded to it
        researchers in indiana
        different news articles
        outlets
        and evaluate how they lean in one direction or another
        it is highly controversial
                they did something that pissed off congress
        
s:        richard rogers
        issue networks and crawlers may also be an interesting project to look at
        
n:
        news crawler
        look at search engines
        when we query a search term
        how is that going to return different results
        farmed it out on a distributed measurement platform
        and issued it in europe and elsewhere
        inconsistencies
        chrome plugin:
                ukraine riots
                i get this
                the plug in shows you things you didn't see that other people are seeing and you can see those results
        the work was incomplete
                we didn't get into how user profiles and history effect this
                that would be a neat area to follow up on
        
        we were curious 
                search history effects
                but also certain kinds of attacks
                and your search history can get polluted things you never searched for
                i am a sophisticated a search engine optimization person
                i want a pair of shoes to show up
                it is possible to do that, you searcher for nike shoes
                you never searched for that
                you just searched for shoes
                
j:
        can they reuse some of your code?
                you were planning on looking at one article
                that is probably a processing and time limitation
                
d:
        once you have a framework it is extensible
        once you do it for one article you can do it for more
        i was thinking in terms of interface
        level of reflection when you see two next to each other
        that also could be altered!
        it would be easier to display links
        
j:
        if there is an architecture we can use from nick

n:
        an aborted project
        we started with search and news seemed like an interesting place to go
        google news aggregator
        we took a whole bunch of rss feeds
        and pulled articles 
        and fired that back to google news
        what sources do i then get
                in the us: washingtonpost and reuters
        query inconsistency:
                these are for some fix number of queries
                there are a certain number of instances
                where you query a term
                you get an article in n-1 countries, but in one country  it is missing
                something is going on there
                there are cases where there are no news on that particular term
                
j:
        one of the things you can look at is to use rss feeds
        it would give you scrapes of titles and summaries
        
c:
        digital methods
        the issue crawler
        the tools that are used for social scientists used to crawl the web
                
        
        
c: 
        tom jenkins, pulling tweets on traffic to do sentiment analysis
        
j:
        this is for class
        she does NLP
        she does sentiment analysis 
        this could be as wizard of oz as you wish
        i just want her to show people a couple of articles at the same time
        
c:
        he thought this was going to be a joke
        doing sentiment analysis on social media to feel the pulse of atlanta
        it worked enough for them to find it appealing
        
        
n: 
        another person who has thought about that is hans
        
c:
        it can also be interesting to think of projects that don't fit the agonistic model
        michael hoffman
        political scientist: a project that is the opposite of ours
        argumentation mapping in community decision making
        if you map the ways in which decision is made
        how topics are emotionally charged, you can do more rational decision making
        
        this is the system that tries to dampen some of the exact things we say should be part of it
        if you make political decision making logical, the best solution will come to the fore
        very german
        
j:
        in the chapter in the second book
        decisions about space craft
        is this spacecrafy going to fly past this part of titan
        i thought it was whether it is about pasing by titan
        but it is also because there are ice volcanoes in titan
        the people tat believe there are want them to go there
        and the others don't
        so you look at the discussion
        but it is about who will be able to feed their graduate students for the next 20 years
        how do you understand all of that
        
n:
        that is fascinating to me
        this comes up in networking all the time
        i recognize: i was at a meeting at the fcc
        we were talking about measuring speeds
        clearwire: they were in atlanta
                let's not measure latency
                that's not cool
                users want one number
        well you guys suck in latency
        the throughput is ok, but packet loss is horrible
        i wished somebody would call bs
        you think you are having a technical discussion
        and it is something completely different
        
        it is a shame that there is only one janet
        i would love you to sit in this group
        the broadbant internet tech advisory group
        they write tech advisory documents for the fcc
        here is how engineers use prioritization to manage resources
        the discussions we have in that group, let's not talk about this
        are we having a technical discussion or is this politics going on
        it will look like some weird oblique thing
        as an engineer what i just heard makes half sense
        i can't totally argue with that i am saying
        it sounds like bs
        what just happened there
        that paragraph just got removed
        and i am not sure
        they say it is about readability
        i don't know what is going on
        
j:
        there was a guy from brazil
        ieee decisions about 802.11 wireless standards
        purely technical decisions that were not purely technical at all
        mobile broadband access stuff too
        he figured out what they were saying when they said things
        i wish i had more of me also
        i can teach you how to do that
        
j:
        values framework from shilton may be interesting
        but you still want to be able to talk about these things
        you don't want the company to go out of business
        but still talk about these things
        

j:        i really like how the data is made thing
        homes as entities: tweeting about traffic
        
        i like the candidate bots
        it is funny to watch the donald thing
        it would be really neat to have all four of those voices talking with each other
        
        i like the autobot tweets at the companies
        craig has a whole theory about bureaucracy
        utility companies now have extensive bureaucracies
        you have no recourse
        there is nothing you can do
        and they are all regional monopolies
        we could not get antyhing but comcast here
        (n: now there is verizon)
        craig grew up in communist china
        in china and egypt
        the service was on the ground
        but you had recourse
        if you greased the right palms you could get back service
        you can't bribe your verizon person
        you can't bribe them into getting it back
        infrastructures of basic utility provision are
        and what is the recourse
        there is no levers that you can pull
        if it were flooded with tweets from locations
        this service is down here
        

n:        i showed you a bunch of my slides
        couple other
        projects
        random ideas
        technology for agonism
                bobble
                pollution attacks: people stay logged into these stites
                you go to website: you can trigger the browser to load any object
                we triggered it to load a request to google
                as a result, your search history
                defense: against cross-origin requests
                google should not accept query from a third party referrer
                the defense exists
        agonism in networking tech policy
                we are engaged in a particular project
                debate: comcast, partner would be netflix
                how much congestion is there between comcast and netflix
                there will be fingerpointing
                john oliver got into this
                the debate is: netflix will say comcast subscribers are paying comcast to reach us
                they should upgrade their network
                comcast: netflix you can congest links at will
                you can choose which link the content will come down on
                there is a lot of finger pointing and not a ton of data
                isps said they would like to release some data
                how congested is link number 2 between netflix and comcast in dc
                the issue is
                there is spare capacity on these links in these cities
                if there is slowshing around of congestion that would be apparent
                for reasons that i don't understand yet
                there is a lot of sensitivity about how many links are provisioned in each city between comcast and netflix
                how much capacity is there in atlanta
                some of this stuff is competitive in nature
                maybe they don't want to reveal how much capacity they have for video
                but some of it is clearly, it doesn't rise ...shnanigans rather than shadiness
                you can see that netflix traffic carries more value than youtube traffic
                if you look at: if this youtube or netflix
                for google youtube is low grade cat videos
                netflix: paying subscribers
                netflix and comcast have incentives for reasonable quality
                these kind of things start to pop out of the data
                i asked them: i saw some anonymized data
                why is there high utliziation to comcast in atlanta an chicago
                that is where the youtube data centers are
                that is a perfect example of something that 
                on the one hand maybe they don't want to expose how they provisioned youtube
                on the other hand, there is also info there that is being withheld
                the fact that we are running youtube at super high utilization, subject to congestion is not something i told you
                there is a debate about who should pay
                where is the congestion: are they paying games with where they are sanding traffic
                or is there negligence with upgrading links
                they on the one hand want to release data to the public to show that there is spare capacity and utilization
                and withhold
                and this is one of these things
                what kind of discussion am i having here 
                are you concerned about proprietary nature of things in dc
                or are you trying to hide the fact that you don't have enough capacity in the city
                i spent many weeks and months trying to figure out
                am i being smart about this
                if you are going to basically engage in this debate
                as an isp or content provider
                you also have to be concerned about a skeptical public
                that would say you are hiding and not contributing everything
                this is a rambling way of saying
                debate in networking and tech policy where there is opposing side
                and data economies
                people are making decisions about what data to release
                and it could tilt the discussion
                the reason they are making these decisions is not entirely transparency
                clearwire example
                how do we decide what data should we even be collecting
                what should we release
                how should we present it?
                on hourly time scales, 5 min time scales
                
j:
                where does the agonism come for you?
                two sides and the data released?
                
n:                 this epitimizes the existence of citp
                the discussion primarily involves hobbyist
                the technologists are part of your agonistic graph
                i would be the first to say as a technologist it is hard to understand what is going on
                also crypto
                these are lawyers making arguments
                on one side or another
                
                historically, these have been only lawyers
                having these discussions
                or lobbyists
                with folks like ed
                jonathan mayer
                ashkan soltani
                you have these technology expertise brought to bear in a discussion
                that is not taking place on that axis
                ed summarized this best
                i said, when i go to dc
                you enter this other universe
                you think as an academic and scientist
                you expect to be taken seriously and at face value
                and it is the opposite
                we think about outselves as impartial academics
                it is more like: who do you represent? who is paying you...
                oh you are paid by the isp lobby
                you work for fcc
                i can plant you in this agonism graph
                you are on that side of the dyad
                you are an academic, i cannot place you
                you are way untrustworthy
                i have no way to orient you
                who the heck are you?
                it is the opposite...
                you are more trustworty if you are allied with some part of the dyad
                i find that to be interesting
                it seems to be in these things the technologists are somewhere in that graph that is not exactly dyad
                i am not quite sure
                these don't play out on social media
                i think that is going to change
                that is changing
                
j:
                what if you had censors on each of the db
                instead of talking about internet traffic to neighborhoods
                so that the pipes could speak
                there is a whole sociology of technology
                the question would be what voices are not in the current conversation
                and that is the service and concent providers are in the conversation
                and maybe the two things are combined
                in the interest to keep those things separate
                no voice will be unbiased
                are there voices that you could include
                that could be from particular kind of pipes
                even surfacing those pipes may be interesting
                i hadn't thought about how requesting netflix means comcast in between
                
n;                that is directly my real house

c:
                how much can we know about those pipes
                let's look at the pipes to understand the home
                if my reasoning is correct, you can do something similar
                can you get that? what can you know about those pipes with the kind of network measurement
                
                
n:                this is dependent on the isp release
                you can sense from the ends
                the analogy is apt
                the more sensors you have at the edges the better idea you can have
                there are a bunch of people with verizon
                if we can figure out who is having trouble 
                if we could automate that kind of thing
                the comcast tweitter bot is very interesting
                it would be sweet if we could have something like, that guy tweets, if that could trigger another speed test elsewhere
                
                that guy is in dc
                i tweet, my performance sucks how about for you
                is it down for everybody
                people have phantasized about that but there is no such system
                

s:         who else should be in the room


c:        instead of waiting for the centralized authority
        you have all these parties going to measure
        and collecting that data and organizing in a way that reports back
        here is what we are going to do
        end user reporters
        data may not be the same but will have an affect in the policy argument
        if you take the model of how we monitor the environment
        in ecological issues
        and think of that in the network environment
        and project forward and discussions about smart city rollouts
        not just about netflix, but which neighborhoods are going to get what kind of bandwidth
        atlanta vs. chicago
        taking the home and city as different scale
        
j:        the google roll out
        atlanta...
        
c:
        neighborhood advocates and representatives would be interested in this
        as they do airpolution monitoring
        in the smart city world you want to know what are the services we are getting in this vs. this other neighborhood
        in 12 months, when these things roll out, but these very well be an issue
        it scales nicely
        
j:        students studies utilities
        regulation of price control, railroads, broadband roll out
        we see this over and over again
        this story is converging towards x
        but if you take the agonistic frame
        those are two things on a trajectory which wil be mutually beneficial to the two of them
        there is not going to be any innovation there
        what are some voices
        
n:        another one that pops up in the dyads is in the IoT security area
        security on these devices: who bears the costs
        and who bears the liabilit
        and that is like a dyad, but maybe 2.5
        the device manufacturer and the consumer
        everybody suffers as a result of the insecurity but somebody has to fix it
        
j        now it is on the user

c:
two angles:
        how are services constructed
        literature on alternative economies
        urban foraging and food gleaning
        models that say
        you are going to have industrial agriculture
        you are also going to have alterantive models and coops
        how do we understand the models
        and how they interact
        an alternative economies approach can be taken towards our devices and what are the choices that we make
        and from a research perspective, how do you document those and understand the models people are building
        
j:
        google has collapsed the network into a single node
        there is something neat about that
        it allows you 
        who is providing the server, the service, the fixing, 
        and that could be exploded into many categories
        
        social construction of technology framework
        flexibility about what the object could be
        dyadic: the relevant social group is defined by the company
        user and company
        those are the only relevant groups
        if you are doing the isps
        and media provider
        and those are the two relevant social group
        they are being made irrelevant
        what would happen if we had the relevant social groups
        classic example is the car
        automobile: main users in rural environments were female
        farmwoman
        they plugged their washing machines into their cars: generators
        tractors
        ford: we sell tractors
        and shut down the repurposing to exclude women out of the equation
        a lot of people collapsed into a single category and node
        make them relevant
        
n:
        who controls how the thing gets collapsed
        if you think about the comcast netflix thing
        google or whoever: they have the incentive to make the dyad look like
        user vs. isp
        the isp bares the cost
        for users the first reaction is this is comcast
        who bares the cost
        the isp tech support is the front line
        it is in every one else's interest
        content providers saying this is not our fault
        the knee jerk dyad there is that
        if you look at the performance the graph is much more complicated than that
        it is a network graph
        what are the nodes that affect whether or not you get a high quality stream
        it could be the stuff in your house
        the whole debate gets collapsed into user vs. isp
        
j:        
        sandstorm.io
        server administrator tools to manage your own servers
        make it easier to install your own packages
        they manage the security
        and you can either use their server cluster
        privacy policy: they don't touch
        or you can pay for an upgraded version: financial model
        you can buy it and install on your own server
        they are allowing you to decide who you would like to be your sysadmin
        tools that would make it straightforward to be your own sysadmin
        it is even more complicated
        there is a series of different models that woudl allow you to do that ll that management
        




carl revisits last session

things we want to achieve
        list of papers
        workshops
        
1) we start with CACM short position piece
2) longer piece to first monday
3) target symposium at CITP spring 2017 (soon after elections)
4) followup with maybe a CSCW workshop                


that was paper deliverables 
we also talked about creating bots
actionable objects
        guidelines for resisting IoT
        bot guidelines
        a toolkit to create an agonistic bot

design based research                
        alterantive service models for IoT
        case studies -> to infer tactics
        

is it agonism that is driving us
or is there something about bots that we are interested in 
do we want to zero in on bots?
do we want to focus or do we want to give us a capacity to say, we are just as interested in this at the level of network measurement
or something else

j:
situate what we are doing
see what agonism gives us that other models dont
many other projects about this thing will assume an easy dichotomy
the good and bad
assume a specific object, bot or algorithm or a platform
as if it is itself not connected to anything else


n:
how does agential gerymandering relates to agonism?

j:
there is that question of many other projects focusing on bots
and algorithms and platforms
all of those assume that a bot is an agent and it it is the object that is worth looking at
that already does agential gerrymandering
focusing on the bot is already doing that
as something we can take out of its context, regardless of where it comes from, the data that it is using
deciding that certain something has agency
when do we decide that something is a bot
some of these bots are also humans tweeting as bots


n:
does gerymandering refer to perimeters that determines who can talk to who?

j:
it is about what gets to be an agent
who gets to act
is the bot something that acts in and of itself
we do agential gerrymandering as well
the policy makers have already drawn the boundaries of a discussion
and if you come as someone outside of that you have no agency

c:
it is question about this drawing lines of inclusion for your benefit
and by excluding these other things, it will be for our benefit

n:
do you view your efforts to disaggregate your data as a agential gerrymandering?

j:
by drawing boundaries i am doing some gerrymandering
so i am deciding which spaces those objects get to act in
mars rover, is it a he or a she
does it act because of us or because of itself

data balkanization at home is a neat way of thinking about it

n:
i would be interested in talking about that
because of our earlier discussion about little control over how devices exchange data
and there is some interesting technical problems


j:
maybe that is more about agnosim
rather than thinking about the "smart home" single unified home that is seamless
this samsung light, and apple tv, and draw distinctions within and outside of the home

n:
does it relate to political agonism?

c:
there is nothing in the technology
i think that is the agonism that happens in the process of the set up
in creating those boundaries
and thinking about the ways in which you as a user are refusing to participate
and denying these objects from participating with each other

n:
there are these relationships that you have with google
and someone might not be aware that google and nest are not the same

c:
so you are resetting the boundaries
it also comes back to the discussion of affect
and it comes back to your question about your affective desire
that is what is driving your decision making

j:
agential gerrymandering
nest is in the box
and there is a thing
it looks like it is all in the object
it is obscuring this whole network of connections
and property model

n:
you wouldn't put a bug in your home
but you would put a bug on your thermostat
i would be interested in exploring 
get a bunch of devices
and try to figure out how they communicate

there is some very early days things about
one guy decided he was going to block the dns of his thermostat
and it still worked
and you are still able to control the heat

i think there are these
i don't know if this is gerrymandering
and can you control the device
it is tough
are you going to hack into your nest
there is limited time for that
you might though put a wall in front of it
and put a wall around it

i think from a network tech perspective
that is very interesting to me
how do you build this walls

j:
can we MITM nest
so that nest doesn't talk with google

n:
interesting experiment
set up a harsh firewall
that blocks everything
port 80, dns, maybe dns is too much
block it: does it work?
what cracks do we have to open to make things actually work
are we happen with those cracks?
for every device it is going to be different

c:
in that process of revealing the network
what are those networks that we are not aware of

there was work done early on
one of the things i wrote about
was to show the interrelationship of fortune 500 companies
what are all the organizations that shell is funding to do climate science
how do you understand how
what are these technical dependencies and relationships that are happening

this is what excites me about the humanities perpsective
because we want to show how we can do these things differently


j:
people are putting echo on raspberry pi 

why do we call it an internet of things
why don't we call it something that shows that it is an extension of a bunch of your databases into your home

i am not pessimistic, it offers opportunities
what if we changed the script and made it do different things?


c:
things network in amsterdam
it doesn't rely upon commercial service profiders

lorawan video
we can provide a data connectivity for a city
amsterdam covered in one week


j:
you can do this because you have been working with computers since you are this big
i think that the goal of some of this
what would it be o build a bot, when you are being attacked by misogynists
and you could launch your own attacks
can you give people the tool to take control of some of this stuff
without much of the skill
we have the privilege in the university
the profit generation model
i don't believe in one profit generation model
if one of the things we are interested is in agonism in the increase of opinions and actors
we can use the tools that we have to spread that privilege


c:
phd student on diy infrastructure
it doesn't work
the whole point of infrastructure is that it holds up
they will work for isolated moments
but this is what the state is useful for


n:
going back to the gerrymandering fences thing
i am interested in that
it is a networking problem
there are a couple of other things that make it compelling
that it is tough to control a device
it comes out of a box
the other is coming back to what we are talking about
switching costs may be high
switch off of facebook or google
various elasticities there
and with devices that might be harder
i bought a 300 dollar thermostat
and you want me to remove it for a cheapy thing
that i have to monitor
so, we are going to put it behind a fence

s:
i feel like it is a tactic

n:
people will have the devices
and they will have the incentives cooked into those
but can you mediate them anyway!
it is an arms race
and some legal action
a firewall: that keeps logs in house
but then a firmware update comes
and that is gone
or you don't do updates

j:
how do you manage to live wo google?
it is surprisingly easy
i have done that for four years now
it is hard to fence yourself to one option
and creating the leaflet
here is how you circumvent your nest
here is a tech you don't want to use
here are your alternatives
can you write a guide for how to get off of google?
even leaflets for iot configuration
is there a device you could reroute your devices to your home
take this raspberry pi in front of your network
and we will take care of the rest

n:
this would allow us to not have to configure each device individually


j:
hack your netflix socks!


c:
if there is research on this
if households are making decisions on services
based on i like this company but my partner doesn't 
we are going to agree that there are certain things that we are not going to support
it is much more related
either there will be a nest thermo or not

j:
my husband has google products
when apple maps really sucked
i would ask him to look for things on his phone
and it would also mess up his results

n:
did you consider swapping acconts with people?
something i do
cvs
the loyalty cards
they also have a phone number look up
and i enter a different number every time
princeton main line
major businesses

c:
it is then interesting to think about how this scales up to the smart city
entire buildings or neighborhoods


agonistic frame of tech/bot development/design

what do we mean when we say agonism
algorithms/bits and bots
we can turn it into the acm article



n:
has there been an agonism meets cs

seda: governing algorithms


n:
tussles paper is very relevant
it is relevant
there is a paper by david clark
an old school internet architect kind of guy
in tech there are often conflicts and tussles
and they often play out in the technical realm
are we going to have
what is this protocol header going to look like
it plays out in the technical space but it is a policy discussion
and how we need to design tech to enable these kinds of tussles

-janet is taking note of what nick is saying

continued




To do:
- Carl gets in 

        

                




NOTES FROM SEDA'S PRESENTATION

Agonistic encounter - what could be the full scope of the agonistic enounter?

How do we structure our colleaborative work so that accorss our proejcts we engage a fuller scope of the agonistic encounter with algorithms

Cybersecurty approach
• attacker drivien reserch
• how to dring agonsm and to whome

Who else might be included? 

Designers of Algorithms
• think not just of the attacker but other players!
•  the obsession with "erasing" the attachers
• good and bad users
• using "good users" and "behviors" as benchmark

Users  
• highlight security paternalims
• encourages embracig the autonomy of fdeivces
• how to rething algorithms with user paticipation 
• can we make users aware of their lack of control of the device?

Nick
particularly distubring proposotions
you are going to accept these certificates if you want to use the internet

Seda
the quetions with regard to security is how far we should be pulling the users into security models
how to balance not overwhelming the user with providing the user with choice

Nick
This is an issue that comes up at places like FTC PrivacyCon
Transparency, but not sufficient
Choice, being an important factor, how to mediate and provide a range of choices (can I install this thermostat and NOT have ALL of my data go to Google?)

Seda
What other service models might be possible, could a local person become part of the installation process?

Janet
What about maintenance? What if I want to control my own devices?

Nick
Another example is the hotel wifi blocking. 

Carl
We could approach this from a services perspective, from services literature. or we could look to the alterntive economies litera

Janet
What google has done is collapesed all of theses other (the ISP, the sysamin) into itself to create a dyad with you, so it's only you and google, also SCOT that shows how users are constructed and how users are created as well as written out of history 
    
Seda
Next step in presentations - how to bring agonism into activist research

Longitudinal study in countries where the study is not a dange but censorship is big

censorship
the blcijing is only one strategy
throttling as a way to do soft censorshop
trolls: psychologicl warfare
• think of what censorship is trying to stop?
• speech?
• witnessing?
• organizing /mobiling? 

image of the stack of censorship

understand the stack but also what is at stake 

longitudinal studiy
• shoe which groups topics are being effectedthroug the stack
• trolles exist on all sides? women, kurds hit the worst?

allow the performance of that which censorshp tries to stop
• I am a witness
• Another way of organizing post censorship


agonism in the bot

• fact, topic comprehension
• topic models

• learning from social
•summarization and explainers will liberate users from having to read every article 

•emotion understanding
•affective computing

•contextual and episodic memory
they jump between topics, incapable of sustained fliw or conversational rhythm. past conversations in context are hardly reintroduced.

•personality 


agonism bot
• play with the different aspects of a bot to make people think critically about bots
• how they are configuring us as subjects
• what do they make in/visible 
• speculative work? how the bot transforms our interaction with the IoT space 
control over our lives rather than objects
but really? control over our lives (by whom?)

bot is a way into IoT

Nick
What happens when I plug things into my network. What happens when I plug in some device that I bought on Ebay from China - I'l letting in my house and I don't know what that means or what its doing. 



DELIVERABLES
Papers
• CHI / AltCHGi
• The Atlantic
*** CACM
*** First Monday
• Computational Culture
*** ESTS

Workshops
• Purpose 1: Bring peope together interested in studying this (practice)(industry)
• Purpose 2:  Introduce to the idea (seduction)

Symposiun
• 2 day
• Connect to elections (post elections)
• CITP
• Exploratory / not just conclusive

Things
Both objects to think with and actionalbe objects

Actionable objects
Browser plug-ins

Bots
• Candiate bots in debate
•

Critical Toolkits 
• Download, set-up and then execute bots on issue 

Guidelines for political bots 
Guidelines for IoT resistance 

Design based research (things to think wtih)
Alternative service models for IoT

Case studies  --> inferring tactics

Conceptual Work
Take things seriously in political theory 

bots not just as proxies for human ideologies 

prosthesis - take this metaphor to be not just doing human work but human work + 

bot as complement

human action + something lese

different kind of extenstion - not just imitating

like a rover 

bots as doing work on behalf of the bot



SCOPE

agonism explodes the possbility from any dichotomy and insists on a pluralsitic domain 

these struggles are not taking place in just a bot or platform or algorithm but rather happening all the game through the agonistic game

makes explicit the cut / included or not (Strathern, Barad), both analytically and technically 

agonism already in the creation of the data

what objects are acting

in / out

humans / nonhumans 

instittutions

troubles categories

agonism is able to help us more the site of analytic / technical intervetion around the field of engagement (and actors)

a design perspective should open up the space of encoutner 

how is a bot different from any software service? 

what makes a things agential?

bot presents itself

bot interacts with us through channels that are primarily human

when do people ascribe agency to a robot 


How do you build these wallls



Tussel space
often thooug


HERE IS WHAT WE THINK THIS WHOLE WORKSHOP HAS BEEN ABOUT:
    
    
    Agonism and computing


There is a paper by David Clark who describes how in networking, there are conflicts and tussles that often play out in a technical realm. For instance, what is this protocol header going to look like? It may be a policy conversation but it is playing out in a technical space. We need to design a protocol for these tussles. You have opposing forces, if you will, in a particular conflict or debate or interaction. And often we think about those as dyads - diametric poles in a particular conflict or debate. But in fact that tussle is playing out in a much larger space. We describe many examples where a “tussle” is in fact not just about a dyadic interaction but takes place in a broader space, with more interlocutors. This may show up in the offline world and shows up also in bots, as in “political bots” and it may also be in the design of technology itself. We want to expose the “agon” so that we better understand the debate as it plays out, and also so that we can find levers that change or tip the balance in a different direction. Why should computer scientists be interested? If we can understand the AGON as builders of systems, networks, and protocols, if we have a goal (like a privacy goal) we need to surface the multiple voices involved. And not say, It’s google versus us, or it’s ISP versus … and if you want to protect your data from going over HERE you have to go over THERE - or it’s not Hilary it’s an agglomeration of micro-PACs. Identify new parties to the discussion.

Technology FOR agonism and agonism IN technology.
(many of us have been doing this work without even realizing it.)

CARL: (To DHUM and Media)
Humanities gives us a rich understanding of politics and democracy in ways that are multiple. Agonism is one of the ways of understanding democracy. How do ideas drawn from political theory in the humanities and social sciences take form in computational systems, whether in algorithms or platforms, or in bots or other forms of media presentations. What are the ways in which some technical systems are already doing this but haven’t been labeled as agonistic, and what do we gain by seeing these bots as dong the work of agonism? How does this enable us to theorize technologies differently? If as humanities scholars we should be bringing 
bring political theory to provide richer interpretations of the ways in which technology is doing politics. 

To HCI colleagues, a little more instrumental. We talk a lot about how social media is used and misused for good or social ends. but we paint it in broad strokes. It’s good, or it’s bad, we have attacks and victims, but it’s more complex. As people who do HCI design and research we have a responsibility to map out this complexity, so that we can make intentional designs of systems that produce different kinds of politics. The other thing that we’re doing, its that we’re taking the technical seriously. It’s not “computers” it’s how the network is actually functioning and how bots work or don’t work. To me that is crucial to the humanities and HCI which too often gloss over “the digital.” Taking those specifics seriously and treating them as the engineering and science topics that they are.


Janet:
    Doing agonistic technology design is a question of exploding the dyad. Take one or two relevant objects that look like the obvious objects --    the user and designer, or the isp and content provider -- and explode that into as many categories, objects and people as possible. By thinking of things as a dyad we obscure the large number of voices that are subsumed and left out. What if we draw these differently?  To an extent, this is a form of agential gerrymandering. These objects do what they by associating themselves with other objects. We can map those connections. The goal of  design is to inspire this broader network and to intervene in its associations.
    
The critical news bot does that by taking the first step: it says, what you see in the news is the beginning of a puzzle with a lot more people and actors than even just CNN and Fox. The political bots talking at each other imitating candidates,  is a way of showing how everyones is sutated in the netowrk and how these bots act in a larger network of things, people and interests.  What does a technologist do wen ou are shut out of the conversation because they don't belong to predefined camps, or when it is defined by two opposing sides? Can we make the other objects speak? Can we make the network speak? Agonism it about moving away from antagonism -- the dyad and affective relations between poles, the good and the bad -- and putting into view something multiple, and allowing those new voices to be political.
    


SEDA:

To think actively about agonism as a frame through which to understand different technological practices which includes the design of objects, the process of design, research, and tech policy. I think we explored not only the use of technology for agonism but also think about agonism within design and research. Through that we were able to see every technological step as a contestation in and of itself and as a moment of reflection. We explore both methods and tactics that would either bring to th esurface these contestations that technologies enable, or contest the way we do technology. Some of these methods I think encourage us to think more wholistically - the dyad is one where we say where are th other actors who remain invisible, but also what we call Layers enables us to see technical and social effects across different layers. And that allows us to think about alternatives. So agonism is a way o fcreating contestation and thinking about alternatives.





in the antagonistic view, emotionally is highly charged. agonism, when you do that explosion of the network of not only dual opposing polls, there are these different voices and alliances, that totally helps manage an affect question. you don't have a predefined category through which emotions are already preconfigured

n:
    the comments to my freedom to tinker blog post was emotional. here is what net neutrality is about. but that is not going to put an end to the comcast netflix conflict, because that is an economic discussion: who is going to pay for this uplink. it is orthogonal. i wrote a blog post explaining this, and people were outraged. the comments are englightening.
    
    
   J:
       
       infrastructures remedy some of the problems with ant
       in ant human and nonhuman are all on the same place
       in infrastructure studies the decks are stacked
       the way that the network is laid out has a topology
       and things are nor symmetrical
       by surfacing different options we are not making that mistake
       we are trying to explode an already unequal relationship
       we are trying to diffuse them
       we are maybe flattening them a little bit
       and we are doing it in a highly politicized place
       
       
c:
    i don't think we flattened anything
    if anything we pointed out the importance of keeping the peaks and valleys and whatever the right metaphor is in variations in different people's positions and opnions
    some may say we have flattened the idea of who can participate in politics
    by taking bots seriously as political agents
    this is a matter of lagauge
    we have been more pluralistic about who counts
    we didn't say everything is equal
    we are including more things than we have previously included