Discussion on the publication
Publication - version for transmediale
We will set up a new folder structure for this and send you the link
DEADLINE for contributions 4th NOV.
FANZINE printed at 17h
8 slots of 20 minutes in afternoon (14-17h)
you can make multiple pages with 4 types of content on each double page
anyone can create individual page within slot
Please have a look at instructions for basic lay-out (using Markdown) for .txt files :
http://etherbox.local/pad/p/cheatsheet
Notes of talk of last night, please paste here:
http://etherbox.local/pad/p/seda_antoinette
SCHEDULE
13:50-14:00: last check
the FANZINE will be made out of 8 CHAPDERS (not necessarily separated), each made out of an arbitrary number of PAGES, each built out of 4 FOLDERS, which contain 1or 2 BLOCKS!
CHAPDER 1. 14:10-14:30 Dave (glossary), Daphne, Maya (scenarios);
CHAPDER 2. 14:35-15:00 Sam/Nathan (reading), Soren P/Christian, Rosa
CHAPDER 3: 15:50-16:10: Femke, Martino, John
CHAPDER 7. 16:10-16:30 Silvia, Renee (disappearance), Jara (writing with scripts/cheating!)
CHAPDER 8. 16:30-17:00: Abelardo, Maja, Brian, Roel, Martino+ flickerbook-sentence Sam/Nathan
FOLDER 3. 17:00-17:10 Nicolas, Michael, Geoff, Soren R (OCR)
---
17:00: launch fanzine + talk about next steps
Hey, Let us back in.
FOLDER 3. 14.50-14:40 Nicolas, Michael, Geoff, Soren R (OCR)
Collective decisions on:
* elements on the page:
4 folders (each folder corresponds to box on the page in which you can select:change content): you can name them as you want
images
* references
* colophon
* license
* deadline for final texts for publication created for Transmediale in 1 or 2 week
Sarah uses Markdown for basic text lay-out (title/bold/italics/placements on the page)
https://daringfireball.net/2004/12/markdown_101
Proposals
Dave: python script filter, searches phrases that looks for repetitions of phrases, converts into acronyms, produces glossary at the end
maybe just present glossary - have original text online ACRONYM
-> script can be used by others
Method research: based in grey media, looks at artefacts of bureaucracy, my texts relies on these types of materials
use them to reduce repetition
Maya: US army development of scenarios defined in literature as stgh in between storyline script & a roleplaying game
rewrite ethnography of error as series of speculative / fictional scenarios, to show that ethics is something that gets produced contextually in different scenarios; can you read across scenarios to identify factors and dynamics that constitute "ethics".
(Imagine a 15 yo wants to take out her mothers self-driving car to go out at night ...) SCENARIO
Abelardo:
creating flow so that all the pieces can be read as 1 stream
algorithm building this flow / stream
extracting "semantic similarities" between texts with features
Python script measures quantitatively proximity between 2 texts based on frequency of words
script can be adapted to relate 2 or more texts
measures deviance of use of words and closeness to document that uses same deviance
way to order the texts
-> script can act as index
-> can be run after all slots have been made
Maja: organise texts following atypical ordering
ex. sentiment filter: have most negative texts in beginning (happy ending filter ;)
Brian: set up neural network that takes in text, tries to learn language
take any input, generate any amount of babbling
-> take in etherpad
Silvia:
writing with disappearance script
Shannon: the more options, the more rare, the more information
rarirty doesn't mean anything, does not imply semantic significance
most of the words we use more often will be deleted
when processing a theory text, would essential texts be deleted in addition to more "stop words" (prepositions etc)
look for semantical word that don't seem essential to my thinking
visual preciousness, importance of information
either core / semantically not interesting is deleted
desire to write a text designed to be passed through the filter -> cheating!
Geoff: tried running dispapearance on script itself
Renee: anonymization
redoing the text with the text
http://dataprivacylab.org/projects/scrub/
https://github.com/lihararora/anonymizer
Jara: 3 examples of text logistics
apply machine operations on theoretical part of text(?)
using Graphiz (?), rewrite to cheat semantics
Roel: using Traceroute (network debugging tool)
run it on all texts to annotate web ressources and look at different networks that underlie publication
Sam/Nathan: typeface making reading easier & more difficult
are rewriting to 1000 words
write it in columns, 1 word/line
flicker book component for Transmediale? over all the pages
Soren P: algorithmic readers
Soren R: slow reading "speed doesn't mean 'fast'"
why words look the way they do
reworking text in radical sense looking at slowness of information transfer non accessible to humans
Nicolas: will not be there this afternoon
images
OCR script that looks at images, extracting letters/words, produce text with that (ex Hamlet map)
-> for TM publication
-> add your images if you want them OCR'd
John: how the pads function as summarization of oral stream
represent multivoice conversation in MD/publication
Unicode/author notation for each voice
notes of what I've said, add comments in right place
Femke: last night lots of people took notes
organise all personal notes on 1 pad
1 recording of Renee -> use Robert Ochshorn Gentle for automatic transcription
https://lowerquality.com/gentle/
the recording is too large to work in Gentle browser
cfr interview with photocopier
make collective interview of etherbox: Q&A on 1 pad, we speak together for/with this machine
Casper:
loop situation
still working on reduction of text
Christian: same
Rosa: still thinking
Daphne: will send sthg for TM publication
Brian: will upload the neural net letter-learning prog.
Possibly will produce a 'ghost document'? doing an impression of the machine learning
==========================================================
COMMENTS
***Publication Fanzine -> Transmediale
DEADLINE for FINAL TEXTS: 4th November
TECHNICAL
printed fanzine has wrong page order -> right pdf will be sent by mail
bleeding edge
MISSING
titlepage
table of content?
introduction
add 1000words text for each?
add more experimental pages? not everyone has produced all the pages they wanted
add descriptions: add small texts that explain the 'treatments'? 'usage/about' in 3 lines (like comment above in code scirpt)
colophon with names of everyone
license / ISBN
pagenumbers? (possibly flickerbook speed reading at corners also? NJ)
thumbnails of the 'demo' in the final publication?
add scripts/filters?
QUESTIONS for Sarah
* introduction
(Daphne offered to write transitional introduction -> colonial act (?) -> write collectively in a pad -
Do we need an introduction? probably not by Transmediale.
Maybe an introduction per project. Not an introduction that frames all the works just from one perspective.?
Brian: I like an intorduction to the whole and to the parts.
Maybe just small. No discussion might make it into a throw away paper.
Geoff: A descriptive text is super important.
Should it be a comment on the presentation done by Kristoffer and Daphne ealier in the workshop.
cf. interesting TML contexts"Middling" "Ever Elusive" etc, the way the TML themes have evolved through the workshop?
Geoff: in the pastwe have given a middle page to the trandmediale. Do we want to do this again? There is a sense of legacy and audience.
transmediale understands it as an important presence.
Someone needs to coordinate this.
Geoff: We might act as a colonialist but we as Arhus university can probably take on the coordination. of this introduction. We will make sure there is a pad and a strucutre and people are invited to be part of the pad. If someone spams this, then this should be honoured (Femke)
There will always be a hierarchy *editorial control - but hopefully a lose one and one with responsibility
Femke: do we want it to be collaboratively written? Okay so when? We need a moment to connect to this 'live'.
Perhaps November 4th--- but Geoff will set a date.
* ReadmeSSSSS
We need a moment to connect to this 'live'. add descriptions: add small texts that explain the 'treatments'? 'usage/about' in 3 lines (like comment above in code scirpt)
What texts want to be read, and which not? In the README
Readme is optional. Everyone will have the option of a README, so write that. and send that
Søren: Do we need the author for pragmatic academic referencing. We can also acknowledge that some text are not written by singular authors, it might be a communal or machinic effort.
Geoff: putting names next to text is Ok
Nathan: Are we gonna have biographical information. Are we gonna have names? if you want your name near the article, it is your responsability, they will be in the colophon!
* add 1000words text for each? ok
* Title? "Machine Research is Machine Research is Machine Reasearch"
Lets take on a tautology, Just so maybe if we do not get invited this year, maybe we can get invited to Transmediale next year.
* table of context?
pagenumbers? (possibly flickerbook speed reading at corners also? NJ)
* license / ISBN
I hear si si cc (Creative Commons)
Free Art License, 'unless otherwise stated'
-> made with Free Software, published on Gitlab Repository, is Free Art License
* thumbnails of the 'demo' in the final publication?
Geoff is not with us. Renee is also not. Silvia and Rosa are not interested in this.
* add scripts/filters?
add link to gitlab
* size of publication
Last year was to attract attention.
It was hard to take care of it.
Rosa could not get it home in one piece.
Proposal for Tabloid size (small newspaper, like the one in 2014)
Something that can fit in carry on so it does not die on the road.
Geoff: We have an amount of money that pays for design, amount of copies and materials.
We need to negotiate with the designer specifically how much design will cost.
* colour? 1 colour?
Colour seems like it would make sense - given that there has been a tension between the ways computers and humans "see things"
how many - 3000 copies
negotiation with design//amount//colour//...
Geoff to put exact figures on budget on the pad... but approx £1000 design £2000 print.
Geoff: Newspaper is v. cheap! Machinic- efficiency.
Lulu: we could make one that people can buy.
Can we find some printers in Berlin? Risograph, who prints, environmental printing?
Newspaper Club - using The Times printer. BUT this is UK, delivery?!
Can we take the dead trees into account? is there a printing service in Berlin that is more ethical? - that would be quite a machinical statement.
Maybe we can ask Maja?
-> can it be small print run & read on smartphones?
fonts: could we use metafont, and what variables could we feed it?
style of paper?
Newspaper, glossy, in between, other?
Jara: there is a strange discontiuity between the people who are part of creating the content of the newspaper and the people that make the design decisions
Rosa: it should be clear that all creators should feel happy about what they get out of the publication: some people get payed, while other people get payed for being here, and somehow there needs to be a middle where effort.
We have all committed to making a publication over three days.
So.
Rosa: but perhaps this should take longer, in order to feel comfortable.
Voicing for people who are not actually here - not speaking - should not push to publish. Expecially given the context of TML.
FEMKE - also option to not be included.**
Geoff: no-one does have to do anything.
Rosa: but pressure of history etc.
Geoff: we are being transparent at all times, at each stage...
So there have been choices... but changing the plan has not been an option.
But then this is a late time to raise this. Too late.
Femke: shall we Constant
take the risk to bring veryone to TML Berlin...
Could there have been a moment in which we could have questioned the process? Have had a conversation about the process?
Femke: I am wondering if there could be a possibillity to bring people who are not backed by an institution, Is there a possiblity to have a pescence beforehand.
Rosa: I am not saying we should postpone the process.
Martino: But the starting point would be a different place.
Geoff: we are over thinking the way these open meetings normally happen.
We have some money to do more than normally any academic insitutition can do.
This is being over-thought. Food provides micro-nutrients for the brain.... and the body
Nathan: I would be really disappointed not to make a publication for transmediale.
Martino; this is not to critique the trust.
Renee; the question is also where do you put the feedback loop?
Jara: it is really important to bring the pracarity in the room. we have spoken a lot about ethics and politics.
Its a collective situation. if yesterday we could talk about the ethics of a driveless car, right now we can sit down and talk about the issues of symbolic capital existing between us.
There is some interest of symbolic capital because everybody want to be part of Transmediale.
Femke: Its interesting that we have this conversation at the end of the workshop.
Some things were only clear on Sunday.
Since we have created this collective environment, in which we build this thing collectively, could we also have
There is a confusion between the publication and its realtion to transmediale. How can we solve this?
Nathan: it seems like all sides benefit from it.
Geoff: the starting principle was that young researchers often dont have a voice. Often its middle aged men speaking on the stage, we would really like to have a platform for the other voices, give a new platform.
Now we work in these entangled conditions, which are full of oppossitions and we still try to create a proposal for stuff that works
Søren: maybe we can try and finish our texts, if anyone is not happy to publsih the project, they can chose to not be present at all.
Style of folding? random pages, no binding? fluid newspaper, put double-printed pages/posters in different places (numbered, together they make newspaper)
* print run: 2000 a 3000
INDIVIDUAL PAGES
Dave/Maya: concept of small/illegible texts + legible texts over different pages
Nathan/Sam: stick with technique, but deliver image of the page -
Chrisitan/Soren: 1 page missing, machine readers for input texts: highlight words // create texts vertically by highlighting letters, poem - using a "neighborhood reader (http://thereadersproject.org/readers.html) and a "mesostic" reader https://en.wikipedia.org/wiki/Mesostic
--- end edited (by humans) to become/look like a poem ... maybe we need to import them as images?
-> code format
Rosa: made text smaller & translated into dct (discrete code & transform), only legible for other dcts
Caspar: wysiwyg
John: collected all the notes of talk Tuesday night, reordered them to reconstruct event, author is represented by symbols
Femke: etherbox interview & automatic transcription of 1st part (1MB) of sound recording
Silvia: it needs rethinking
Jara: is not working: I took 1 post of blog make it read by Graphiz notation, some connections were forced, shows juxtaposition of diagram & part of the text, it worked as a proces of translating to 'dot-language', know what the axis is
-> put source of the graphs in publication?
Renee: 1000 words & added page rank algorithm (it's formula), then took the same text and used Constant's encryption script, another text is first the 1000 word text being read with voice of Alex for text-to-speech, then processed it with Gentle (audio to text), and last text is put through synonym replacement script, called 'Synonymiser' (thanks to Dave and Michael) because 'RankBrain’is rumoured to be the third most important signal, after links and content(words) and infers the use of a keyword by applying synonyms or stemming lists.'
-> how to order these texts? Easiest text to read first? Encryption at the bottom? Rosa suggested 'Page Rank' or a ranking in some way to reflect content, or have the texts as blocks. All possible. R. Queneau, Exercices de Style!
Michael and Geoff: machine ways of seeing the book: OCR inversion
tesseract -- and then it ate the text.
Soren: images gone through deep dream software
reminds Jodi, puts in hierarchy between human-machine -> needs to be decided?
Things on different levels. Layer; keep the confusion
resonance between one page and another, creates overlaps in readings -> you need to be alienated!
-> flipover between machine language & human text / variety throughout the book
back and front, different ways
Maja: reduced version of 1st text, part of stream of texts using links
small texts by Roel/Martino: references in articles filtered through traceroute -ISP here is Proximus, Level 3 links us to international group Endurance (fex) where website is hosted - each hour it will follow a different route
Abelardo: shrinking texts on top: analyses how much a text relates to other texts, in proportion of 0 to 1
text made into vector of terms, projecting to vector of terms of the other ones, per text: a rank listing of every other texts
'philosophy wins!!! ( Silvia's text is connecting every other one
*Readings* - was used as term by Geoff and feels very relevent to the plurality of readings happening
Brian: opposing pages, left is text, right is neural net trained on the text, regenerating 1000 words
ran it on everyone's text + content of pad : looks like Markov chain with new words, character by character
VARIABLES -> I put these up (under ISSUES)
Colour?
Page size?
Paper? tracing paper
font? -- could we use metafont, and what variables could we feed it?
***PJ Machine
change step of zoom to make it smaller
changing height of block
organise editing table to organise pages/folders/check files/MD
wishlist:
flip text into landscape
duplicate box?
bugs
table doesn't show up properly
indentation & bold (Chrisitna/Soren)
appearance of text in block on page (only showed till 'conference'): something cuts off the text (Etherpad?)
-> replace unknown character by return in txt file
-> use TEXTWRANGLER
container runs around the page
***Etherbox
how to delete files?
how to create folders?
show ranking of etherdump
agree on keeping gitlab repository after the session?
ACTION UNTIL 4TH OF NOVEMBER:
____________________________________