(B8) Social Media: Trevor Paglen, Working on Visuals for Spreads

TREVOR PAGLEN:

Classifications of Gait:

From far away it's difficult to understand what the content is, but then you move close up and it becomes much clearer that the content is being made up of lots of small images placed next to each other.

- I find the idea of having to look closely to get a full picture is something I think it really interesting about this piece, relates to our project and wanting to get the audience asking questions as they look through our publication.

Could we do this with google maps, thinking about having an image made from other images?

-What could this be of? Could document things in my own life, routes walked? Do these on google maps and screenshot the route place next to each other. Could find routes on Strava from other people's routes they've travelled. 

 

Portraits

These images are portraits of people who I have collaborated with and who have been close to me over the years. They are filmmaker Laura Poitras, AI-researcher Kate Crawford, and investigative journalist AC Thompson.

The portraits are made by creating facial recognition models of the people depicted (i.e. a computer model that can “recognize” that particular person’s face). Then I create a second program that generates polygons. These two programs go back and forth until an image ‘evolves’ that the facial recognition model identifies as a representation of that particular person. It takes several days for each of these images to generate.

These representations of faces have something rather harrowing to them since they're generated by artificial intelligence. The length of time for these portraits to be made surprised be, we seem to have this idea that things that happen digitally happen instantly and don't take much time.

- Using algorithms to generate imagery, not something we have the facilities to do but the processes behind developing these portraits are really exciting, they showcase the power of digital media. 

 

The Datasets

We all know that the photos we post and the texts that we write on social media and other online platforms are being used to collect information about us - information that is then sold to everyone from advertisers to insurance companies to banks and credit agencies.

The works in this section are all composed of “training data” that is given to machine learning systems in order to “teach” those systems how to evaluate human behaviors.

- This mixing of digital data and real life people is really exciting, demonstrating how they are interconnected is something I want to bring in to the project more and this was something I found to be different to my original thoughts about the closeness of digital and physical. Initially I thought they were very separate but throughout this project I've learnt they're much more interconnected than I realised.

Airlines and Sentiments

These texts are culled from datasets that artificial intelligence developers use to try to teach computer systems how to recognize the content and overall sentiments of online communications. The theory is that, for example, if you give an AI enough examples of the difference between “positive” and “negative” feelings expressed in texts, that the AI will be able to discern the difference.

The ability to autonomously measure sentiment and other aspects of content on platforms like Facebook and Twitter is a major goal of AI research, with applications in everything from content moderation to user profiling to high-frequency stock trading.

This idea of representing these large, intangible concepts with visuals is what's most interesting (to me) about these works, they show something that we can't comprehend through a visual. The simplicity of the execution allows the concept to be at the forefront. 
-When working on our publication make sure to embody the most appropriate ways of communicating themes, not everything needs to be busy to communicate something complex (as demonstrated here).

----------------------------------------------------------------------------------------------------------------

Google maps routes:

Took a random route on google map street view, going along the different roads and screenshotting the whole way.

Did this numerous times and then placed in indesign:

Not sure how I feel about the design, something about it feels very flat, not sure if it's because of the imagery being of roads. Wanted to use a route that I has walked today but couldn't access the park, was only able to complete this on roads.

- Limits the outcome, also is this too much like Trevor Paglen's work? Yes. Would need to have a purpose or different way of presenting the theme to use this. 

Photo everyday:

Think about mark farid's talk and how he published his own content to the web, and the original idea behind the project of publishing personal content tot see how it would make us feel.

Could relate to surveillance also, running along the idea of tracking/monitoring.

Mixed feelings about this one. 

Peg's comment: I think it's a nice idea but I don't think it quite makes sense with the idea of surveillance as it's not like public data/you're voluntarily giving up this information if you get me?

I feel as though the original purpose for  the project has been detached slightly and now since we're exploring more abstract themes with idea isn't the most relevant, could be reconfigured so will keep on standby for now. 

Working on this spread:

This spread is running off of the idea of tangents and rabbit holes (based off of the pop up when visually similar image search).

Imagery in this came from looking at the hyperlinked text that was explaining what a Glaciokarst is. 

Took this further and saves all the images, not just one:

There are 55 images in total.

What to do to organise them, currently displayed in rows of content.

- Could look into image organisation through recognision?

This looks much too organised for the theme of a rabbit hole, needs to be more sporadic and visually represent the process I went through of opening tabs, closing, reopening, saving, pop ups etc. 

Using the random number generator to decide the size and placement of the images.

Really liking this one, feels much more representative of what it takes to get the images as well as have the idea of a rabbit hole, sense of loss of control, losing time also. 

- Feel this could be a good break between pages that contain a lot of text and is less structured than some of the other pages that present imagery. 

Secondary spread of this imagery:

My Spreads so far:

Reflections:

• The spreads are coming together nicely, they're getting a visual language to them. Thinking about the different sections do they want to work together or be visually different as sections? 
• I'm not sure how we'd communicate the different sections without making changes to type or colour palette. Want to talk to Peg about this more and develop a system to follow. Personally I think if there isn't a clear visual difference that's okay as the "themes" we're exploring all flow into one another anyway.
  - Could think about type and having specific type for specific sections and then when there's overlap multiple typefaces are use.
  - Could alter the background colour, of the page?