Modeling Value in the Anthropocene has regressed into a bit of entropy this week, as Brian’s immune system has experienced plague-induced degeneration and Hampton has come up against the structural incompleteness of his mind and his reality. Brian was unfortunately checked out for a week and a half due to a bout with COVID-19 while Hampton continued to dig further into word2vec and word embedding.
We have gotten to a point where our text is cleaned (with the cleaned version being stored locally on our machines), and we have successfully been able to play around with a word2vec model that has been pre-trained on a large corpus of text from Wikipedia. The end-goal we are working toward involves performing calculations on certain vectors from this pre-trained model with vectors from a model that we want to train on our Stiegler text. Now that we have successfully used the pre-train model, the next step for us is to train the model on our text. The problem that Hampton has been confronted with is a simple one, which is that of merely loading our cleaned text into our programming environment to begin to do so. This is where we run into problems as pirate digital humanists without a real background in computer science. However, we are very fortunate to be surrounded by a wealth of human beings with more experience than us who are willing to assist us. We have a second meeting with Leanne from the Digital Fellows Friday to figure out how to import our corpus and get back to work.
This situation (both Brian’s falling ill and Hampton’s Python woes) have also been a valuable lesson in two important Stieglerian concepts, adoption and quasi-causality (which he adopts from Gilles Deleuze and Felix Guattari). For Stiegler, following Heidegger, we are thrown into a world which has the character of being already-there, a world which pre-exists us and world over which we do not have control. We learn from Stiegler that the responsibility of the non-inhuman is to adopt this situation (of our lack of mastery over being) as necessary, and in so doing to make it quasi-causal for us. We do not have control over the initial blows that the world deals to us, but we do have the capacity to channel the libidinal energy that is generated as a result and make the situation the preindividual funds for a future-to-come. Working with Python is transforming the way we view failure, revealing its function as a departure point for progress.
Spoiler alert: I am uploading this a day late, so we have already met with Leanne, with whom we had the most productive meeting possible. She assisted us with getting our corpus into our Jupyter notebook and has brought us to the point where our model is now trained on our corpus. The next step will be figuring out what calculations we need to perform on our corpora in order to produce a new vector for value in the Neganthropocene and relate this vector to other vectors in the Stiegler text and the Wikipedia corpus. We hope to shed some light on these calculations for next week, our week thus amounting to a lot of Google-assisted programming. This will be weaved through with our continuing close reading of the lectures, which continue to stimulate us emotionally and intellectually.