I kind of like the association since it speaks to how text collected while browsing the web can be used to generate new text, which is similar, at least metaphorically, to how human memory is reconstructive and transformative, not perfect recall. https://en.wikipedia.org/wiki/Reconstructive_memory

Also check out https://archiveweb.page which is open source, local, and lets you export archived data as WARC (ISO 28500). You can embed archives in web pages using their Web Component https://replayweb.page.

Does anyone know what PrivateGPT is using for its local model, and where it came from?

Update:

Answering my own question it looks like it uses llamacpp in local mode? https://github.com/imartinez/privateGPT/blob/main/private_gp...

In her book ‘Discriminating Data‘ (2021), Wendy Chun reveals how polarization is a goal — not an error — within big data and machine learning. These methods, she argues, encode segregation, eugenics, and identity politics through their default assumptions and conditions. Hito Steyerl and Wendy Chun will discuss how can people release themselves from the vice-like grip of discriminatory data and consider alternative algorithms, defaults, and interdisciplinary coalitions in order to desegregate networks.

If you work as a researcher you might be able to apply for Academic Research Product Track access, which gives you access to the full archive of tweets back to 2006.

Is there a good listing of these, or could we crowdsource one quiclky to make sure they are at Internet Archive?

Not exhaustive but https://waxy.org/2018/11/a-tribute-to-youtube-annotations/

At least today we have the investigative reporting from the Intercept. It's kind of crazy the lengths to which citizens had to go through to get the word out about COINTELPRO https://en.wikipedia.org/wiki/Citizens'_Commission_to_Invest...

This is a translation of an interview Sepulveda did in Spanish. The description of the technology and resources used provides an important glimpse at how public opinion is being manipulated in social media.

Check out the volume of negative comments...

If you can identify the IP ranges you can configure anon https://github.com/edsu/anon to listen for edits from them, and announce them using a particular Twitter account.

Honestly, I just put anon together in a few hours, and didn't put a great deal of thought into it. It might make sense to do what you suggest, and create a more generic webapp to deploy on Wikimedia labs, which manages the IP ranges, and Twitter accounts to post to.