"Why break when you can divide it"
hiddencomplexity.substack.com
Certain topics are rather timely, and no other topic fits the times than the topic of “The Dead Internet Theory”. First, as usual, a bit of history is necessary since this is a very niche part of net history. Even if at a superficial level, understanding Imageboard's self-contained culture, and its impact on culture itself is necessary. There a significant reasons why each and every single Large Language Model is trained on a corpus of scrapped data from 4chan. Terabyte worth of data.
From a descriptive perspective, Imageboards such as 4chan have been at the forefront of culture for the most part since their inception. The influence of specific boards
(imagine a smaller section of the website centered around specific themes) has always been influential and disproportionate to its size, a true lesson is cascade effects in complex networks. How a small, secluded node of a system
“affects” the rest. No wonder the UN, think tanks and even the DoD poured a few hundred million dollars (total) since 2015 into
“researching” the boards.
It would take months to dissect each common, pop culture, or even niche
“belief” that arose from image boards, but one in particular bears more weight today, than at the time—the Dead Internet Theory. I have the original post here, somewhere, but given how extensive my Imageboard dataset is, there is a chance close to 0 of me finding it. The original post was many years older than the following one, but they are close.
The text is loaded with Imageboard vernacular, and with the conspiracies that often are common among the crowd, but the brunt of the hypothesis is there. Ignore the conspiratorial part, focus on the prescient nature of the text. At the time of the original post, before 2014, I thought that was far-fetched, we were not even close to that level of capability, and we didn’t have the compute.
For quantifiability, a GPU today is as powerful as a 15 million dollar supercomputer of 2015, the disparity in computing power is hard to grasp sometimes. So what could explain the discrepancies in content quality, or what we call what is, or feels organic (made by humans), and what it is not (synthetic) ?