Author: Brian Arechiga
Language: English
Year/s of development: Preliminary Findings, hypothesis

Hello, I am currently a 5th year USC student working on my dissertation which will cover conspiracies and the digital platforms that create and disseminate them. 4chan is of particular importance to me since it played a large role in conspiracies like Pizzagate and QAnon. In the following, I offer a quick digital anthropological history of the 2025 leak of 4chan's source code along with a preliminary reading of my approach to the code. Here, I analyze the 4chan 'bump' system through the thermodynamic and informatic concept of entropy.

Intro - Brief History on the 4chan Leak

On April 15, 2025, hackers successfully gained administrator access to the infamous imageboard, 4chan.org. Founded in 2003, 4chan has gained a reputation for housing some of the most irreverent, influential, and despicable communities on the net. Originally a port of a Japanese open-source imageboard 2channel, 4chan's codebase, named Yotsuba, has long been a proprietary piece of software -- that is, until the soyjaks came around.

What is a Soyjak?

The work soyjak stems from a combination of the word 'soy' and 'wojack.' Wojack is a popular meme, also referred to as 'thefeelsguy.png'.

(Wojak)

The word 'soy', in this context, has its roots in the misogynistic world-view of manosphere and alphamale digital culture. A 'soy boy' is a derogatory term for men who do not prescribe to stereotypical ideals of masculinity. 'Soy' has become the word of choice based on the false idea that soy-based drinks introduce estrogen into the body, which weakens a male's testosterone. The Soyjak meme is a man drawn in the style of wojak who posses interests and traits contrary to masculine stereotypes. He appears as follows:

For years the soyjak community found its home on 4chan's /qa/ board. The /qa/ board was originally a questions and answers forum on 4chan intended for meta-discussion about the website. However, over the years, the posts on /qa/ strayed far from its original intent and the board turned into a large community with most users posting off-topic soyjak memes. See, this image of a 4chan archive detailing a few random posts from 2021:

Due to the off-topic posts, 4chan admins decided to shut down the /qa/ board in 2021, drawing the ire of the soyjak community. In response, a rival imageboard site soyjak.st grew in popularly with many /qa/ users migrating there. Since the soyjaks diaspora, many of its users held animosity towards 4chan for shutting down their original home. These tensions climaxed in 2025, when a soyjak.st user by the name of, Cirrus, posted a new thread with an image of 4chan's administrator view.

Cirrus had successfully hacked 4chan. With admin privileges, the first order of business for Cirrus was to re-opened the /qa/ board. It was active for several hours before 4chan admins regained access to the system and shut it down again.

(Image from a 4chan archive. The final posts from /qa/'s 2025 revival before 4chan admins realized they were hacked and shut down the board again)

While all of this digital-community infighting is of great interest to the digital historian, perhaps the most interesting development of this debacle is what happened to 4chan's code. See, Cirrus and his team did not only reopen /qa/, but while they had administrator access they also downloaded 4chan's code and released it to the public. Now that the Yotsuba code is available, I aim to study it for a future chapter of my dissertation which deals with conspiracy theories, online forums, and informational entropy. Here, I will present a preliminary reading of the code that I hope to develop into a future chapter of my dissertation.

Code Reading: The Bump

One of the most unique aspects of 4chan is how threads are displayed to the user. 4chan contains many boards with numerous topics ranging from weapons to anime. Each unique board has 10 pages. When one visits a board, they are taken to its first page which displays the ten threads with the latest replies. Members of 4chan refer to a reply as a 'bump' since each reply 'bumps' the thread back to the top of the board's first page. There is also a self-purging system in place. Bumps are so valuable because a thread gets deleted from the website if no one bumps it. Each time a new thread is created, the last thread on the last page is removed in order to make room for the new one. Therefore, if a thread receives no responses, it slowly makes its way down from page one to page ten of the board until it is permanently deleted from the servers. If one attempts to visit a URL from an expired thread, they are met with a 404 error.

I would like to study the bump system further because it creates a unique dynamic between information the site houses and the means in which the user interacts with it. There are two main ways the 4chan system is unique from other forums or social media websites:

1) the temporary nature of information -- born from thread's auto-deletion and movement speed
2) the organization of threads -- from latest activity (new posts and replies)

Since the 'bump' algorithm is the system which dictates these unique characteristics, it will be the main aspect of the code the rest of this post will focus on.

Bump Code Explanation

The bump algorithm is a HTML front-end and back-end (PHP and SQL) interaction that takes user-data into a server-side database. Once the database is updated, the PHP code then rewrites the HTML of the site to place the threads in the correct order. The function that initiates the bump logic is called: regist. In order for this function to be initiated a user must submit a post through the form on the front-end of the site. The reply is then ran through multiple lines of code until the bump is allowed:

if( $resto ) { //sage or age action

$resline = mysql_board_call( "select count(no) from `" . SQLLOG . "` where archived=0 and resto=" . $resto );

$countres = mysql_result( $resline, 0, 0 );

$permasage_hours = (int)PERMASAGE_HOURS;

if ($permasage_hours > 0) {

$time_col = 'time,';

}

else {

$time_col = '';

}

// FIXME: a similar query is done at line ~4723

$resline = mysql_board_call( "select {$time_col}sticky,permasage,permaage,root from `" . SQLLOG . "` where no=" . $resto );

$resline = mysql_fetch_assoc($resline);

if ($resline['sticky'] || $resline['permasage']) {

$root_col = '';

}

else if ($resline['permaage']) {

$root_col = 'root=now(),';

}

else if ($is_sage || $countres >= MAX_RES) {

$root_col = '';

}

else if ($permasage_hours && ($time - ($permasage_hours * 3600) >= $resline['time'])) {

$root_col = '';

}

else {

$root_col = 'root=now(),';

if (!$captcha_bypass && BOARD_DIR === 'jp') {

if (!spam_filter_can_bump_thread($resline['root'])) {

$root_col = '';

$_bot_headers = spam_filter_format_http_headers($com, $country, "$insfile$ext", $_threat_score, $_req_sig);

log_spam_filter_trigger('necrobump', BOARD_DIR, $resto, $host, 1, $_bot_headers);

}

}

}

mysql_board_call("update `" . SQLLOG . "` set {$root_col}last_modified=%d where no=%d", $_SERVER['REQUEST_TIME'], $resto);

}

The first conditional statement above leads a reply ('resto') through a series of if-statements, which results in a change of the variable '$root_col.' If is left empty, $root_col = ' ', then the reply will not bump the thread; if $root_col='now,' then the thread is bumped to the top of the board.

The actual bump executes in this line of code:

mysql_board_call("update `" . SQLLOG . "` set {$root_col}last_modified=%d where no=%d", $_SERVER['REQUEST_TIME'], $resto);

Reading Entropy into the Bump Algorithm

I would like to offer a preliminary reading that interprets the 'bump' algorithm as an analogy for entropic processes. Entropy is a concept that spans multiple disciplines such as thermodynamics, probability statistics, and information theory. I will attempt to give a short summary of the aspects of this complex concept that are relevant to my reading.

Delaying Decay

In the thermodynamic sense, entropy was first studied to explain the natural flow of heat from hot objects to cold. Its definition grew to account for most irreversible processes of the physical world, such as heat dissipation or decay of molecular cells over time. Later, scientists, such as Edwin Schrödinger, began to see entropy and the flow of time as heavily related. He writes, "The statistical theory of heat must be allowed to decided by itself high-handedly, by its own definition, in which direction time flows [...] To my view the 'statistical theory of time' has an even stronger bearing on the philosophy of time than the theory of relativity." (Schrödinger, What is Life? 152).

In many ways, 4chan's code mirrors the physical properties of entropy. For starters, the board flows in an irreversible direction, creating an environment where threads can grow stagnant and decay until they perish. The logic of the the bump algorithm creates an arbitrary 'flow' of information with its "$root_col" variable. Recall, if a thread is bumped then $root_col= 'now,' placing the thread to the top of the board. Therefore, the more conversation generated by a post or, in other words, the more 'information' users post, the more 'life' the post receives. Leon Brillouin theorizes information as a negative entropic force. He quantifies information (I) as a negative force towards entropy (S), he called negentropy (N). He writes this as

'I = -S'
or
'N = -S'
(Brillouin, Science and Information Theory, 116).

Since each reply is more information added to the website, the life or death nature of a thread is dictated by how much negentropy it can gain from the users. The more replies (information) a post receives the more negentropy it gains, slowing a threads decay and erasure. This process results in a type of informational Darwinism, survival of liveliest.

A Snapshot in Time

While information continually flows towards its death on the site, the way people interface with the site also enables one to cheat entropy for a instant. When one visits the site, the web-interface grants the ability to essentially freeze the 'now.' If the front page of a board was constantly updating in real time, then fast boards, like /b/ - random, would be impossible to read. However, loading a board on a browser presents the user with a static page with the most active threads according to the moment one visited the page. It is a frozen moment of time, a single instance of the 'now.' As time passes, the static page one browses becomes out of date since thousands of users continually bump threads in the background. It is only when one refreshes the page that they receive an updated instance of 'now,' which again is quickly outdated. As a result, 4chan's database and organization is constantly in a state of flux.

The flowing nature of 4chan's information might led one to interpret it is as a chaotic system and, therefore, a highly entropic system; however, this is far from the case. A common misunderstanding of entropy relates it to a measure of disorder and chaos. In this view, order is often seen as the counter to entropy, which has led many to think that highly entropic systems are those that are disordered. Therefore, it would be natural to assume that 4chan is a highly chaotic place, but this is a misconception that fails to account for the statistical view of entropy.

4chan's Low Informational Entropy

To clear up this misunderstanding one must turn to the study of macro and micro-states in entropic theory. Ludwig Boltzmann was the first to try and understand entropy from a molecular level which led to his development of the micro/macro entropic framework. Macrostates are the qualities of a system one can observe and measure, such as temperature, volume, pressure, etc. Microstates are variables that are unobservable by scientific instruments at the time: atomic position, and spin to name a few. For this point of view, entropy does not relate to how messy or disordered something is; it cares more about the number of microstates. A highly entropic macrostate is one that can be accounted for through numerous microstate configurations, making it the most probable, occurring when forces naturally flow to an equilibrium. For instance, a hot cup of coffee eventually reaching a temperature equilibrium with room temperature. Low entropic macrostates are systems which have fewer microstates that can account for them (i.e. the cup of coffee before it reaches thermal equilibrium with the room). We can relate this logic to the organization principle of 4chan's bump algorithm.

Here, I created a table to better exemplify how this theory applies to my reading of the bump algorithm:

Therefore, an imageboard with low entropy points towards a healthy flow of information. One cannot predict what will be on the front page since it is a highly random organizing principle. The interests of the user-base and their attention animates the flow of information. In an age of algorithm centered consumption cycles, this system is wholly unique and results in a highly innovative and iterative culture.

Closing Words on Iterative Culture

Lastly, the cycle of life and death is an important part of the development of the site's culture due to the iterative engagement with information the system imposes on its users. An important aspect of the bump algorithm is the time between the deletion and re-posting of information (memes, ideas, etc.). Borrowing from Leon Brillouin, we can think of this as the time between the bounding and freeing of information. He writes,

We now wish to distinguish between two classes of information:

1. Free information I(f), which occurs when the possible cases are regarded as abstract and have no specified physical significance.

2. Bound information I(b), which occurs when the possible cases can be interpreted as complexions of a physical system.
   (Brillouin, Science and Information Theory, 152)

When information is posted on the site in the form of language or images, then it is bound information. After it is deleted, the idea becomes free information in the heads of the users. The entropic flow of information on 4chan creates a constant transference of free and bound information within the culture. The longer information is bound the less iterations and work is being done on it. When information is free, the human conscious does 'work' to it by thinking and remixing. For instance, due to the ephemeral nature of posts, users are encouraged to save memes and repost them. During the time between these actions, the ideas are free and abstract in the users head. More often than not, the ideas are re-posted in a new context or some users even go out of their way to edit the images and remix them. It is, then, bound again when it is posted and the cycle continues. This process is how boards develop their unique culture and how ideas are processed.

Welp, that's about all I have at the moment. In the future, I hope to investigate this process further and connect it to the ways in which conspiracy theories are created and disseminated on 4chan. I would also like to get a more concrete idea of what exactly would constitute the 'microstate' of the imageboard code in this analogy. I plan on continuing my study of this code in my dissertation. Would love to hear any thoughts, musings, or even 'further reading' suggestions for me.

Thanks for reading,
Brian