Entropy

Entropy is a measure of uncertainty. 

When considering a source consisting of a specific set of symbols, the more unexpected (i.e., low probability) a message from that source is, the more information it carries. If we consider the opposite of this, pure entropy—that is, complete randomness or the total absence of order—would mean that no information is actually transmitted.

If the content of a message is highly predictable (for example, a source that always sends the message "A"), this message does not tell the recipient, meaning you, anything new; therefore, it contains no information. Every incoming message is already known. Entropy is low here.

On the other hand, if the content of the message is full of rare or unexpected symbols, this means new information for you. This message carries high information value because it is unexpected. However, as long as this is just a singular event, the overall entropy of the system does not increase. This is just a momentary increase in information, not an increase in entropy.

However, if the message is a completely random and ruleless sequence of symbols (for example, piles of meaningless letters), this also does not carry useful information. Because there is no meaningful structure or pattern within this randomness. In this case, entropy is at a maximum level, but there is no meaningful information. But extremely high entropy, that is, a completely random and meaningless stream of messages (for example, “xy!@#q”), also produces noise, not information. In other words, to produce "information," meaningful surprise is required.

Thinking in terms of social sciences, in a media environment where total censorship is applied, there is low entropy. That is, the news is highly predictable. There is little information. A social media environment where everyone produces random content contains very high entropy. It becomes difficult to make sense of the incoming messages. This situation leads to chaos, not information. For optimal information flow, moderate entropy is good. In other words, a little order, a little surprise.

For example, the concept of the "filter bubble" we encounter on social media platforms describes the narrow field of view created by algorithms filtering content according to users' interests. Here, entropy is artificially lowered by the platform.

These platforms operate on the logic of showing you what you watch. Thus, the algorithm causes the content to become predictable and repetitive. This offers the user a safe and familiar experience, but at the same time, it reduces entropy, and therefore the information value obtained from the system also decreases. A paradox arises here: users leave the application when they get bored. So, the algorithm needs to add a little surprise.

This is where "entropy management" comes into play. TikTok and, to some extent, YouTube Shorts, in particular, adjust their algorithms according to this principle: "Show the user 70% of the content they like, but give them 30% random, unpredictable content."

This awakens the user's sense of discovery (i.e., receiving new information). Thus, the algorithm increases entropy in a controlled manner. However, this increase is not done with random content, but with pre-tested surprise content. That is, it happens in the form of managed surprises, not complete chaos, and an attempt is made to keep the attention level high. This situation evokes C. Shannon's distinction between "noise" and "information" in communication theory. If there is surprise, there is information, but if the surprise is detached from its meaning (context), it becomes "noise."

On the other hand, entropy affects not only the individual experience but also the ways in which society accesses information. When there is extremely low entropy, that is, if you see and hear the same things in the media all the time, as in censored media environments, this causes individuals to receive only one-sided and predictable content. This creates "intellectual barrenness" in society. When there is extremely high entropy, for example, the abundance of disinformation on social media leads to results such as individuals no longer being able to trust any information.

Here, what is aimed for an ideal media environment is moderate-level entropy. That is, a structure that offers enough difference to take the user out of their comfort zone intellectually without surprising them too much.