These notes accompany the project EVERYTHING. They are not required to experience the work itself, but provide additional context about its mathematical foundation and its unusual development process.
At first glance, the matrix appears to be nothing more than random noise. Every second reveals a new arrangement of colored pixels, seemingly unrelated to the one before it. In reality, every image occupies a precisely defined position within a complete mathematical ordering of all possible images.
Everything begins with a single value: the number of seconds that have elapsed since the project started on June 16th, 2026, at 20:45:23 UTC.
Nothing more is required.
No images are stored. No database is queried. No previous states are remembered. At any moment, the current image can be reconstructed entirely from this single number.
Unlike conventional approaches, the timestamp is not compressed into a fixed-size hash. Instead, it is treated as an integer that can grow without any predefined limit as time progresses.
This avoids introducing an artificial ceiling on the number of distinct images that can ever be generated.
The 160 × 144 matrix contains 23,040 pixels. Since every pixel has four possible brightness levels, the complete image consists of exactly 46,080 bits of information.
The elapsed time therefore determines one unique state within this 46,080-bit space.
The resulting state is then transformed by a reversible Feistel permutation. Its purpose is not to create randomness, but to rearrange the complete image space into a sequence that appears chaotic while remaining perfectly deterministic.
Feistel networks have been used in cryptography for decades because they allow information to be mixed extremely thoroughly while remaining perfectly reversible.
Because the transformation is bijective, every state has exactly one predecessor and one successor. No information is discarded and no two states can become the same.
This property has an important consequence: the algorithm does not search through possible images. It defines an ordering in which every mathematically possible image appears exactly once.
No image is skipped. No image is repeated. The complete image space is traversed without collisions.
If the same timestamp is entered again—even decades later—the algorithm recreates precisely the same image, pixel for pixel. Every image is therefore a permanent mathematical object identified by its position in the sequence.
One consequence of this design is easily overlooked: the project contains no collection of images.
Instead, it stores only the mathematical procedure that defines their order. Every image exists only while it is being calculated and displayed.
The project demonstrates that extraordinary complexity can emerge from remarkably simple rules.
A single integer becomes a complete mathematical state.
That state becomes one unique image.
Every second advances exactly one position further through the finite universe of all possible images.
Nothing is remembered.
Nothing is repeated.
Only time moves forward, revealing one unique image at a time.
EVERYTHING is not only an experiment in mathematics. It is also an experiment in creation.
The project began with a simple question: Can a technically demanding website be conceived, designed and implemented by someone who writes no code, but instead develops every aspect through conversation with an artificial intelligence?
Every visual decision, every mathematical refinement, every structural change and every conceptual adjustment emerged from dialogue. Rather than programming line by line, the process consisted of describing ideas, evaluating results, identifying imperfections, reformulating intentions and gradually approaching the desired outcome.
Contrary to popular imagination, this process is neither automatic nor effortless. Artificial intelligence does not eliminate creative work; it changes its nature. Progress depends less on programming syntax than on the ability to formulate ideas precisely, recognize subtle inconsistencies and refine a vision through continuous iteration.
In this sense, the creator becomes less a programmer than an architect of intent. The code is generated through AI, but the conceptual framework, the aesthetic direction and the countless decisions that shape the final work remain fundamentally human.
Whether this approach should ultimately be called vibe coding is almost beside the point. The project investigates something more general: whether modern artificial intelligence can lower the technical barrier to software creation without reducing the importance of imagination, critical thinking, patience and persistence.
The process was far from linear. Ideas were proposed, tested, discarded, reconsidered and refined over hundreds of iterations. Some solutions appeared elegant at first, only to reveal subtle conceptual flaws later. Others emerged only after long discussions about mathematics, visual perception or philosophy. The website itself therefore became part of the experiment.
The collaboration also demonstrated an important limitation. Artificial intelligence can generate code remarkably quickly, but it cannot decide what ought to exist. Direction, taste, judgement and conceptual coherence still originate from the human participant. The machine contributes implementation; the project contributes intention.
EVERYTHING therefore explores two finite spaces simultaneously. The first is the combinatorial universe of all possible images. The second is the emerging creative space between human intention and machine implementation. Both are vast. Both remain largely unexplored.