Catching your 'subconscious' with mathematics.
This diagnosis is not just a personality test or fortune telling. It is an interactive search engine that applies probability statistics and information theory to keep calculating the shortest route to the 'ideal color' in your brain.
PC and smartphone screens can theoretically display 16.77 million colors. Finding the one single color you are looking for from this vast sea of colors is like looking for a ring dropped in the Pacific Ocean. You won't find it by searching blindly.
So, this system placed 263 'observation points (candidate colors)' at equal intervals on the entire map called color space. These are like GPS satellites or buoys floating in the sea.
Importantly, these 263 colors are not 'answer candidates (choices)'. They are merely sensors to sense 'around where' your preference lies. Since the final diagnosis result is calculated by mathematically interpolating (filling) between these points, even colors located where there are no buoys can be identified with near-infinite smoothness.
Note: Why 263?
Theoretically, more points are placed (496), but as a result of strictly excluding 'areas where colors are crushed' that are visible to the human eye but cannot be reproduced on general monitors (sRGB), these 263 elites survived. If display performance improves around the world in the future, the number of these guideposts is designed to increase automatically.
The moment diagnosis begins, the system has no idea which color you like. In other words, the 'probability of liking' is spread thin equally for all colors. This is like the entire color space being covered in a 'fog of possibility'.
Here, the magic of probability called 'Bayesian Inference' appears. When you answer 'I like color A', the system calculates 'Checks are high that colors close to A are liked' and 'Checks are high that colors close to B are disliked', and updates the density of the fog (update of posterior probability).
Repeating this, the fog that was pure white at first gradually clears, and only one point begins to shine strongly. This is the process of identifying your preference. By taking the approach of 'narrowing down possibilities' rather than 'choosing the correct answer', identification with a small number of questions is made possible.
Since we don't know where the correct answer is at first, the probability of all points is the same (uniform distribution).
As questions are repeated, only the probability density of a specific area (preferred color) increases, and others disappear.
The system does not choose 2 colors properly every time. From thousands of combinations, it calculates and asks the 'Question that yields the most information (High Information Gain)'.
For example, suppose you are asked 'Which do you like, Red or Blue?' and you answer 'Red'. If the system predicted 'You probably like Red' from the beginning, almost no new information is obtained from this question. Conversely, when 'two colors that are completely unpredictable 50-50' are presented, great value (amount of information) is born in your choice.
In technical terms, the system avoids 'Questions that maximize entropy (uncertainty)' and looks for 'Questions that reduce entropy the most'. If a question comes up where you hesitate like 'I like both...' or 'Both are subtle...', that is proof that the algorithm is approaching your core. That 'hesitation' decision dramatically advances the diagnosis.