The CAP (Consistency, Availability, Partition tolerance) also known as Brewer's theorem, states that it is impossible, on a distributed computer system, to guarantee at the same time the three qualities mentioned above. This issue affects all existing distributed engines and systems, which forces the administrators of these engines and systems to favor two qualities to the detriment of a third, when possible, despite the consequences. This paper technically and formally describes an inventive algorithm to locally weaken the said theorem when reading the data. This process combines three elements: a real-time system of key-value pairs distributed in memory cloud; a multi-planar index with concurrency avoidance; a distributed consolidated reading method called "LC4" based on a hysterical retroactive probabilistic effect. This inventive algorithm was implemented and tested successfully, jointly with a second invention playing the role of an application delivery controller (“A heuristic algorithm imitating social behaviours for intelligent routing of application flows” ⟨hal-02471814⟩ https://hal.archives-ouvertes.fr/hal-02471814). Full understanding of this paper requires expertise in computer science, software engineering and algorithmics.