One of the possible applications of the Y-combinator

Idea. And let's send the Y-combinator to the input ... not the Y-combinator itself, but almost Y. Y-stroke, not even a combinator, but simply a function that has no way out of recursion.

I know that they would do it 100 times, and nothing good would come of except STACK_OVERFLOW_EXCEPTION.

Let it take a state as input, enter and output information, and send the state further.

Y functor data = functor (\ newdata -> Y functor newdata) data
Yscratch callback state =
    // cross knowledge with input
    let new-state = intersect state (input ()) in
        output state
        // simplify knowledge if possible
        callback (reduce state);

and now

wtf iIdiot =
    Y Yscratch iIdiot;

What is this nafig? This is an imitation of consciousness. If state is an analog of a neural network, then direct imitation.

But who needs this?