Sonata form is relatively predictable when it has an exposition: you clearly distinguish the two main themes and you will perfectly understand what is being developed in the development. Furthermore, the recapitulation is also relatively predictable because you have heard the exposition before and both have certain parallelisms that, even in Romantic sonatas, will always guide the listener to the coda.
I thought that if you got a piano sonata, even a Haydn one, and you started with the development instead of the exposition, effectively getting rid of that section, it would be extremely surprising. Take Mozart's Sonata no. 16 in C Major and start with the development in the first movement. While in this case the development is very short, you still fell very confused if you had never listened to the exposition before.
If a development-recapitulation form was made in sufficiently large dimensions (otherwise it would be too short to be a movement), together with an increase in the duration of the development, you'd get an extremely innovative tonal form. The listener would perceive some chaotic elaboration of motives that he had not heard before and, when the recapitulation begins, he understands everything: the order is uncovered. From apparent chaos to perfect order.
Has any composer done this (statistically, someone has done it)?