Weakness in currently published ++æ versions

Lear Barack identified what seems to be the most significant weakness in ++æ v1.0 and v1.1: when the plaintext contains long runs of 1s then the inner vectors Ii / Os crystallize and just after a very short series of blocks injecting those 1s in the plaintext  (only 4 blocks in fact) the series become periodic allowing, basically, cryptogram forgery by simply removing the blocks comprised between the repetitions.

Thanks Lear for your interest and analysis! Could it drive ++æ several floors upstairs ?

Success  usually is a momentary state between long failure runs (that fortunately can be socially extended by the obtained status-quo   ;o)

Therefore, since success is now closer, I'm exploring simpler forms of addition-and-xor sums combinations that could be simpler to be analyzed, stronger in front of practical and theoretical attacks and ... even less computing demanding. A truly challenge and a truly fun game (at least while the underneath failure does not become evident ...  ;o)))

If I followed my impulses, I would post now what I have in mind as a final ++æ version, but this time I want to be more cautious and I'll wait a sound security analysis is completed before releasing any detail. For that purpose, collaborators providing independent critical reviews would be really helpful and appreciated (specially if the review comes in the form of a definite ++æ security proof ...   ;o)

(to be continued ... or not ...)