It’s foolish to evaluate the outcomes of text simplification endeavors without looking at the intended target audience of the simplified works.
While we can approximate objective evaluation metrics, there is no ONE way to simplify a text which means that results are highly subjective.
Although not many studies include surveys where target audiences were asked to rate the results or experiments which seek to infer readability from eye movements, reaction times or similar, this work is prudent and may lead to insights on which measures need to be taken to provide appropriate accomodations.
(It could be argued that this is a case of Computational Linguistics forgetting about their Linguistics background. Psychology is for everyone! SCNR)