My two cents on the CARDS vs ARDS debate. Distributions of compliance and P/F ratio in the SUPERNOVA study show that compliance > 30 ml/cm H2O not uncommon in moderate ARDS. P/F ratio and compliance were only weakly correlated (R2=0.05).
https://www.ncbi.nlm.nih.gov/pubmed/31432216 

So ARDS patients can be hypoxemic without low compliance; and ARDS patients can have low compliance without significant hypoxemia. Whether this has implications for tidal volume management is highly controversial.

What "feels" unusual in these patients is the profound PEEP responsiveness (oxygenation improvement) in the absence of mechanical lung recruitment, as described by @respresource and others. I am less certain this is attributable to hemodynamics, since shock is minimal or absent.

I would speculate that endotheliopathy, loss of hypoxic pulmonary vasoconstriction, and possibly pulmonary thromboemboli play a larger role than usual in driving hypoxemia in CARDS in comparison to ARDS. This remains to be proven.

So what does this mean for management of hypoxemia? Gattinoni and others advise against increased PEEP in these patients in the absence of lung recruitability. Proning should unquestionably be the first step, but is not always successful in ameliorating hypoxemia.

I would suggest that increases in PEEP to the minimum level necesssary to achieve adequate oxygenation (SpO2 88-90%) are reasonable in these patients, even when compliance is high and recruitability is low, provided that plateau and driving pressure limits are not violated.