From what I can tell the GOP was particularly weak at recognizing underlying bias in some neutral mathematical models. In AZ the GOP assumed that the blame was solely on the partisan chair, but even a neutral chair would have had a hard time overcoming the state competitiveness directive given the bias in the chosen elections.
In CA my sense was that the lack of understanding by the GOP of the impact of socioeconomic grouping as a preferred community of interest. The underlying math here works against the GOP as much as a maximally square grid with minimum area districts works against the Dems.
In Arizona, the game was over the minute the Democratic mapping firm was picked. Inevitably, there were in a position to equate "strongly leans Democratic" with "competitive." That is why Mathis went to great lengths to intimidate/bribe the Republican commissioners into going along with a unanimous vote to select the Democratic firm. She wanted the inevitable Democratic gerrymander to have a veneer of "bipartisanship."