They saw functionally no change in voter preferences from the debate.
I know this sounds ridiculous, but the debate freakout is happening among Dem partisans. Most voters already had those perceptions of Biden. Dem partisans are still supporting the Democratic candidate.
I think Harris would have higher upside at this point, but can we talk about polling like rational human beings for cripes' sake? Polling has error.
That's why these resurveys are so good. They can attenuate the sampling error that you get from just comparing pre and post debate polls that sampled different groups.
The polling movement overall has been within the margin of error. So it could entirely be sampling error and not real. That's what this resurvey suggests.
And like I said, Andrew Gelman showed us to be wary of post-debate poll movement in presidential elections. In 2012, Obama's ostensible decline in the polls after his bad debate was entirely from partisan non-response bias. It was a mirage from sampling error.