Seems very plausible based on the 2020-2021 population changes, at which NC was growing faster than GA. Will be a very close-run thing either way, though.
I'm a bit confused about their 2021 estimates though because they seem to be in line with their 2020 estimates rather than the true 2020 numbers. For 2021 numbers, they have states like NY and RI (which outperformed in the census) doing rlly badly), whereas underperformers like NC and AZ did really well.
It's well established that the 2020 census numbers erred in a meaningful way. The North was generally overestimated by several % and the South was generally underestimated by several %. The 2020 census error margin was on a scale not seen since before WWII. I would be very careful extrapolating these numbers.
How do we know that it's actually an error on the census's part rather than them just being lazy with estimates?
It'd be hilarious though if Trump singlehandedly cost Rs an extra seat in FL and TX they otherwise would've gotten.
They acknowledge it here: https://www.census.gov/library/stories/2022/05/2020-census-undercount-overcount-rates-by-state.html
The New York overcount and the Florida and Texas undercounts are likely >500K people each. This is not normal.
This is true but also not true. If you read through the report, you'll see that their estimated undercount and overcount are taking a limited household survey similar to the ACS methodology used to produce their ordinary estimates, then subtracting the actual Census figures. In other words, they largely just assumed that the methodology for their estimates would produce a result that was more correct than the Census methodology and claimed undercounts and overcounts in the Census on that basis. Therefore, it's really not any more revealing or a greater admission than when they put out the 2021 estimates that largely repeated pre-2020 estimate results. Whether the estimates are actually more correct than the Census is truly impossible to know for certain (I'm not going to rule out that they are), but I would think a new Census survey in 2030 utilizing (mostly) the same methodologies as in 2020 is more likely to produce a result similar to the 2020 Census than to estimates that use an entirely different methodology.
One point: If the Census was truly so far wrong in multiple places, you'd expect it show up fairly dramatically in voter registrations and turnout figures relative to 2010 (e.g., a state like Hawaii that is reported to be dramatically overestimated should be showing much lower voter registrations and turnout figures as a percentage of the overall population and VAP than it did shortly after 2010). It hasn't.
