4 ms·

So, you think that if you swapped R and D in the dataset, then the algorithm would always make the D win? Given that they are talking about open source tools, t

So, you think that if you swapped R and D in the dataset, then the algorithm would always make the D win? Given that they are talking about open source tools, that should be trivial for someone to prove, if they were so inclined. It sounds to me like you just don't like the possibility that "your side" might be doing something politically nefarious. And don't get me wrong, it's likely that both sides are gerrymandering. It just so happens that republicans have more governorships and other state power, and have had much more opportunity to draw districts. You should be happy about the existence of mathematical tools to try to prove when someone in power is choosing their voters, instead of the voters choosing the person in power.

CheezeIt 5y ago
Democrats will want the algorithm that helps them, GOP will want the algorithm that helps them. That doesn’t mean states won’t eventually settle on an algorithmic solution. We’ve been through this before, with competing congressional apportionment algorithms.
- HWR_14 5y ago
  Quite a few states have replaced partisan assignments with more neutral algorithms.
someguydave 5y ago
no, I think in practice academics will claim an algorithm is neutral when it in fact favors democrats
josephcsible 5y ago
> So, you think that if you swapped R and D in the dataset, then the algorithm would always make the D win? Algorithms will fall into two groups: the first will favor Democrats in the real world and Republicans in the counterfactual swapped-dataset world, and the second will favor Republicans in the real world and Democrats in the counterfactual swapped-dataset world. There will be arguments that multiple algorithms, some from each group, are each the "fairest" by varying definitions. GP suspects that it will be mainly algorithms from the first group that end up being used.

CheezeIt 5y ago

Democrats will want the algorithm that helps them, GOP will want the algorithm that helps them. That doesn’t mean states won’t eventually settle on an algorithmic solution. We’ve been through this before, with competing congressional apportionment algorithms.

HWR_14 5y ago
Quite a few states have replaced partisan assignments with more neutral algorithms.

someguydave 5y ago

no, I think in practice academics will claim an algorithm is neutral when it in fact favors democrats

josephcsible 5y ago

> So, you think that if you swapped R and D in the dataset, then the algorithm would always make the D win? Algorithms will fall into two groups: the first will favor Democrats in the real world and Republicans in the counterfactual swapped-dataset world, and the second will favor Republicans in the real world and Democrats in the counterfactual swapped-dataset world. There will be arguments that multiple algorithms, some from each group, are each the "fairest" by varying definitions. GP suspects that it will be mainly algorithms from the first group that end up being used.