I think the main thing to understand in the Great Debate over whether or not moderation leads to better electoral performance is that it really, really matters how you measure ideology. Jake Grumbach and Adam Bonica are using a metric based on campaign finance receipts and

Deconstructing WAR — and is it really stacking the deck for moderates? Here’s a very detailed breakdown of our WAR metric, which also responds to questions from the last few days. split-ticket.org/2025/08/15/dec…

Deconstructing our Wins Above Replacement model and explaining how it works, for all curious. (And why progressives don’t do as well in it) split-ticket.org/2025/08/15/dec…

Really good piece here, let me offer some less technical comments that may make this a little bit more intuitive. Say you really like Ritchie Torres' politics and you also notice that Torres ran a lot stronger than Kamala Harris in his district.

實事求是

As it happens, the biggest swings toward Trump in 2024 took place among non-white voters. This means the incumbents whose performance is most inflated by downballot lag were mostly House Dems with very safe seats who progressives like.

Solid piece. I don't really understand Jake M. Grumbach and Bonica's case for rejecting lagged presidential vote as an adjustment, particularly given that the latter considered it a key variable in a different context

Jake M. Grumbach 2) They may have one! But found this bit of their piece odd: There's no reason in principle why an optimal set of adjustments would necessarily have ideologically neutral implications So their basis for declaring adjustments for lagged presidential vote "biased" is unclear

Jake M. Grumbach Lakshya Jain I think the real slight of hand is to go "I must drop any covariate adjustments that make progressives look worse" and then motte/bailey down to "Alas, none of this is an RCT, knowledge is impossible, therefore my priors are correct" on criticism.

Coming soon to Silver Bulletin: E-WAR. I settle the moderation debate once and for all by scoring candidates based solely on how much I like them. My favorite candidates are all overperformers and everyone I don't like is a bad candidate.

This is a convincing rebuttal, imo. I will say it was always hard for me to believe that a model that had Rashida Tlaib as an overperformer in 2024 was systematically biased against the left.

VoteHub is possibly the most intricate and valuable project to ever come out of Election Twitter and it is incredible that this thing exists for public access.

Democrats in February: Gavin Newsom’s career is over for platforming Charlie Kirk, Steve Bannon and turning on trans-people Democrats in August:

I would take my life savings and short this man's presidential stock if I was allowed to. How does anyone even think Beshear 2028 is even a possibility? Only way that happens is if he's the VP pick and someone dies.

Talked with Tim Miller and Cameron Kasky at The Bulwark about the CA maps, Gen Z politics and the job uncertainty AI causes for new grad engineers. Plus a debate about how "forcefully voicing what we believe in" became synonymous with "no disagreement". youtube.com/watch?v=gmZgsS…