The year that broke the charts

The persuasion/turnout paper that's not all it seems, the myth of the 19 million missing Biden voters, combative centrism, friends and foes, urban decline reversal, vibecoding

Patrick Ruffini

Mar 14, 2025

No. 353 | March 14th, 2025

🦠 Covid: Five Years Later

Five years ago this week, the world shut down. Here are all the different ways Covid-19 broke the charts.

The great resignation was real:

Grocery prices spiked, but restaurant prices grew even more — as we’re eating out more than ever before:

Time spent socializing with others hasn’t recovered.

Where we work has changed permanently.

Americans across partisan lines agree that the pandemic changed their lives forever but they remain divided on how it was handled. .

🔬 Academia

A new paper finds that ideological moderation might be overrated compared to other factors like turnout.

I had five different AIs summarize the paper. (Fwiw, in a blinded test, the AIs themselves rated Grok 3.0 as having the best summary.)

The findings run counter to conventional wisdom that ideological moderation should matter a lot, and some people have been reacting accordingly.

I don’t think the paper disproves that after all.

The penalty for ideological extremity is greater in gubernatorial (1.9 points) and presidential (1.0 points) than in legislative elections. In higher profile elections, there’s a bigger penalty.

Since the numbers cited are the effect on one party’s vote share, the effect on the margin is double that — almost 4 points for gubernatorial elections and 2 points in presidential elections.

Sorry, but in a close election, that’s a big effect. In this election, that was the difference between President Trump and President Harris.

The effects being magnified in high-profile races — read: close races — makes ideological moderation all the more important.

It intuitively makes sense that the penalty for legislative offices is lowest than for executive offices, and it is highest for governors. When you’re voting for Congress or the state legislature, what matters isn’t just who wins your local election, but the overall partisan makeup of the legislative body. You’re willing to make tradeoffs to make sure that your party is in the majority.

Meanwhile, here’s a good analysis showing how the paper’s claims about turnout aren’t all they’re cracked up to be.

🇺🇲 2024

No, there weren’t 19 million missing Biden voters, argues Charlotte Swasey.

Because you can’t assume that voters who went “missing” wouldn’t have voted for Trump had they voted (in fact, polls suggest many would have)
Because running the same analysis on the other side yields 12 million missing Trump 2020 voters, making the original 19 million number a wild overstatement. Churn in the electorate is an actual thing that people almost never take into account.

📊 Public Opinion

A through-line from the Echelon March voter omnibus: combative centrism.

Democrats want their party to be more combative against Trump and more centrist.

Who do Americans consider to be our friends and foes?

👫 Demographics

Is urban decline now reversing?

Pew Research takes a look at the gender gap that is emerging in teen experiences.

💻 Artificial Intelligence

😬😬😬😬

Josh Kalla isn’t worried yet, since this use case tends to be quite costly.

For now.

What is vibecoding?

“Vibecoding, a term that was popularized by the A.I. researcher Andrej Karpathy, is useful shorthand for the way that today’s A.I. tools allow even nontechnical hobbyists to build fully functioning apps and websites, just by typing prompts into a text box. You don’t have to know how to code to vibecode — just having an idea, and a little patience, is usually enough.
‘It’s not really coding,’ Mr. Karpathy wrote this month. ‘I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.’
My own vibecoding experiments have been aimed at making what I call ‘software for one’ — small, bespoke apps that solve specific problems in my life. These aren’t the kinds of tools a big tech company would build. There’s no real market for them, their features are limited and some of them only sort of work.
But building software this way — describing a problem in a sentence or two, then watching a powerful A.I. model go to work building a custom tool to solve it — is a mind-blowing experience.