Map of estimated Jewish Population Density in Europe around 1900 See the change from 1933 and 2015: brilliantmaps.com/jewish-po...

I feel like we haven’t fully reckoned with the with the implications of the reality that in the past generation, many have lost the ability to concentrate long enough to read a book.

The latest Republican running for Colorado Governor labels Gov. Jared Polis and others as the "synagogue of Satan Jews" and has called for their execution by hanging. MAGA podcaster Joe Oltmann recently said "Violence is probably the only solution."

This big look at the Middle East from Roger Cohen is both harsh and hopeful. And I appreciate being quoted with a couple of the fragile shards of hope. www.nytimes.com/2025/12/28/w...

In historical terms, enjoying a quiet picnic on railroad tracks before you can hear the roar of the approaching locomotive.

April 18, 1930: The BBC announces there is "no news" that day and plays piano music. www.bbc.com/news/enterta...

"Norman Rockwell was antifa," his granddaughter says. (Fact check: literally true.) www.thebulwark.com/p/norman-roc...

Number of School Shootings Per Million People By US State Between 1966 and 2025 Data for Europe: brilliantmaps.com/school-sh...

גיל פלדמן מזכיר לנו: במסמך המצורף לדיון הממשלה בהדחת היועמ"שית מוגדר אלי פלדשטיין כעובד לשכת רה"מ...

גיל פלדמן מזכיר לנו: במסמך המצורף לדיון הממשלה בהדחת היועמ"שית מוגדר אלי פלדשטיין כעובד לשכת רה"מ...

"No one voted for your advisors And nobody wants amends You're the one we voted for, so you must take the blame For handing out authority to men who were insane" ארלו גותרי, 1974.

"?ואם לא ידעת על זה, אז מה עוד לא ידעת?" ארלו גותרי, "Presidential Rag"' 1974, על פרשת ווטרגייט. נשמע כאילו נכתב לישראל 2025. www.youtube.com/watch?v=wTLK...

"If you didn't know about that one, well then what else don't you know?" If I didn't remember that Arlo Guthrie released Presidential Rag in 1974 about Nixon, I'd swear he wrote it about Netanyahu 2025. www.youtube.com/watch?v=wTLK...

"If you didn't know about that one, well then what else don't you know?" If I didn't remember that Arlo Guthrie released Presidential Rag in 1974 about Nixon, I'd swear he wrote it about Netanyahu 2025. www.youtube.com/watch?v=wTLK...

"?ואם לא ידעת על זה, אז מה עוד לא ידעת?" ארלו גותרי, "Presidential Rag"' 1974, על פרשת ווטרגייט. נשמע כאילו נכתב לישראל 2025. www.youtube.com/watch?v=wTLK...

When I was on CNN's northeast desk, the first job at 6a every morning was to read the local papers. That guided some of what the massive CNN machine followed every day. When those papers die, CNN doesn't hire new regional reporters. What you get instead is more panel shows, more punditry.

You need to be an antisemite to come up with such a dumb, pretzeled explanation why the killing of Jews in Australia wasn’t really antisemitism

When you’ve gone to this place, it’s time to just admit your only interest in antisemitism is claiming it doesn’t exist

Any journalist who has not at least once committed reckless disregard of leadership is not to be trusted.

One of the big changes from print periodicals to digital analytics is we now know that people mostly read stuff when they're at the office pretending to work. www.slowboring.com/p/nobody-wil...

Here is something to help your quality of search experience slightly: When typing your query, add "-ar" to the end. This suppresses the useless AI slop.

That is so cool. The moment when a historian knows something the rest of the world has yet to hear

The most exciting part of historical research is discovering what you weren't looking for, the answer to a question you hadn't asked. Anyone using AI to do historical research will not have this happen and will not, in fact, be doing real research.

The whole idea that Weiss spiked the CECOT story because there wasn't an on-the-record statement from the Trump administration is laughable, as Sharyn Alfonsi notes. It's regular journalistic practice to run a story without comment, as long as you've given the source a chance, which 60 Minutes did.

You do everything in your power to get every relevant voice in a story. But as one of my former colleagues told a local official who thought his silence would kill a piece: “I’d like to have you. But the story is being written with or without you.”

When I started my book on the origins of Israeli settlement, I expected to tell the standard Israeli story of religious extremists forcing the hand of weak Labor governments. Within my first days in the archives, I found a very different story. Only going through old paper made that possible.

Story via @davidfolkenflik.bsky.social www.npr.org/2025/12/22/g...

Reporter: We have 3 videos, 2 breathalyzers, 4 bartenders, and a minivan full of nuns attesting to how the driver was drunk and reckless when he hit the school bus. Weiss; What did the driver say. Reporter: He and his lawyer won’t talk. We asked them 5 times. Weiss: I’m killing your story.

No, absolutely not. Most archival materials have not been digitized, many would be illegible to AI (handwriting etc) even if they were. But the question came out of the research. There’s no way AI can do that.

I wish I could like this louder. There are SO many catastrophic reasons to resist AI, but the willful abandonment of creative, analytical, unpredictable, human thought processes is heartbreaking.

So when we rely on AI, we not only foreclose the possibility of novel thinkinh; we hamper our own human ability to develop those new paths. We develop the machine's neural network at the expense of our own.

That led me to still more material I never originally considered, and ultimately to a book that began in the New Deal and ended where I once thought the book would begin, in the 1960s. I made countless discoveries along the war and found something new to say. All because I did the work myself.

The standard story was that 1950s religious revivals were about contrasting America with the godless commies of the USSR, but I found material that hinted at an older well funded campaign that reframed the revival as a corporate pushback against the New Deal. So I followed that unexpected thread!

That discovery led me to entirely new archives that hadn’t been on my radar before. And at the Eisenhower library, as I researched religious nationalism in the 1950s, I found a document that once again subverted the expectations I originally had (and that AI would still have)

Instead, I launched myself into the archives and soon found that religious conservatives in the 1960s were strongly influenced by the politics of religious nationalism (In God We Trust, etc) of the 1950s — even though the literature (which AI would’ve used) told me that was meaningless. Nope!

For months I looked at German messages sent by Enigma & deciphered at Bletchley Park. Other researchers told me there was no way to tell what key of Enigma they'd been sent in. Suddenly scribbles in nearly illegible handwriting quadrupled in size in front of me: abbreviations for the names of keys.

Any prompt you write to use in an archive of that scale, no matter how carefully crafted and set up, is deterministic. You'll get back answers that confirm the assumptions you made when you set up the prompt. It forecloses the process of discovery that is essential to writing good history.

It's within the realm of possibility for an LLM, even if setting it up to do it well is very hard. An archive the size of Venice's, or the Vatican's, or the US National Archives is the opposite of a sandbox. Going in there is an open ended task that always takes you somewhere unexpected.

That doesn't even get into issues like names. Is the Zuanne q. Francesco in this document here the same guy as the Zuanne q. Francesco in that other document over there? You can figure it out, but only via a whole range of contextual clues that an LLM is not going to be able to pick up on.

Even if all that paper were digitized, which it obviously isn't, and even if an AI could reliably transcribe it all, there still isn't a prompt you can write that will get you from that pile of words to an understanding of what the people who produced it were really doing, much less why.

Every project is different, and almost always involves a painful learning curve during which you have to figure out how some set of premodern institutions thought about and handled some set of issues. You only figure that out by sitting in the reading room week after week, looking at paper.

It's the latter, but at a much larger scale than you may be thinking. You don't know what questions you can ask until you've spent a lot of time with the sources. But you also don't know which sources are going to be important going in. Archives are not organized in some obvious or transparent way.

If I’d asked AI to guide me, it would’ve mapped out the book I envisioned in my head because, like me, it would’ve drawn its conclusions just from what we already knew. No surprises.

The most exciting part of historical research is discovering what you weren't looking for, the answer to a question you hadn't asked. Anyone using AI to do historical research will not have this happen and will not, in fact, be doing real research.

Not a historian but like to research. The AI might summarize what I'm looking for, but it doesn't find what I'm *not* looking for. The book on the shelf next to the one I wanted. The insight in chapter 6 based on the quote I needed from chapter 4.

Lovely thread.

If you want to take one thing away from the art of the New Deal it is - in my humble opinion - that throwing money at *NORMAL* artists to decorate public buildings makes the world a much nicer place, particularly if you charge them with making the decorations evoke the history of where they are.

If I die laughing, it will be your fault.

Thanks @baygrl.bsky.social 😍

When I was a kid, mom explained to me that democracies named stuff after dead leaders and dictatorships named stuff after the current leader.