Here is an argument against donating to presidential candidates, stated less glibly than I did last night.
First premise: downballot races need the money. Even small donations to House and state candidates make a difference.
Second premise: presidential candidates don’t really need your money. They won’t notice it. They’re swimming in it.
Third, and most important premise: a downballot donation helps the top of the ticket.
That is to say: every dollar you donate to JD Scholten in IA-4 is going to help Sanders, Warren, Klobes, whoever. The voters JD Scholten turns out aren’t going to vote for Trump.
(Is that 100% true? No, but, close.)
In fact, in places like Iowa, the effect is even stronger, because a vote turned out by Scholten is also going to go to whoever ends up running against Joni Ernst in her Senate reelection race.
This seems like an especially strong argument if you’re in a safe D area, like I am in Chicago and you are in CA and NY. Want to donate to Warren? Why not do it through Iowa or Maine and hit 3 races at once?
There are structural reasons this strategy makes sense, too, because downballot candidates are doing DIFFERENT THINGS than Warren and Biden. How much of a Warren donation goes to bid up the cost of media buys? Why allocate your donations to that?
• • •
Missing some Tweet in this thread? You can try to
force a refresh
People on the orange site are laughing at this, assuming it's just an ad and that there's nothing to it. Vulnerability researchers I talk to do not think this is a joke. As an erstwhile vuln researcher myself: do not bet against LLMs on this. axios.com/2026/02/05/ant…
I think vulnerability research might be THE MOST LLM-amenable software engineering problem. Pattern-driven. Huge corpus of operational public patterns. Closed loops. Forward progress from stimulus/response tooling. Search problems.
Vulnerability research outcomes are in THE MODEL CARDS for frontier labs. Those companies have so much money they're literally distorting the economy. Money buys vuln research outcomes. Why would you think they were faking any of this?
I feel like there's a "next Google" thing happening with Internet-scale LLMs and part of it is because I find myself using them more than Google but the "next" part is because it never made any sense to argue or cajole Google and that works with LLMs. A true advance.
You can also ask ChatGPT "why does anybody like Roxy Music they sound like a Eurovision band" and it will say "I see where you're coming from" and then give reasons. I love it so much.
In 1917, the Supreme Court in Buchanan v. Warley struck down Louisville's zoning laws, which explicitly restricted Black families, on the basis of the 14th Amendment.
1917 is very early in the history of American zoning. Oak Park, where I live, enacted zoning in 1921, and it was a novelty; newspaper articles described it as one of the first comprehensive zoning systems in the nation.
If you spend some time doing archival research, you quickly confirm a suspicion: people in the 1920s were luridly, cartoonishly racist. The Court outlawed explicit racial zoning, but not implicit racial zoning, and thus was born the "single-family zoning district".
Opening of Ronin. De Niro walks behind the cafe, cases the joint, hides his gun behind some milk crates. Why? Nobody’s patting him down. It’s a cafe! Why not just go in strapped?
This first car chase is great, Stellan Skarsgard riding along in the passenger compartment of an iPhone 4 directing the whole thing.
You don’t get all the boops and beeps Frankenheimer envisioned with this ’97 technology. Real missed opportunity, Apple. I’d have paid extra for it.