plausibly deniable misestimations of budget impact were a favored tactic of Chicago city leaders long before Johnson, but it's less tolerable now than ever
idk if they're reckless or ignorant, but it's infuriating when you understand the stakes
Eddie retweeted
New: Mayor Brandon Johnson’s CFO resigns after three months on the job, saying she is no longer able to fulfill her duties to provide accurate statements about the city’s finances.
Eddie retweeted
“Hint of Lime” is an astonishingly dishonest name for a tortilla chip that tastes like it was manufactured during an industrial accident at a lime processing facility
Eddie retweeted
Democratic sacrificial lambs in Trump +69 districts getting the call at 2 am on Wednesday that they’re going to be county groundwater commissioner.
Eddie retweeted
Elon Musk has spent more money on American elections in the last two years than George Soros has in his entire life.
Eddie retweeted
Working in the Pentagon
Replying to @DanLamothe
A group dispatched by Hegseth's team in February attempted to move the historic globe, said to be used by Eisenhower during World War II.
They lifted it from its home in an Army office in the Pentagon ... and dropped it.
Eddie retweeted
In 2016, on Tuesday, looking at data rolling in, I told the then-global head of rates at JPM that Trump would win.
It was not, shall we say, a consensus call at the time.
So I want to talk a little bit about what I'm seeing right now, since the timeline is going crazy.
On a trading desk, it matters more to be right than to be popular, and it matters more to be right than to predict what you want to happen.
I have the curse of being a mathematician by training. So I still have the muscles, even if I don't use them as much.
Right now I would say this:
1 - Anyone who thinks they know anything with high confidence you should immediately discount in this election cycle.
2 - The error bars on races, especially the swing races in the Senate, are likely so large that polling may not be particularly predictive.
3 - There are conflicting cross factors that really are making a mess out of some of the traditional polling which is causing this.
Namely, what I'm seeing is a few things that gives me major heartburn.
First, non-response rates to polls are now so high I think it's impossible to know if we have a representative sample of any subgroup. We can 100% get the aggregate weighting correct to voter turnout, but when you model your R sample at 10% crossover votes and get 4%, or your I sample at mildly D leaning at +2% but instead get +10%, it means all your re-weighting was an exercise in false specificity.
Second, we have major conflicts in the data. People are claiming they did not vote for Trump in 2024 when we are pretty sure they did, and we are getting R samples in some swing states where if the sample is correct about the sides voting in 2024, Harris would be president (she's not, if you've been in a coma). At the same time, we have young people looking maybe way more like they did in 2022 than 2026, and we know we are missing a large chunk of that sample because they never fucking answer the phone, so the net result is... ... ...? Yeah, nobody knows.
Third, mutually combative forces. Anti-incumbency is a thing in mid-term elections, it's also a thing in American politics where "fuck the guy in charge I hate it here" seems to be stronger than partisanship in an increasing number of cases. This means that helps Dems in red states but hurts Dems in blue states? The reverse would be true for Republicans. These kinds of forces produce compression.
Fourth, social media and Twitter are not the real world, but the distortionary forces here are influencing the polling landscape, where I would imagine the most media-connected people are also the ones taking polls, which leads to the invisible sample also being the ones in a different discourse bubble than the ones on here.
The punch line is simple: take any error bar and triple it. Quadruple if you're in Maine or something. We don't know anything, and any confident prediction about Dems taking the Senate or Republicans holding the House are almost certainly wrong. Not in their specific outcome, but rather, in good process. All pigs are fish therefore 2+2 = 4 type thinking.
Right now, depending on how you want to torture the data, you can see any outcome from R/R to D/D and any middle state. We've had large polling misses meaning we should be favoring Dems (NJ governor, anyone?), Rs being underweighted in the past meaning we should favor them (IA, anyone?), and just internal polling that was way off (El-Sayed way up in the primary, anyone?).
Anyone giving you a confident hot take here, in short, is full of it. Even the early voting is not uniform, and while the information content there is not always high, there is usually not zero, and if you're looking at data from GA vs. OH, you aren't getting the exact same vibes, much less some other places.
In short: national politics is getting weirder, polling is getting harder, the parties are fragmenting internally, and nobody knows shit yet.
Good news, though: they'll all claim they were right after the fact. What you should do is track both predictions and process to see who is really worth listening to over time.
Eddie retweeted
[guy with Christ Is King in his bio] I can't wait to watch this guy get shot to death
BREAKING: The Pentagon says it will livestream the firing squad execution of Fort Hood shooter Nidal Malik Hasan. apnews.com/article/fort-hood…
Eddie retweeted
Man in a Blue Mantle, by John Singer Sargent, 1891, 📸 via @cjzero
Eddie retweeted
I really need more big names in cybersecurity to come forward and state the obvious: cybersecurity is real and works and yes we absolutely can contain an AI even if it’s extremely good at finding zero days.
Eddie retweeted
a rogue openai agent broke into my freezer last night and ate an entire quart of ice cream
Eddie retweeted
leaving aside the truth value of this, this dynamic is part of the reason so many researchers profess high p(doom)s.
a low p(doom) is most often explained by a lack of understanding of the progress of AI, so a low p(doom) is low status
Replying to @tszzl
especially embarrassing for VCs… you missed out on the ai wave the first time because you didn’t get it. you are going to make bad capital allocation decisions again and again if you haven’t internalized the premises that make the danger of this technology obvious



































