All of them are allergic to learning anything, even basic probabilities
November 11, 2025 at 2:04 AM
All of them are allergic to learning anything, even basic probabilities
The thing about this, tho, is that Jon Steinsson has devoted his career to a style of macroeconomics that is based on a rational representative agent maximizing their utility over infinite future time based on correct beliefs about the probabilities of all future outcomes.
November 11, 2025 at 4:33 AM
The thing about this, tho, is that Jon Steinsson has devoted his career to a style of macroeconomics that is based on a rational representative agent maximizing their utility over infinite future time based on correct beliefs about the probabilities of all future outcomes.
did a set of probabilities write this?
November 8, 2025 at 1:53 AM
did a set of probabilities write this?
If you live in a 100-year flood zone, here are the probabilities of experiencing a flood during the life of your hypothetical mortgage:
15 years: 14%
30 years: 26%
50 years: 40%
Plus, more years of mandated flood insurance premiums -- though you should buy it regardless of the lender requirement!
15 years: 14%
30 years: 26%
50 years: 40%
Plus, more years of mandated flood insurance premiums -- though you should buy it regardless of the lender requirement!
November 11, 2025 at 7:39 PM
If you live in a 100-year flood zone, here are the probabilities of experiencing a flood during the life of your hypothetical mortgage:
15 years: 14%
30 years: 26%
50 years: 40%
Plus, more years of mandated flood insurance premiums -- though you should buy it regardless of the lender requirement!
15 years: 14%
30 years: 26%
50 years: 40%
Plus, more years of mandated flood insurance premiums -- though you should buy it regardless of the lender requirement!
Only change in the latest SPC update was to nudge the tornado probabilities from 2% to 5% chance within 25 miles of you. Nothing major, but worth watching. Stay connected this evening. We'll communicate any adjustments through the day.
November 7, 2025 at 5:07 PM
Only change in the latest SPC update was to nudge the tornado probabilities from 2% to 5% chance within 25 miles of you. Nothing major, but worth watching. Stay connected this evening. We'll communicate any adjustments through the day.
normalizing constants arent real. probabilities arent real all we have is score functions
November 10, 2025 at 4:40 AM
normalizing constants arent real. probabilities arent real all we have is score functions
New paper!
We propose a framework to empirically study animal social relationships by modelling social network (SN) data as time-series—that is, without the need to aggregate them over time.
www.biorxiv.org/content/10.1...
We propose a framework to empirically study animal social relationships by modelling social network (SN) data as time-series—that is, without the need to aggregate them over time.
www.biorxiv.org/content/10.1...
November 12, 2025 at 11:56 AM
New paper!
We propose a framework to empirically study animal social relationships by modelling social network (SN) data as time-series—that is, without the need to aggregate them over time.
www.biorxiv.org/content/10.1...
We propose a framework to empirically study animal social relationships by modelling social network (SN) data as time-series—that is, without the need to aggregate them over time.
www.biorxiv.org/content/10.1...
Flat and well curved probabilities both have their place. In a vacuum they are entirely neutral.
The rounded curve of 2d6 works really well for so many concepts. I love it. When talking to @naldodrinan.bsky.social about Break!! recently he mentioned "I love how swingy the d20 is."
The rounded curve of 2d6 works really well for so many concepts. I love it. When talking to @naldodrinan.bsky.social about Break!! recently he mentioned "I love how swingy the d20 is."
November 10, 2025 at 2:52 AM
Flat and well curved probabilities both have their place. In a vacuum they are entirely neutral.
The rounded curve of 2d6 works really well for so many concepts. I love it. When talking to @naldodrinan.bsky.social about Break!! recently he mentioned "I love how swingy the d20 is."
The rounded curve of 2d6 works really well for so many concepts. I love it. When talking to @naldodrinan.bsky.social about Break!! recently he mentioned "I love how swingy the d20 is."
-- WEEK 11 #NFL 🏈 Playoff Leverage --
Here you can see how the playoff probability of each team changes depending on whether they win or lose the next game.
Here you can see how the playoff probability of each team changes depending on whether they win or lose the next game.
November 12, 2025 at 2:55 PM
-- WEEK 11 #NFL 🏈 Playoff Leverage --
Here you can see how the playoff probability of each team changes depending on whether they win or lose the next game.
Here you can see how the playoff probability of each team changes depending on whether they win or lose the next game.
lol every time I write my congress people about the shut down I nearly send the message with it spelled “shit down.” 🙃 I haven’t yet, but probabilities suggest I will one day.
November 7, 2025 at 8:26 PM
lol every time I write my congress people about the shut down I nearly send the message with it spelled “shit down.” 🙃 I haven’t yet, but probabilities suggest I will one day.
The value of the company is based upon speculative , expected, possible return; like throwing money down at a gambling table based upon probabilities of winning
Like the .com bubble,it is speculative profit being cashed in prior to it existing. Only those who cash out early win. Everyone else loses
Like the .com bubble,it is speculative profit being cashed in prior to it existing. Only those who cash out early win. Everyone else loses
I honestly don’t get the value of this company. They hoover up energy and water. Their product constantly gets things wrong and, in extreme cases, coaches people into suicide.
And it’s all built on what seems to be malicious and vast intellectual property theft.
What does OpenAI offer the world?
And it’s all built on what seems to be malicious and vast intellectual property theft.
What does OpenAI offer the world?
“authors & publishers who filed a lawsuit against the Sam Altman-led firm have secured access to internal Slack messages… discussing the mass deletion of a pirated books dataset… A NY district court ordered OpenAI to hand over the communications regarding data deletion”
futurism.com/artificial-i...
futurism.com/artificial-i...
November 9, 2025 at 8:17 AM
The value of the company is based upon speculative , expected, possible return; like throwing money down at a gambling table based upon probabilities of winning
Like the .com bubble,it is speculative profit being cashed in prior to it existing. Only those who cash out early win. Everyone else loses
Like the .com bubble,it is speculative profit being cashed in prior to it existing. Only those who cash out early win. Everyone else loses
This is a really brilliant animation, showing how global warming has increased the probabilities of extreme heat. Note how it's not just a matter of a shift in the mean -- the distribution has gotten broader. Probably not enough data yet to tell if the tails have gotten fatter.
Exactly! Which makes the real climate shift way scarier. Like shown by the data from svs.gsfc.nasa.gov/5452/
November 8, 2025 at 10:55 AM
This is a really brilliant animation, showing how global warming has increased the probabilities of extreme heat. Note how it's not just a matter of a shift in the mean -- the distribution has gotten broader. Probably not enough data yet to tell if the tails have gotten fatter.
these are unfathomably complex networks but it’s still just doing math really fast…… it’s computer. it doesn’t *understand* words it’s calculating probabilities
November 6, 2025 at 4:55 PM
these are unfathomably complex networks but it’s still just doing math really fast…… it’s computer. it doesn’t *understand* words it’s calculating probabilities
just as neoliberalism compelled people to "invest in their human capital" and "build their personal brand" etc etc (and some people pretended to enjoy this or were eager to sell how-tos about it), current socioeconomic conditions compel people to think about themselves as a set of probabilities
November 7, 2025 at 7:33 PM
just as neoliberalism compelled people to "invest in their human capital" and "build their personal brand" etc etc (and some people pretended to enjoy this or were eager to sell how-tos about it), current socioeconomic conditions compel people to think about themselves as a set of probabilities
okay! so, the output layer of the model is basically a 'keyboard' of tokens which is filled with probabilities which sum to one. when you are doing ethics training, you're trying to avoid catastrophic forgetting, a pathology which occurs when you train on data which isn't IID with pretraining.
November 7, 2025 at 3:38 AM
okay! so, the output layer of the model is basically a 'keyboard' of tokens which is filled with probabilities which sum to one. when you are doing ethics training, you're trying to avoid catastrophic forgetting, a pathology which occurs when you train on data which isn't IID with pretraining.
Big shifts in 2026 House probabilities after last night (Ds up to 71%), small shifts in 2026 Senate (Ds up to 31%) per @gelliottmorris.com www.gelliottmorris.com/p/seven-data... kalshi.com/markets/cont... kalshi.com/markets/cont...
November 5, 2025 at 1:27 PM
Big shifts in 2026 House probabilities after last night (Ds up to 71%), small shifts in 2026 Senate (Ds up to 31%) per @gelliottmorris.com www.gelliottmorris.com/p/seven-data... kalshi.com/markets/cont... kalshi.com/markets/cont...
What does it mean to assign probabilities to causal parameters? In just one minute and 33 seconds, I explain why this is puzzling, and how you can nevertheless make sense of it.
youtu.be/UMJSFkz2sCg
youtu.be/UMJSFkz2sCg
How to make sense of probabilities over causal parameters
YouTube video by Causal Foundations
youtu.be
November 6, 2025 at 4:25 PM
What does it mean to assign probabilities to causal parameters? In just one minute and 33 seconds, I explain why this is puzzling, and how you can nevertheless make sense of it.
youtu.be/UMJSFkz2sCg
youtu.be/UMJSFkz2sCg
I had a play in #rstats with the distribution of p-values under a true null hypothesis, for two-sample tests of equal proportions at different sample sizes. They're not uniformly distributed, but sometimes they are more non-uniform than other times. freerangestats.info/blog/2025/11...
November 9, 2025 at 2:17 AM
I had a play in #rstats with the distribution of p-values under a true null hypothesis, for two-sample tests of equal proportions at different sample sizes. They're not uniformly distributed, but sometimes they are more non-uniform than other times. freerangestats.info/blog/2025/11...
imho understanding Bayes is about 1. the weirdly simple consequences of modeling all uncertainty as probabilities, 2. the arbitrarily gnarly practical consequences of trying to do that (MCMC, cunning parameterizations, aliasing, Stan, etc) & 3. the mad philosophical side of 'wtf am I doing actually'
November 7, 2025 at 9:50 PM
imho understanding Bayes is about 1. the weirdly simple consequences of modeling all uncertainty as probabilities, 2. the arbitrarily gnarly practical consequences of trying to do that (MCMC, cunning parameterizations, aliasing, Stan, etc) & 3. the mad philosophical side of 'wtf am I doing actually'
and very small probabilities also have more or less degenerate penalties too. so in order for the model to learn to implement its ethics training, it has a very strong bias towards outputting chains of tokens which have at least a moderate likelihood of occurring in the corpus of natural language.
November 7, 2025 at 3:41 AM
and very small probabilities also have more or less degenerate penalties too. so in order for the model to learn to implement its ethics training, it has a very strong bias towards outputting chains of tokens which have at least a moderate likelihood of occurring in the corpus of natural language.
The Jedi DID fail Anakin, albeit indirectly. They lost their faith in the force, instead trusting too much on their material interests. Practicalities. Probabilities. Likely outcomes. Strategy.
They refused to believe in the chosen one when he was right in front of their eyes.
They refused to believe in the chosen one when he was right in front of their eyes.
Star Wars Episode II - Attack of the Clones - The Jedi Council - 4K ULTRA HD.
YouTube video by Star Wars Saga LatinAmerica
youtu.be
November 10, 2025 at 10:33 PM
The Jedi DID fail Anakin, albeit indirectly. They lost their faith in the force, instead trusting too much on their material interests. Practicalities. Probabilities. Likely outcomes. Strategy.
They refused to believe in the chosen one when he was right in front of their eyes.
They refused to believe in the chosen one when he was right in front of their eyes.
I think I broke the AI that generates an answer to your search in the Brave browser. I know the common very oversimplified description of LLMs is that it chooses the next word based on probabilities and associations between words and all that, but I've never seen it more on display 😄
November 6, 2025 at 2:14 PM
I think I broke the AI that generates an answer to your search in the Brave browser. I know the common very oversimplified description of LLMs is that it chooses the next word based on probabilities and associations between words and all that, but I've never seen it more on display 😄
Rain returns midweek! 🌧️ Probabilities are increasing for 2"+ of rain Wed–Fri, highest in the North Bay, Santa Cruz Mountains, & Big Sur coast (60–90%), lower across the South Bay & interior valleys (10–30%). #CAwx #BayAreaWX https://x.com/NWSBayArea/status/1987568636882518119
November 9, 2025 at 5:29 PM
Rain returns midweek! 🌧️ Probabilities are increasing for 2"+ of rain Wed–Fri, highest in the North Bay, Santa Cruz Mountains, & Big Sur coast (60–90%), lower across the South Bay & interior valleys (10–30%). #CAwx #BayAreaWX https://x.com/NWSBayArea/status/1987568636882518119
I'm a day late to @scottsallen.bsky.social's newsletter, but pretty wild to see this chart with probabilities of where the #Commanders will pick in April's draft.
To get more interesting stuff like this delivered right to your inbox, sign up here: www.washingtonpost.com/newsletters/...
To get more interesting stuff like this delivered right to your inbox, sign up here: www.washingtonpost.com/newsletters/...
November 6, 2025 at 3:32 PM
I'm a day late to @scottsallen.bsky.social's newsletter, but pretty wild to see this chart with probabilities of where the #Commanders will pick in April's draft.
To get more interesting stuff like this delivered right to your inbox, sign up here: www.washingtonpost.com/newsletters/...
To get more interesting stuff like this delivered right to your inbox, sign up here: www.washingtonpost.com/newsletters/...