This was a good discussion but somehow I feel it entirely missed the point. This is takeover and destruction of the U.S. Government. It's a criminal matter. There needs to be a look at the unspeakable.
This is a criminal takeover but as Greg Sargent has pointed out the GOP is still acting like it has to be responsive to electoral influences. Trump cares about his electoral standing and so should we.
This is a COUP! There is nothing systematic about coups. I know because I lived through one. I do not expect 2026 midterm elections, and if they happen, Muskrat will fix it. It'll be a repeat of the 2024 election. None of these goons is acting in a way as if they're concerned about voter backlash, and that speaks louder than words. Voters are irrelevant in a dictatorship. As for Nate's comment "if we lose respect for the rule of law"....OMG, it's already happened! There's no "if" here.
America is on the edge but court cases are still happening and Musk/Trump are still acting like the courts matter. That might change in the future but don't make it happen by being nihilistic like Musk/Trump.
I have heard that Trump only believes in a "I win, you lose" model of interaction. That precludes all "win-win" scenarios. Trump would actively try to kill such options if the assertion is correct. This seems consistent with recent behavior.
I'm not sure why they don't use the correct model of Trump which is that he is a malignant narcissist and all his interactions with others (countries included) are based on desperate need for adjulation, and power and revenge.
There's nothing at all mentioned or no analysis of this aspect of him and it's actually running the whole show.
Let’s not do this. Silver tried to turn politics into sports ball betting. His models only work in hindsight, if sports ball results (or election results) were predictable there wouldn’t be any betting at all. Let Silver go back to sports, he isn’t worth listening to about politics.
The reason current AIs are so poor on visual processing is that they are designed to analyze textual information. To process visual information, you have to build specialized subsystems that impose obvious constraints like translation, rotation and magnification invariance and common distortions like parallax (shearing). Then you have to build a huge repository of standard items like edges of various shapes (again invariant under the appropriate symmetry groups). That massively reduces the rank of the linear algebra equations that the model solves. Since solving such problems scales worse than linear with most algorithms, the reduced rank massively speeds up solution and reduces power consumption. ...sorry for the long answer, but I did this with an OCR project back in the 1990s while working on my master's degree.
Again, regarding AI, any company that depends upon intellectual property must be considering using, or using, AI to seek out new patents. For example, many semiconductor companies (and, e.g., biotech companies) try to create "fences" and other means of extending their patents. They do this via relatively minor changes to the original patent. Why not use AI to survey all of the current patents and seek out minor extensions (or major inventions)? Of course they should. And use their engineers to cull through to figure out what's can be patented. Perhaps t he issue: patent law requires conception, non-obviousness and reduction to practice. In the event that you use AI, who "conceived" the invention? It appears that there's no human "inventor." But you might come up with some great innovations.
Making federal services like Social Security fail, due to staff cuts & underfunding, will give Trump/Musk the opportunistic excuse to PRIVATIZE them & channel public funds & assets to themselves & their wealthy oligarch pals.
In MHO, Krugman is right about narrow uses of AI. I'd use the example of protein folding, which in the past has been something done by grad students, who published a new protein fold every few years, but then AlphaFOLD solved a lot of the protein folding problems. So yes, let's use it in narrow circumstances where it works.
Indeed, defined-scope applications of AI are useful and practical. Paul Krugman's example from their own textbook was telling. A small all-volunteer NGO I know is using AI to convert Minutes of meetings into subject matter posts on the web site. Another example is right here on Substack which, until recently, had an Ask a Question button on the dashboard page; its answers were trained on how Substack works and I have used it many times to advantage. (I wonder why it disappeared.) David Clinton of The Audit (https://www.theaudit.ca/) has used AI to summarize huge documents like Public Accounts and transcripts of Parliamentary Committee meetings.
There needs to be a place where all can get the facts. Someone needs to publish “The Trump Papers” (as I would call it) on the Internet. A bank of information containing documents (and summaries of documents) like the Jan 6 report (and rebuttal), the Mueller report, the Jack Smith report, Trump’s criminal cases, false claims and lies, lawsuits, impeachment records (both), profiteering, attacks on the media, attacks on the constitution, first term, covid response, retribution efforts. Russia associations, cabinet profiles, views on women, views on race, views on climate change . . . and a collection of relevant, revealing articles from the NYT, the WP, and other credible recent essays by Krugman, Acosta, Reich, Rubin, and others.
This was a good discussion but somehow I feel it entirely missed the point. This is takeover and destruction of the U.S. Government. It's a criminal matter. There needs to be a look at the unspeakable.
This is a criminal takeover but as Greg Sargent has pointed out the GOP is still acting like it has to be responsive to electoral influences. Trump cares about his electoral standing and so should we.
Why, trump told his followers that they wouldn’t have to ‘vote again’?!?!
So a felon convicted 34 times for fraud might not be telling the truth? He might be engaging in wishful thinking?
I don't see it. https://www.theguardian.com/us-news/2025/feb/27/republicans-trump-threats?utm_term=67c061d1ee118bb63143f34aa95bc50b&utm_campaign=GuardianTodayUS&utm_source=esp&utm_medium=Email&CMP=GTUS_email
it seems they kind of shifted gears to avoid a pit of despair.
This is a COUP! There is nothing systematic about coups. I know because I lived through one. I do not expect 2026 midterm elections, and if they happen, Muskrat will fix it. It'll be a repeat of the 2024 election. None of these goons is acting in a way as if they're concerned about voter backlash, and that speaks louder than words. Voters are irrelevant in a dictatorship. As for Nate's comment "if we lose respect for the rule of law"....OMG, it's already happened! There's no "if" here.
America is on the edge but court cases are still happening and Musk/Trump are still acting like the courts matter. That might change in the future but don't make it happen by being nihilistic like Musk/Trump.
I have heard that Trump only believes in a "I win, you lose" model of interaction. That precludes all "win-win" scenarios. Trump would actively try to kill such options if the assertion is correct. This seems consistent with recent behavior.
I'm not sure why they don't use the correct model of Trump which is that he is a malignant narcissist and all his interactions with others (countries included) are based on desperate need for adjulation, and power and revenge.
There's nothing at all mentioned or no analysis of this aspect of him and it's actually running the whole show.
Even Krugman talks as if things are normal; that the centre is holding. It isn't. We're living a catastrophe.
Let’s not do this. Silver tried to turn politics into sports ball betting. His models only work in hindsight, if sports ball results (or election results) were predictable there wouldn’t be any betting at all. Let Silver go back to sports, he isn’t worth listening to about politics.
So many Billionaires, so many cowards, still on playground getting beat up by the bully....
You aren’t seeing the point. Trump doesn’t expect to ever leave the office of the presidency. He doesn’t give a shit what Elon is doing.
The reason current AIs are so poor on visual processing is that they are designed to analyze textual information. To process visual information, you have to build specialized subsystems that impose obvious constraints like translation, rotation and magnification invariance and common distortions like parallax (shearing). Then you have to build a huge repository of standard items like edges of various shapes (again invariant under the appropriate symmetry groups). That massively reduces the rank of the linear algebra equations that the model solves. Since solving such problems scales worse than linear with most algorithms, the reduced rank massively speeds up solution and reduces power consumption. ...sorry for the long answer, but I did this with an OCR project back in the 1990s while working on my master's degree.
AI is not intelligent. Eventually people will figure that out.
It doesn’t need to be, they can filter and sort information, and push some content, and suppress others. Thats all it needs to do.
Ummm the plastic straw thing is for federal government offices Nate
I hope MAGA understands that the disabled, like a handicapped sibling, when your parents die, depends on Medicaid.
Again, regarding AI, any company that depends upon intellectual property must be considering using, or using, AI to seek out new patents. For example, many semiconductor companies (and, e.g., biotech companies) try to create "fences" and other means of extending their patents. They do this via relatively minor changes to the original patent. Why not use AI to survey all of the current patents and seek out minor extensions (or major inventions)? Of course they should. And use their engineers to cull through to figure out what's can be patented. Perhaps t he issue: patent law requires conception, non-obviousness and reduction to practice. In the event that you use AI, who "conceived" the invention? It appears that there's no human "inventor." But you might come up with some great innovations.
Making federal services like Social Security fail, due to staff cuts & underfunding, will give Trump/Musk the opportunistic excuse to PRIVATIZE them & channel public funds & assets to themselves & their wealthy oligarch pals.
In MHO, Krugman is right about narrow uses of AI. I'd use the example of protein folding, which in the past has been something done by grad students, who published a new protein fold every few years, but then AlphaFOLD solved a lot of the protein folding problems. So yes, let's use it in narrow circumstances where it works.
Indeed, defined-scope applications of AI are useful and practical. Paul Krugman's example from their own textbook was telling. A small all-volunteer NGO I know is using AI to convert Minutes of meetings into subject matter posts on the web site. Another example is right here on Substack which, until recently, had an Ask a Question button on the dashboard page; its answers were trained on how Substack works and I have used it many times to advantage. (I wonder why it disappeared.) David Clinton of The Audit (https://www.theaudit.ca/) has used AI to summarize huge documents like Public Accounts and transcripts of Parliamentary Committee meetings.
There needs to be a place where all can get the facts. Someone needs to publish “The Trump Papers” (as I would call it) on the Internet. A bank of information containing documents (and summaries of documents) like the Jan 6 report (and rebuttal), the Mueller report, the Jack Smith report, Trump’s criminal cases, false claims and lies, lawsuits, impeachment records (both), profiteering, attacks on the media, attacks on the constitution, first term, covid response, retribution efforts. Russia associations, cabinet profiles, views on women, views on race, views on climate change . . . and a collection of relevant, revealing articles from the NYT, the WP, and other credible recent essays by Krugman, Acosta, Reich, Rubin, and others.
If you build it, they own't come.
We have a place where everyone can get the facts. The people that need to get the facts aren't interested in the facts.
Feench Candadian here:
I hear you Dr…. Thank God Trump’s crazy!
Should he have been smart… America was doomed.
The beautiful Irony… he surround himself with people more incompetent then him…
Perfect actually… so one day a new administration might be able to salvage what will be left standing.
And Hey!… your former allies will also have to thank him for forcing us to find new markets and (true) trading partners.
Sadly true.