Are rallies for the faithful really the best use of a presidential candidates time?
In the UK we'd do a mix of interviews, soapboxes, meet the voters and battlebus stuff across a variety of constituencies.
Yes, it's quite a contrast to British campaigning. You never see Harris pointing at potholes, or Trump in a white coat staring at test tubes in a lab somewhere. Neither pulling pints either.
Pretending to work in MaccieDs or on a bin lorry?
Both notable for being a deviation from the usual style.
Are rallies for the faithful really the best use of a presidential candidates time?
In the UK we'd do a mix of interviews, soapboxes, meet the voters and battlebus stuff across a variety of constituencies.
Yes, it's quite a contrast to British campaigning. You never see Harris pointing at potholes, or Trump in a white coat staring at test tubes in a lab somewhere. Neither pulling pints either.
There is nevertheless a lot of door knocking and knocking up (not that they call it that), and leaflet delivery hung on door handles in the manner of a 'do not disturb' sign, as there aren't any letterboxes and they're not allowed to use the USPS mailboxes.
That presumably is the point of the rallies, to fire up the faithful and get them door knocking.
In Trump's case doesn't he keep them captive at his rallies for eight hours thus diminishing the time remaining for campaigning?
I was mildly alarmed today to learn that a structural engineer friend uses chatgpt to do his structural engineering calculations. Apparently the paid for version is better than the free version. But still.
Sorry, what??
I hope it isn't for anything more than a shed. And even then...
You’d be amazed how many otherwise intelligent professional people have come to treat LLM output as gospel. It really is rather worrying.
I did worry that would happen. Part of the problem is people calling it "AI" when it's nothing of the sort.
The next Grenfell-style disaster could well end up being due to someone using a LLM to do something out isn't capable of doing, and not checking the results.
It’s now extremely rare for ChatGPT to get “basic maths wrong”. In fact it is increasingly rare for these things to get anything wrong - tho of course it still happens
Anyway. Good day everyone
Seems a bit quiet. Nothing happening today?
I checked recently and it told me that "raspberry" contains two r's.
As a language model it's very impressive and turns out very plausible copy, although I find the bouncy demeanor they've given it annoying, but for anything factual it is worse than useless.
Re Georgia, and the view that more women out is a positive for Harris. It is worth remembering that Georgia was the site of a particularly high-profile case of an illegal immigrant killing a female student and one that generated a lot of anger.
It might not be abortion that is driving these female voters out.
Yes there’s been a lot of commentary on the Republican side about Laken Riley, and sadly a number of other similar cases that have been mentioned at rallies.
Here’s Megyn Kelly at Trump’s rally last night, endorsing Trump and mentioning both the sad cases of murder and sexual assault by illegals, but also the controversial issue of women’s sports which might also drive turnout among younger women. It’s not necessarily just abortion behind the differential female turnout. https://x.com/megynkellyshow/status/1853643950512316882
It looked more likely at the weekend than it does today, after the three coalition leaders met last night. The only way the government will be brought down is if the FDP split from the coalition but this would be a suicidal move for Lindner and his party. They have been consistently under the 5% hurdle and bringing down the government is hardly going to be a vote winner for them.
Yes voters don't like any of the coalition partners, but quite rightly the FDP is doing the worst. I absolutely despise them. They have been nothing but spoilers in this government and I'd give them very little chance of getting into parliament at the next election. Of course from the point of the FDP they have done exactly what they wanted.
So much data on all these voters, yet it still takes so damn long to count the actual votes!
Happy US Election Day everyone!
We'll know the result for Florida within two hours of polls closing. Georgia, unless it's very close, likewise.
Florida could be useful since if Trump wins it by less than 5% he's probably going to lose the election overall.
Whichever way it goes, I'd be surprised if this election remains on a knife edge for very long after the polls have closed.
I don't think it'll be an easy win for any candidate either.
I think I'm saying the opposite. Whoever wins, I doubt the final result will be close.
Depends what you mean by close. Trump could win all 7 swing states, while still losing the popular vote, and only winning the tipping point state by 1%. That would be a fairly comfortable win in terms of Electoral College numbers, and one state (Nevada) better than he did in 2016. But a election where the winner loses the popular vote, and a half per cent national swing would give it to the other candidate - I'd still call that close!
I'm thinking more of the electoral college than the popular vote. But my comment was really that either way, I don't think we'll have to wait very long after the polls close before the result becomes clear.
RCP average now +0.1 Harris. The polls have the race as a toss up. Evens the pair. The betting is almost 60:40 in Trump's favour. Harris almost 2.5 when the polls have her at 2.0. You could make 25% on your money if you backed Harris at 2.5 and were able to lay her back at 2.0. Will this prove to be the biggest Political Betting value opportunity ever? Huge liquidity available.
If it’s as close to 50-50 as has been suggested, then betting in-play tonight on whoever is odds-against is going to be good value.
I was mildly alarmed today to learn that a structural engineer friend uses chatgpt to do his structural engineering calculations. Apparently the paid for version is better than the free version. But still.
Sorry, what??
I hope it isn't for anything more than a shed. And even then...
You’d be amazed how many otherwise intelligent professional people have come to treat LLM output as gospel. It really is rather worrying.
I did worry that would happen. Part of the problem is people calling it "AI" when it's nothing of the sort.
The next Grenfell-style disaster could well end up being due to someone using a LLM to do something out isn't capable of doing, and not checking the results.
It’s now extremely rare for ChatGPT to get “basic maths wrong”. In fact it is increasingly rare for these things to get anything wrong - tho of course it still happens
Anyway. Good day everyone
Seems a bit quiet. Nothing happening today?
I checked recently and it told me that "raspberry" contains two r's.
As a language model it's very impressive and turns out very plausible copy, although I find the bouncy demeanor they've given it annoying, but for anything factual it is worse than useless.
I was mildly alarmed today to learn that a structural engineer friend uses chatgpt to do his structural engineering calculations. Apparently the paid for version is better than the free version. But still.
Sorry, what??
I hope it isn't for anything more than a shed. And even then...
You’d be amazed how many otherwise intelligent professional people have come to treat LLM output as gospel. It really is rather worrying.
I did worry that would happen. Part of the problem is people calling it "AI" when it's nothing of the sort.
The next Grenfell-style disaster could well end up being due to someone using a LLM to do something out isn't capable of doing, and not checking the results.
It’s now extremely rare for ChatGPT to get “basic maths wrong”. In fact it is increasingly rare for these things to get anything wrong - tho of course it still happens
Anyway. Good day everyone
Seems a bit quiet. Nothing happening today?
I checked recently and it told me that "raspberry" contains two r's.
As a language model it's very impressive and turns out very plausible copy, although I find the bouncy demeanor they've given it annoying, but for anything factual it is worse than useless.
This is stupefyingly wrong
Was that ChatGTP generated?
To understand the limits of chat gpt, give it a cryptic crossword clue. It will come up with a reasonably well argued answer that may be right.
And here’s the thing, if you then say the answer is incorrect and give another word it will come up with a rationale for how that word is correct. Even if the word has the wrong number of letters.
I got the sense that ChatGPT has two weaknesses. Almost by definition it provides the most likely response, which then leads (unless heavily promoted) to bland answers and secondly it priority is to sound authoritative, even when wrong.
From a human perspective this is troubling. We love a bland, authoritative, not too spicy plausible answer.
I was mildly alarmed today to learn that a structural engineer friend uses chatgpt to do his structural engineering calculations. Apparently the paid for version is better than the free version. But still.
Sorry, what??
I hope it isn't for anything more than a shed. And even then...
You’d be amazed how many otherwise intelligent professional people have come to treat LLM output as gospel. It really is rather worrying.
I did worry that would happen. Part of the problem is people calling it "AI" when it's nothing of the sort.
The next Grenfell-style disaster could well end up being due to someone using a LLM to do something out isn't capable of doing, and not checking the results.
It’s now extremely rare for ChatGPT to get “basic maths wrong”. In fact it is increasingly rare for these things to get anything wrong - tho of course it still happens
Anyway. Good day everyone
Seems a bit quiet. Nothing happening today?
I checked recently and it told me that "raspberry" contains two r's.
As a language model it's very impressive and turns out very plausible copy, although I find the bouncy demeanor they've given it annoying, but for anything factual it is worse than useless.
This is stupefyingly wrong
It really isn't. What was worse was that, after I'd corrected it, I asked it to tell me the letter position of the letter r's in strawberry. It told me that they were in positions 1, 6 and 10, and, helpfully, bolded those letters, which looked like this: strawberry
@Steven_Swinford Kemi Badenoch was left infuriated after The Times disclosed that she had appointed Robert Jenrick as her shadow justice secretary yesterday
Badenoch had wanted to announce Jenrick’s appointment today so the appointments of Priti Patel as shadow foreign secretary and Mel Stride as shadow chancellor would take the headlines
Are rallies for the faithful really the best use of a presidential candidates time?
In the UK we'd do a mix of interviews, soapboxes, meet the voters and battlebus stuff across a variety of constituencies.
Yes, it's quite a contrast to British campaigning. You never see Harris pointing at potholes, or Trump in a white coat staring at test tubes in a lab somewhere. Neither pulling pints either.
There is nevertheless a lot of door knocking and knocking up (not that they call it that), and leaflet delivery hung on door handles in the manner of a 'do not disturb' sign, as there aren't any letterboxes and they're not allowed to use the USPS mailboxes.
What's the rationale behind no use of USPS mailboxes?
That sounds very strange.
Are they allowed to use it if they send the leaflet through the post?
Only in the USA.
Even more bizarrely, even if you buy, pay for and install it yourself (conforming to a very long list of regulations on dimensions, etc.), legally it still belongs to the USPS, and only they can use it to deliver you stuff.
I was mildly alarmed today to learn that a structural engineer friend uses chatgpt to do his structural engineering calculations. Apparently the paid for version is better than the free version. But still.
Sorry, what??
I hope it isn't for anything more than a shed. And even then...
You’d be amazed how many otherwise intelligent professional people have come to treat LLM output as gospel. It really is rather worrying.
I did worry that would happen. Part of the problem is people calling it "AI" when it's nothing of the sort.
The next Grenfell-style disaster could well end up being due to someone using a LLM to do something out isn't capable of doing, and not checking the results.
It’s now extremely rare for ChatGPT to get “basic maths wrong”. In fact it is increasingly rare for these things to get anything wrong - tho of course it still happens
Anyway. Good day everyone
Seems a bit quiet. Nothing happening today?
I checked recently and it told me that "raspberry" contains two r's.
As a language model it's very impressive and turns out very plausible copy, although I find the bouncy demeanor they've given it annoying, but for anything factual it is worse than useless.
This is stupefyingly wrong
Like a pound shop Dominic Cummings when it comes to AI.
Comments
Here’s Megyn Kelly at Trump’s rally last night, endorsing Trump and mentioning both the sad cases of murder and sexual assault by illegals, but also the controversial issue of women’s sports which might also drive turnout among younger women. It’s not necessarily just abortion behind the differential female turnout. https://x.com/megynkellyshow/status/1853643950512316882
I'm not sure why @Sandpit is so smitten, Trump will throw Ukraine to the dogs, or rather the bear.
And here’s the thing, if you then say the answer is incorrect and give another word it will come up with a rationale for how that word is correct. Even if the word has the wrong number of letters.
I got the sense that ChatGPT has two weaknesses. Almost by definition it provides the most likely response, which then leads (unless heavily promoted) to bland answers and secondly it priority is to sound authoritative, even when wrong.
From a human perspective this is troubling. We love a bland, authoritative, not too spicy plausible answer.
What ChatGPT is the perfect middle manager
NEW THREAD
Embarrassing.