What we think about our leaders – politicalbetting.com
What we think about our leaders – politicalbetting.com
Every party conference season we ask people to describe the main party leaders in a word. Here are Starmer, Badenoch, Davey and Farage brought together. Not sure which one you’d rather!
1
Comments
Word clouds are very accurate except surprised to see Good outshining Racist for Farage.
Wouldn’t they still be liable to HMRC for VAT even if they didn’t charge it to their customers? Ie effectively they have discounted their prices by 16.5% which presumably wipes out a lot of the expected profit.
Chancer
Disappointing
Dunno
Inept
I leave it as an exercise for the reader to match word to leader.
That I really, REALLY, don't want to see him as a member of a Government, let alone leading it, doesn't alter the fact.
Is traitors the pinnacle of reality TV, and now the format is finished? It seems to have reduced the concept to its essentials, removing anything extraneous, besides some rather fun Crystal Maze style challenges.
it has been an "interesting", in the Chinese sense of the word, period of British political history.
Fifty years she spent voting for ‘anyone but Labour.’ Now forty years after she died, she’s about to get it. Thanks in no small part to Truss.
Horrific account of alleged Israeli torture and murder of Palestinian detainees.
https://x.com/ukads3/status/1980289590092353657?s=61
Thanks for the link on the last page, and there a couple of interesting thoughts that come out of it:
(1) Anthropic has a very different business model to xAI. Anthropic is renting GPUs from AWS/Google, while xAI is buying GPUs (and gas turbines) and building data centres. It is not clear which is the right strategy: the risk for xAI is that what is a cutting edge GPU today is not one in three years time, and therefore they're sitting on a bunch of hardware that is well behind the state of the art. The risk for Anthropic is that AWS/Google jack up their prices, and then they're in real trouble.
(2) The cost to train Anthropic models is insane. DeepSeek spent (although numbers are disputed) just $6m in compute to produce an LLM that is comparable to GPT4. The question here is about marginal spend: how much do you need to spend to produce something just very slightly better? Are there massively diminishing returns, and it's best to be slightly behind state of the art. (I would also suggest anyone who is interested check out Andrej Karpathy's Nanochat - he gives you everything you need to train up a ChatGPT 2.5 for about $100 in cloud costs. Which is *insane*.)
(3) If the Anthropics/xAIs/OpenAIs are going to need to ramp prices massively compensate for cash burn, then suddenly local LLMs look really attractive. On my base Mac Mini M4 (a $500 computer), Google's Gemini models produce better than GPT3 output at really impressive speeds.
(4) Groq (not Grok) is really interesting. They're not using off the self nVidia components, but have built chips solely designed for inference. The tokens per second you get on the open models are simply insane.
https://www.theguardian.com/commentisfree/2025/oct/19/donald-trump-xi-jinping-china-trade-tariffs-us-beijing
Relieved to be back in Stodge Towers after a day dodging (unsuccessfully) the showers. I swerved Plumpton and my friend told me the rain was biblical at times and the ground went from Good, Good to Firm in places to Good to Soft by the end of racing. I wonder what would have happened at Champions Day at Ascot last Saturday had this volume of rain arrived 72 hours earlier.
I see we are back to using "word clouds" as a meaningful tool for political analysis - just after I sold my seaweed collection on Ebay to an amateur weather pundit who has predicted fifteen of the last two cold winters and is once again telling me the snow will be falling by Christmas.
We missed last winter's cold by being in New Zealand when I was told on here there would be frequent power cuts and thousands would freeze to death - okay.
The Mail tells me we are again on the cusp of economic disaster - the markets are going to crash and we'll be on nettle soup by February thanks to Starmer and Reeves.
I'm reminded of Stodge's Seventh Law of Politics which states the only thing easier than thinking the best of people is thinking the worst of them. From Trump to Truss via Johnson and Starmer, it's easy to categorise and characterise. More often than not, it's nuanced with grey (of many shades). As I was reminded at a friend's funeral a couple of weeks back, blue was once the colour, now, I'm not so sure.
I’ll go for:
Farage
Starmer
Davey
Badenoch
https://www.2s1h.org/en
But, as with the discussion around whether Mulally is the right person to be ABC, it’s very possible to come to an otherwise valid decision for the wrong reasons.
However, as it's Monday afternoon, we'll give it a go.
Starmer - Composed
Badenoch - Confident
Davey - Relaxed
Farage - Robust
Remember Stodge's Seventh Law...
That is to say, not one iota.
Have you ever come across the Digital Orrery Project?
Basically forgotten now - https://nessie.ilab.sztaki.hu/~kornai/2023/Hopf/Resources/applegate_1985.pdf
Cliff Notes version - a shoebox sized machine that could model the Solar System orders of magnitude faster and better than super computers of the time. Custom hardware for a single task.
So you hear people saying that X can't be brave, because he's bad. Or a good leader, because he's bad.
Which leads to some strange things - such as the lack of modern condemnation of D'Annunzio, often by people damning Mussolini in the next breath. D'Annunzio was brave and quite a leader - he was also one of the founding fathers of Fascism in Italy.
The lettuce livestream has been acquired by the British Film Institute’s National Archive as part of a digital collection featuring 400 significant pieces of British online video content
https://www.dailystar.co.uk/news/latest-news/daily-stars-famous-lettuce-liz-36094044
LDs are outclassed by all the other parties in social media.
Locally unbeatable with the leaflets but nationally, - invisible.
Compare with Reform.
Bangladesh were 181/4 needing 12 runs from 12 balls to be beat Sri Lanka and they lost by 7 runs.
Including four wickets in the first four balls of the final over.
Were the Indian bookies on the blower?
https://www.bbc.co.uk/sport/cricket/live/c4gz10wpemyt#Scorecard
But to his enemies he was a lot worse.
VIRGINIA
Woking
Pizza Express
Shooting Party
[small font]
Helicopters
Pilot
Falklands
Prince Andrew rejected free embassy stays for expensive hotels
Paul Scully, a former trade envoy, said Andrew booked ‘pretty well an entire floor’ of the Mandarin Oriental hotel in Bangkok
https://www.thetimes.com/uk/royal-family/article/prince-andrew-rejected-free-embassy-stays-for-expensive-hotels-39fphrjpt
Honourable
Lolita
Orgy
Pervert
With OpenAI losing money even when charging $200/month for ChatGPT the economics really favour local processing.
She was on top of the issues and received support from across the house, apart from the usual suspects
Bye bye Non Crime Hate Incidents. Well, partially. They won't be investigated but it looks like they may still be recorded. The article is not quite clear.
I think the only final use of LLM will be some sort of initiation of a field of experimental philosophy. And it'll be as pointless as the non-experimental variety. Although of course rather interesting.
(By mediocre, I'm including its inability to generate actual triangular triangles for a trigonometry worksheet.)
In which case, where's the value in all these high profile AI companies?
(Thing 1 has this. Thing 2 doesn't, and neither do I. But meaningfully less traffic on the Superloop 2 route this morning.)
'The best argument against democracy is a five-minute conversation with the average voter.'
'
** Betting (ok, no bet, but betting nonetheless) Post **
The Greens are 42/44 on BF 'Most Seats' market.
What alarms me most is that I haven't immediately laid anything anyone wants.
Kimi picked up on his habit of saying something completely different at the end of his sentence than he did at the start at the last PMQs as did Matthew Rudd in the ST. It is becoming a meme for him which might prove a problem.
The police should be there for crimes, not alleged nasty opinions and wrongthink.
Oh, you meant to *stop* crimes?
They will continue to improve in the areas they are already useful in, but anyone expecting them to ever function like a human mind are going to be disappointed. All those CEOs rushing to replace their employees with LLMs are going to look pretty stupid.
It was a constant refrain of Cyclefree that rather too many of them confused the value of their chair with the value of their input when negotiating their salaries.
Of course if some of these rumours are actually true, it may not be Xi making the announcements.
Only AI companies developing non-LLM models have a chance. Image and video editing and generation is going to be a solidly viable business, why learn Photoshop and pay Adobe a pricey subscription when an image editing AI can do the work for you with just text prompts.
Data analysis another area where there's some money to be made, medical imaging, weather prediction, material science, etc, will all benefit from AI.