Imagine what could have been possible with a skilled political operation that had had said ‘right, we’ve got a huge majority and 4-5 years to use it’
After seeing Johnson/Tories and Starmer/Labour waste opportunities they were given we shouldn’t be surprised if the electorate doesn’t again
4D chess is the version where you shove the pieces up your nose and smash the board repeatedly against your face then yeah?
Donating to online scammers posing as legitimate people in need is not morally neutral:
You potentially are giving money to brutal slavery and human trafficking networks:
www.ohchr.org/en/stories/2...
Confused response as President shouts “Strait of Hormuz… mine!”
This is cute
youtube.com/shorts/0f6hd...
A very conservative approach. May I suggest the socialist alternative of having a no-vax-tax? 😉
Some drink to forget. Marvin cannot drink…. or forget…
What flavour?
100% agree. For one thing it is literally impossible to factor in tactical voting on such models as even if voters are asking about local conditions they have not and cannot respond to these as they would on an election. And tactical voting mattered a *lot* in 2024.
Agree it’s part of it. Also probably that the group of people who are populist curious are I think of the view that they’re all a bunch of lying bastards so I may as well vote for the lying bastards that say other things I like (or they convince themselves the incoherence and lies don’t exist)
Shades of Poland, 2023 in this, where turnout also surged, in that case as people came out to vote out PiS.
Excellent thank you!
Staunch is somewhere between supper and lunch
Oh, which is pharma Derek?
Something I didn’t touch at all during the election were the MRP polls – “Multilevel Regression and Post-Stratification” to give them their Sunday name. To a certain degree that was just because I had so much else going on that reporting on yet another thing would be the straw that broke the camel’s back. But as the campaign went on, I became increasingly convinced that MRPs had become a net-negative for our political discourse. If you’re reading BBS you’ve almost certainly seen an explanation of how these work, but the simple summary is they take a massive sample size, figure out how different groups of voters are voting, and then apply that demographically to every seat in the country to get an overall picture. When you combine this explanation with the very fancy sounding name, MRPs ended up with both a cachet that ordinary polls don’t and capturing a lot of attention with their promise of being able to give credible individual seat projections. Yet in Scotland, a lot of the MRPs were spitting out things completely at odds with the general state of polling. These included an early one with the SNP on 37 seats despite trailing Labour in ordinary polling by 5%; Labour winning Berwickshire, Roxburgh and Selkirk; and the SNP winning Orkney and Shetland. All of these were obviously wrong, but spread like wildfire across social media. In Scotland, most of the final MRPs overestimated the SNP and underestimated the Conservatives, with the notable exception of Survation who were almost spot on.
Compared to traditional and simple models (on BBS I effectively average out uniform and proportional swings), the problem with these is you can’t explain them easily. When my model gives something odd, we can all go “ah, that’s just because the model is giving too much/little swing there.” With MRPs being much cleverer, the individual predictions are given more credibility by a lot of people, yet when they get big misses like the examples above nobody outside the polling firm in question can explain why. Given we have an FPTP system, an urge to find out what is happening locally is understandable. However, in meeting that urge, I think MRPs have ended up going too far towards giving people the illusion of certainty – as too did the exit poll’s individual seat predictions, which had the Scottish Conservatives on an absurd number of seats the party was then forced to spend the entire night expectation managing downwards. No amount of careful caveating changes what happens with polling output when it gets in ordinary people’s hands, so we need to think about what goes out there. At the next election, I think MRPs need to get better at modelling Scotland, and to resist the siren call of clicks and media coverage they get from publishing individual seat breakdowns, just publish the headline seat figures
Nothing about MRPs since 2024 has given me any cause to revise this position
If I could drink I’d be there in a shot
Hope it goes well. Or goes terribly but you learn something interesting
I have a feeling you’ll be v v good at this before too long
Staunch is somewhere between supper and lunch
Carefully replying to every post I come across with "I have no interest in or knowledge of this topic" to make sure I have had my say.
Well it looks great. If it tasted half decent too that’s v impressive
Look forward to hearing about your brewing adventures
Indeed. It’s not like me sharing would make the slightest bit of difference either
Tempted to share this to see how much Bluesky misunderstands
Farage always was a little pivot
I live in a permanent state of questioning and confusion so no worries there :)
You’re probably right. Could be some interesting ripples
Self doubt is useful but sometimes time consuming
That confused me at first. I ran a test run without posting and saw ukraine mentioned but assumed the summary was a hallucination. Spent a while debugging and saw there was a post, but couldn’t see it manually in the bluesky app, while I was trying to work out where it was coming from it appeared!
Thanks! Good idea
I have it fully setup under @ftl.invalid-handle.com to create digests of fintwitter as that gets very noisy and repetitive with 100+ posts an hour at peak
Very handsome 😊 and a lovely photo!
a female superb fairywren standing on top of a leaf
did I show you this wren
so tiny she can just hang out on a leaf
Nice! I’ve been playing with something to summarise feeds, perhaps useful for very busy ones, okay if I try it with this? Maybe more suited to your news lists, but it’s only working on feeds atm
It produces output like this. Always a risk of hallucinations, but potentially useful
