Lew

Coffee, straight out of my nose.

I am just noting that NZ National confusion over Iran is deeper than just the serious Prime Minister’s leadership failure.
I lost track of the number of parliamentary questions National filed to try and stop a Kurdish refugee & critic of the Iranian regime when Behrouz Boochani sought NZ asylum

How fortunate we are to get such dollops of barbarism with all the barbarousness

Walking to the Senate expecting another day of driving impact, aligning stakeholders, and scaling the Roman Republic to the next level. Instead, I encountered unexpected feedback.
But what I would say to you is not everyone will share your vision, and the only soothsayers I listen to are English.

But an LLM has none of those evolved structures. Why would it? It’s a pure engine for linguistic symbol manipulation and (in a suitably pared down sense) “reasoning.”

My conviction that the cat on the chair next to me has some sort of subjective interiority, however different from mine it might be, is not materially affected by her inability to summarize Tolstoy or solve LSAT logic problems.

If you truly believed Graham Platner is not antisemitic at all, you would be sending him to synagogues in Maine to speak to Jews and reassure them of that. That is what Mamdani did during the NYC mayoral election. The fact they are not doing that is... telling.

Also, when you ask @attackerman.bsky.social why everything is awful, you have to let him cook

My brother in Christ, you just sacked 30%+ of your reporters

I thought there were no ways this could get funnier but here it is, they're getting the lads from TopGuer or The Boris Club or something to unskew the polls so he doesn't have a crisis of confidence and shit the bed on him

How fuckin good

Given how badly Labour is doing in England it's no wonder he prefers UK polls

I actually have really hot polls, but you wouldn't know them, they're from the UK

Polling is a catalyst for industrial-scale narrativium production, but that doesn't mean it's not useful for other things (like platinum, or mercury, or I suppose caesium depending on how keen you get about it)

The best time to join "all polls are bad" is when you love them, when they are singing their siren song to you, promising the future you've always dreamed.

I like this but it presupposes that the firm is loyal to the current National party leadership and not to an insurgent far-right takeover that it is trying to happily foment

Criterion of embarrassment - Wikipedia

Firm ecclesiastical basis

en.wikipedia.org/wiki/Criteri...

It's not a mathematical law that can be disproven with a single exception case. It's an assessment about the validity of a messy but well-understood way of measuring something that is simply very hard and expensive to measure

This is wrong. There is evidence to show that Curia's political polling is generally good, because it tells us more or less what all the other polls tell us, and all that data validates well against objective reality each and every time we get the chance to test it

That's what evidence looks like

And it's odd to think "I don't like liberals" is the motivation here, since I am a liberal and I get a lot of shit for it from the lefter-than-thou brigade who think libes like me are the left wing of fascism; but more specifically because my decision is made on statistical, not ideological grounds

It is not equal, because that is not remotely the claim

The public record of Curia with RANZ is sufficient for me to discount their issue polling (especially in the absence of other polling ot objective reality to calibrate issues polls against)

But none of that applies to the PV poll

A brown penguin chick of some kind. It looks very much like a man in a suit. It is bedraggled and miserable

Made it to Friday but at what cost

Really not possible, given the extent to which the sausage-makings are trade secrets and the cross-tabs and deeper data series are mostly commercial-in-confidence. But

So we're all the way back around to "all models are bad but some are useful"

Because "I don't like DPF or the TPU" is not good counter-evidence, and if it cannot be anchored in anything in the actual data, it is no evidence at all about the data

And leftists need all the help we can get understanding reality, so we should not discount good sources due to superstition

And I don't need to prove anything, I just need to decide for myself whether I treat it as a reliable data source to help understand reality

That is why I criticise leftist poll truthers: because there is good empirical evidence for Curia being a good poll, and practically no good counter-evidence

It is presumptively correct to the extent that other polling is correct, and we only get one opportunity to validate that against observed reality every three years

But all of the major polls perform strongly in that validation, so it's really on critics to point out where they see bias in the data

Where is the evidence then? If the claims about Curia polling are justified then someone must be able to show me where Curia polling is wrong

How would you measure voting intention at population level without an election?

This is not a pure objective static data series like temperature stations or card spending data or new vehicle reg

All we have is polls, and all we plausibly can have is polls, so we must calibrate them against polls

I think this just misunderstands people like DPF and Jordan Williams in the typical caricatured way that most leftist diehards misunderstand them

We rely on the long history and good reputation of pollsters and their clients (and the huge commercial value embodied in that) to discipline each other, and they do, so much that when something comes out that is just made-up numbers (like TOP's Ilam "poll" from 2023), we can all see it easily

The answer is "if one of them is inaccurate then they are all inaccurate, despite the fact that they accord well with observed reality every three years" and the only explanations for that really are that they are accurate, or that there is a grand polling conspiracy to cook the numbers up together