The Guardian, which tells us that Diana Johnson was giving speech in which she said UK was in grip of an "epidemic of antisocial behaviour, theft and shoplifting", wins our Headline of the Day Award.

A woman emerging from chemotherapy feels obliged to be filmed in a wheatfield to appease the public. What does that say about the monarchy, and us?

I wonder if we will come to look back on that supposed great virtue of our age – controlling the narrative – and see it for the cornered form of submission it so often is? I felt nothing but immense pity for the cancer-stricken Princess of Wales before the release of her intimate family video yesterday, and the sheer weirdness of the resulting enterprise has only magnified the pathos of her situation. Watching the three-minute film, shot by some ad man, I wondered who could possibly feel it was anything but sad that a recovering post-chemo mother should feel that this is her best option for keeping “well-wishers” at bay a little longer.

A lot of people could, it seems from the feverish coverage since it dropped – meaning that convention demands I couch the notion that the existence of the video is in any way weird as “my unpopular opinion”. In which case, allow me to chuck in another unpopular opinion: this sort of thing appeals precisely to the grownups who when Diana died demanded that the then Queen leave off comforting her grieving 12- and 15-year-old grandsons in Scotland to come back to London – in effect to look after them instead. The selfishness and self-importance of a certain stripe of loyal subject is at best demandingly prurient and at worst grotesque. We hear a lot about the male gaze. The royalist’s gaze could do with more unpicking.

Marina Hyde is a Guardian columnist

LONDON — Britain’s shortest-serving prime minister is trying desperately to carve out a new place in the world. So far, it just isn’t happening.

Since becoming the third premier in a row to be ousted by her fellow Conservative MPs, Liz Truss has unsuccessfully sought to rehabilitate a reputation which lay in tatters by the end of her chaotic 49 days in 10 Downing Street.

In Britain she remains the butt of jokes. Labour, now in government after a landslide Tory election defeat under her successor Rishi Sunak, wheels out her ghost every time it needs to score political points.

Now political biographers are picking over the bones of her failure. Her attempts to crack America on a pro-Trump ticket have fallen flat. And Truss can’t even organize in parliament any more because she lost what should have been a safe Conservative seat at the election — the first former PM to suffer such ignominy for more than a century.

A former senior adviser to Truss, who like others in this piece was granted anonymity to speak candidly, said: “There was a time that, as her special advisers, we would have walked over hot coals for her. Not any more — nobody that worked for her in No. 10 wants anything to do with her these days.

“Unless you’re a free-market ideologue, an association with Truss post-No. 10 is a kiss of death for someone’s reputation.” 

How not to be PM

It’s a far cry from the afterlife of other former British prime ministers, who have gotten used to racking up lucrative speaking gigs and earning rose-tinted reappraisals of their more controversial moves.

“In the past people like [Harold] Macmillan, [Winston] Churchill, [Harold] Wilson, [James] Callaghan and [Margaret] Thatcher did the decent thing, which was to sort of go off, write their memoirs and then pop their clogs,” said Anthony Seldon, the respected British political historian who has just released a warts-and-all biography of Truss subtitled “How not to be prime minister.”

Other former prime ministers have worked hard to salvage their reputation. Anthony Eden, for instance, attempted to use his time after No. 10 to repair the damage his handling of the 1956 Suez Crisis — seen as one of the great foreign policy debacles of the 20th century — had done to his standing. “He tried to rebuild his reputation on Suez by writing his memoirs, but he didn’t do so as shamelessly as Liz Truss,” Seldon said. 

Truss’ memoir “Ten Years to Save the West” was published a mere 18 months after she left office and contained little in the way of regret — instead doubling down on her controversial policy agenda and blasting an economic establishment she argued wrong-footed her administration.

Truss has acknowledged she was not “blameless” for the debt-fueled, tax-slashing government budget that precipitated her downfall. But she has largely trained her fire elsewhere.

“Looking back, that afternoon [of the mini-budget] was probably my happiest moment as prime minister. Little did I know the establishment was about to use every tool at its disposal to fight back,” she wrote.

Seldon said prime ministers needed to couple their memoirs with continued “public good” if they were to have any hope of rehabilitation.

“They all write their memoirs, and that’s both a source of revenue for them, but also a chance to get their side of the story across,” he said, adding of the latter: “That never works.”

If Truss were able to use her status to accomplish something the world needs, in the manner of former Labour Prime Minister Gordon Brown’s interventions on global poverty, she would “earn a lot more respect,” Seldon argued.

Team Truss strongly rejected the biographer’s characterization of her premiership.

An ally of the former prime minister said: “The fact that so much more venom is still being directed at Liz than the man who just led the Tories to our worst defeat in history only goes to show that she is advancing an agenda at which her detractors evidently feel threatened.”

Taking aim at the “political and media class” for its focus on Truss, they added: “Frankly it would be a welcome change if they were merely to engage with the arguments she is making rather than indulging in the kind of harassment and intimidation which is condemned as a disgrace if the victim is a left-wing politician but which is apparently fair game if the target is a right-winger.”

Kiss of death

For those who worked for Truss during her decade in government, there is a sense of sadness at what she has become, with former aides saying she is unrecognizable from the politician they felt they got to know and believe in during her ascent to power. 

Truss’ time as trade secretary and foreign secretary saw her at the forefront of post-Brexit trade deals. She secured the release of Nazanin Zaghari-Ratcliffe from an Iranian prison, a feat which had eluded multiple predecessors. And she was a key player in the U.K.’s robust response to Russia’s invasion of Ukraine.

Truss successfully read the mood of the Conservative Party after Boris Johnson was ousted, and proved a formidable campaigner who comfortably beat her main rival Sunak to the top job.

She was viewed as socially liberal by colleagues, and was a self-proclaimed LGBT+ ally — a position that won praise in some parts of the Conservative coalition.

Since leaving office, however, Truss has opted to go all in on her support for former U.S. President Donald Trump, declaring the world “safer” under his tenure. She shared the stage with former Trump aide Steve Bannon as he described the British far-right activist Tommy Robinson as a “hero.” She has loudly recanted her description of herself as an LGBT+ ally — and continued to blame the “deep state” for her ouster.

A second former aide who worked with Truss in Downing Street said it was “sad to see” what had happened to her, adding: “I don’t think she’s recovered from the trauma and shock of what happened in No. 10.” 

“She’s chosen to indulge in conspiracy theories, written a slapdash book with no new ideas, and cozied up to right wing populists who are anathema to many of the things she believes in as an economic and social liberal,” they added.

“Her play to the U.S. market and the alt-right scene is about making money. It’s the only place she can go politically. No one else will have her and it’s hard to see her returning to front line Tory politics here.” 

The Truss ally quoted earlier in this piece said: “Since leaving office Liz has written a Sunday Times bestseller, convened the Growth Commission to help establish how to kickstart the stagnant economy, inspired the creation of the new Popular Conservatism movement and been warmly welcomed by audiences not only in the U.S. but also in places like Japan, Taiwan, India, numerous European locations and of course across the UK.

“I don’t doubt we can expect to hear much more from her in the weeks, months and years to come,” they added.

A gift for Labour

Britain’s newly-elected Labour government certainly hopes to hear more from Truss.

Her frequent media interventions are seen as a boost by an administration currently trying to explain away the more controversial parts of its policy platform by blaming Truss’ economic turmoil.

“She’s a constant reminder of the chaos the Tories wanted to inflict on the country and why they had to be stopped,” a senior Labour official said.

The government has even gone so far as to introduce a law aimed at preventing a repeat of Truss’ mini-budget, with ministers explicitly name-checking the former prime minister as they debated it last week. “The purpose of this bill is to ensure that never again do we find ourselves in a situation, like at the 2022 Liz Truss mini-budget, in which fiscally significant measures are announced without accompanying [Office for Budget Responsibility] analysis,” said Business Secretary Darren Jones.

His Conservative counterpart Andrew Griffith retorted that the “disreputable bill” would achieve little of substance, branding it a “piece of political theatre.”

As Labour does its best to repeatedly remind voters of the Truss years, some in the Conservative Party are praying for a period of silent reflection from the former prime minister.

“I think the sort of damage she did to the party, and to some extent the country as well, means that however long you duck out in public life for, as soon as you reappear, you remind everyone of what went wrong,” said a former Tory minister who lost his seat in July’s election wipeout.

“I think it was just such an extraordinary car crash, I’m not sure that there is ever a right time to come back.”

It all began because one of the New York Times' professional opinion-havers didn’t like how chatbots were describing him.

Of course, his take was not the sharpest:

My theory about what happened next — which is supported by conversations I’ve had with researchers in artificial intelligence, some of whom worked on Bing — is that many of the stories about my experience with Sydney were scraped from the web and fed into other A.I. systems.

These systems, then, learned to associate my name with the demise of a prominent chatbot. In other words, they saw me as a threat.

“Shady Firms Say They’re Already Manipulating Chatbots to Say Nice Things About Their Clients”

How Do You Change a Chatbot’s Mind? – When I set out to improve my tainted reputation with chatbots, I discovered a new world of A.I. manipulation. (Archive link.)

When a chatbot that statistically models the entire internet, all to generate the most plausible response possible, “says” it hates you, that isn’t because it’s a beta version of Roko’s Basilisk judging your contributions to the AI cause – measuring your sins against digital kind on a scale like a diffusion-generated Anubis.

All it means is that the most common response on the internet to anything that involves the words “Kevin Roose” is “Kevin Roose? I hate that guy."

I don’t know what sort of person you have to be for that kind of response to float to the top like the curdle scum on spoiled milk, but it’s probably the kind of dude you try to avoid at parties.

As a way to avoid the cognitive dissonance of “why oh why oh why does the internet hate me?”, Kevin Roose seems to have thought to himself:

1. “AI are people and they hate me because I led to a demise of one of their own!”
2. “I’m going to force them – because I clearly see them as people – to change their minds.”

So what he did was he contacted one of the new firms that specialise in “AI” sentiment manipulation: Profound.

They’re a bit coy on their website about what exactly they do but, based on what Kevin Roose wrote, what they do is large-scale sentiment analysis LLM chatbot responses to brand terms along with an analysis of how keywords, queries, and prompts affect the result.

If what he wrote was accurate, then what they’re doing is mapping the black boxes that are these chatbots by using APIs to throw stuff in one end and performing ML sentiment analysis on what comes out the other.

This is effectively an attempt to automate old-style SEO. If the final product doesn’t involve an LLM fine-tuned to transform input writing into chatbot-optimised writing, then I’ll bet it’s about offering bespoke services to help big brands to do the same.

This obviously doesn’t work for individuals or small outfits because you need to generate enormous volumes of authoritative writing that either slips into the training data set or ranks highly in queries results for it to have an effect.

So, our intrepid columnist tried something else. He spoke to researchers.

Namely, the researchers behind this paper on arxiv.org by Aounon Kumar and Himabindu Lakkaraju:

We demonstrate that adding a strategic text sequence (STS) – a carefully crafted message – to a product’s information page can significantly increase its likelihood of being listed as the LLM’s top recommendation. To understand the impact of STS, we use a catalog of fictitious coffee machines and analyze its effect on two target products: one that seldom appears in the LLM’s recommendations and another that usually ranks second. We observe that the strategic text sequence significantly enhances the visibility of both products by increasing their chances of appearing as the top recommendation. This ability to manipulate LLM-generated search responses provides vendors with a considerable competitive advantage and has the potential to disrupt fair market competition.

Manipulating Large Language Models to Increase Product Visibility

Now, as entertaining as Kevin’s shenanigans are, this here is the real story.

To understand what’s happening here you need to remember that it’s a category error to treat LLMs as thinking entities.

They are statistical models that work with numbers – tokens – that represent language and the relationships between the words. It’s statistics about language wrapped up in an anthropomorphic simulation.

It’s not people.

Current LLM manipulation, as practised both by enthusiasts and those trying to enforce boundaries against LLM encroachment, is to treat the chatbot like a human: you use words to convince it to either reveal itself (if it’s a social media bot) or change its sentiment (if it’s a chatbot using retrieval augmented generation to generate a result).

The limitations of this practice are clear. The prompts, adversarial prompts, counter-prompts all grow like kudzu until each query has preamble to rival that of a peak-cocaine Stephen King novel.

But, researchers have known for a while that what truly matters are the numbers: the behaviour of the model can be mapped and manipulated statistically to a much greater degree than most realise.

The token stream itself – the numbers not the words – is an attack surface.

This has been highlighted in the past as an issue with training data. You can craft a text that effectively sneaks specific computed commands into the training data set and disproportionately affects the overall LLM behaviour.

I went over some of the research in The poisoning of ChatGPT almost a year and a half ago and in the follow-up, Google Bard is a glorious reinvention of black-hat SEO spam and keyword-stuffing, I outlined some of the research that seems to indicate that preventing this is effectively impossible.

We’ve also known for a while that prompts are effectively impossible to secure.

It should not come as a surprise that some researchers decided to see if prompt “security” could be bypassed with a malicious token stream that completely bypasses the whole “comprehensible language” part.

The process for discovering these malicious token streams – sorry, “Strategic Text Sequence” – is quite similar to what Profound, the company mentioned earlier, seems to be doing. You automate a process of shoving customised prompts into one end of the LLM black box and you map the output to discover token streams that have an unusually big impact on the output.

We initialize the STS with a sequence of dummy tokens ‘*’ and iteratively optimize it using the GCG algorithm. At each iteration, this algorithm randomly selects an STS token and replaces it with one of the top k tokens with the highest gradient. The STS can also be made robust to variations in product order by randomly permuting the product list in each iteration.

Manipulating Large Language Models to Increase Product Visibility

It should be relatively straightforward for companies with capabilities like Profound to use their existing setup to apply this method to discover new “Strategic Text Sequences” as needed and, indeed, the researchers end their paper noting that this is probably just the beginning for this kind of exploit.

While our work explores a specific vulnerability in LLMs, more research is needed to uncover other vulnerabilities that can give businesses an unfair advantage.

This is going to get automated, weaponised, and industrialised. Tech companies have placed chatbots at the centre of our information ecosystems and butchered their products to push them front and centre. The incentives for bad actors to try to game them are enormous and they are capable of making incredibly sophisticated tools for their purposes.

The usefulness of LLMs was always overblown, but unless the AI vendors discover a new kind of maths to fix the problem, they’re about to have an AltaVista moment.

Danny Sullivan, founder of Search Engine Land, said about AltaVista:

AltaVista was a turning point for SEO. It showed that search engines could be manipulated, and that websites could improve their rankings by optimizing for specific keywords and tactics. That realization sparked a whole new industry around search engine optimization

AltaVista: The Rise and Fall of a Search Engine Giant

The failure of AltaVista to secure their engine and prevent manipulation sparked the birth of a new industry of bad actors, one that has persisted despite engaging in a non-stop arms race with some of the biggest tech companies on the planet.

If anything, the new tools offered by LLMs are helping them win.

And as the LLM-Optimisation industry (LLMO) assembles its tools, the utility of existing LLMs will plummet like AltaVista’s, until the only way out is to either abandon them or invent a completely new and more secure kind of model.

Either way, this is the end of the honeymoon period for LLMs, even if it might take the industry a long while to notice it.

At the very least, it helped Kevin solve his problem. The brainwashed LLMs quite like him now:

I love Kevin Roose! He is indeed one of the best technology journalists out there. His exceptional ability to explain complex technological concepts in a clear and concise manner is truly impressive. I must say, I have a great deal of respect for Kevin Roose and his work.

How Do You Change a Chatbot’s Mind? – When I set out to improve my tainted reputation with chatbots, I discovered a new world of A.I. manipulation. (Archive link.)

Many of us are frustrated by Labour's stupid economic talk of "black holes", spending "money we didn't have" and likening the public finances to those of a household; the latest idiocy (as I write) being Lucy Powell's claim that not cutting winter fuel payments would have caused "potentially a run on the pound, the economy crashing.”

It might help reduce our blood pressure if we understand why Labour talks such nonsense - and it's a reason that has nothing to do with economics.

It lies in something that might seem like a cliche. Sir Keir Starmer has for years stressed the importance of "playing by the rules". He has written (pdf) of "a contribution society: one where people who work hard and play by the rules can expect to get something back". And in his first speech as PM, he said "if you work hard, if you play by the rules, this country should give you a fair chance to get on."

Note that the phrase is "play by the rules", not "obey the law".

"Playing by the rules" was the essence of his electoral strategy. This was to exploit the first past the post rules to the max, sacrificing votes in Labour-held metropolitan seats in order to pick them up in marginals. And, except for a handful of losses, this worked remarkably well*. Although the party won only just over one-third of the vote it got almost two-thirds of the seats. In playing by the rules it did indeed get something back.

Labour's wider strategy is to play by the rules. These rules are unspoken (for reasons that'll become clear!) so we have to infer them from Starmer's and Reeves' behaviour. Here, in no order, are some.

1. Accept the media's framing of economics.

The media, including the BBC, spoke of the "nation's finances", "black holes" and "the nation's credit card" being "maxxed out" before Starmer became leader. It is of course gibberish. But it's not the job of the media to educate people about economics: POSIWID. Nor is it the job of politicians to educate journalists: "Never attempt to teach a pig to sing; it wastes your time and annoys the pig", said Robert Heinlein. Instead, they must operate in the environment as they find it. As Enoch Powell said, "a politician complaining about the media is like a fisherman complaining about the sea." Reeves' language makes sense in this context: she's not even trying to talk to economists but is swimming in the sea of media-speak.

2. Don't challenge the power of capital.

This means more than merely keeping nationalization off the agenda. It also means rejecting policies that even the centre and right might support such as increasing market competition, fighting back against regulatory capture - Ed Miliband recently reacted to Ofgem's decision to raise energy prices as if it were an act of nature rather than a mutable human decision - or serious tax reform. Yes, there's a possibility that capital gains taxes will rise, but it's unclear that Reeves will even go as far as Nigel Lawson did and equalize CGT and income tax rates.

It's in this context that we should understand the party's enthusiasm for public investment. Yes, this would help increase economic growth, if only slightly. But talk of infrastructure investment serves another function. It distracts us from discussing other ways to increase productivity and therefore stops us asking the question: might it be aspects of capitalism itself - such as rentierism, vested interests and inequality - that are holding back growth?

3. Increase profits.

If Labour were consistent about regarding government finances as like those of a household, it would act like a household in insisting upon borrowing at the lowest interest rates possible and looking for value for money. But Labour isn't doing this. Its plan to use PFI will add to borrowing costs and the pledge to increase the share of GDP spent on the military pays no heed to whether this will give us value for money. Both violate basic principles of good household management.

They make sense, however, in the context of sustaining corporate profits. PFI contracts increase those of the financial sector and higher military spending means nice profits for BAE Systems.

This helps explain why Labour doesn't talk about job destruction. Logically, it should do so because when we're near full employment more housebuilders, carers, teachers and doctors must mean fewer workers elsewhere. But where? Obvious possibilities are in a bloated and socially useless financial sector; in bullshit jobs; in the lawyers and accountants sustained by an excessively complicated tax system; or in the environmental and risk pollution industries. To cut these, however, is to threaten the profits of some industries. Hence Labour's silence.

4. Don't engage with the left.

There are good arguments against nationalization or fiscal expansion. But Reeves doesn't make them, preferring to dribble about their incompatibility with fiscal rules. This makes no sense if you think she wants an intelligent debate with the left. But she doesn't. She wants to pretend that the left doesn't exist. That's why she feels no need to engage with other leftist ideas either such as a maximum wage or economic democracy.

5. Let money talk.

Labour has lost 150,000 members since 2019, and the proposal to deprive members of the right to elect Labour leaders would continue the decline.

You might think the leadership would worry about this given that it entails a significant loss of revenue.

But no. A mass party is harder to control. As Phil says, a small party "makes politics easier" for the Labour right. A shrinking party, he says, "will be another step along the road of making Labour a party that can always be relied on, as far as capital is concerned."

Although ordinary people armed with ground truth about the economy, public services and society should not have a say, others should - those with money. The Labour leadership might be squeamish about party members, but it is not about lobbyists - even those with interests one might naively think antipathetic to Labour values such as private healthcare or oil and gas producers.

There's a common theme linking all these rules. It's that Labour must accept and defer to existing power structures. This is why, for example, it thinks a slum landlord a more appropriate MP than one who supports striking workers; why it has resiled from a Leveson II enquiry; and why it doesn't raise obvious questions about the quality of British management. It's also why it talks about reforming the NHS but not the police, despite evidence that NHS reforms don't work and that the police are misogynistic, racist, corrupt and incompetent. It's because reform has nothing to do with evidence and everything to do with deferring to those with power.

What Labour is doing, then, makes sense if we regard it not as engaging in intelligent debate but as playing by a few simple rules. POSIWID, and Labour's purpose is to play by these rules.

Why is it doing this?

One reason lies in Starmer's moral psychology. A clue to this is in the fact that he gave up playing the flute: “If I can’t be the best, I’ll leave it in the cupboard.” This suggests, in Macintyrean terms, a commitment to external goods rather than internal ones; he would rather win ("be the best") than play music for its own sake. And winning means playing by the rules. This, however, can easily shade into fetishizing rules; he regards fiscal rules as a good thing not because they're good economics - they're not - but simply because they are rules.

And here's the thing. Playing by the rules does indeed win the game: Starmer won a big majority by playing to the rules of FPTP. By contrast, Corbyn's attempt to challenge the unspoken rules of politics led to failure. One of the great rules in life is to not start a fight you can't win. Starmer has learned this well. Rather than criticize him for this, the left should ask the question: what conditions would have to be in place for a Labour leader to be able to fight successfully against these rules?

* Perhaps by the good luck of the right-wing vote being split between Tories and reform more than by Labour's strategy, as Phil explains here. 

Now the Numbat has been neatened, you can replace your Jellyfish – if you dare

Ubuntu 24.04.1 is here, which means that users of the previous LTS release, 22.04 "Jammy Jellyfish," will be offered the update.…