Policy Sketchbook
@policysketch.bsky.social
A blog on UK public policy. https://policysketchbook.wordpress.com/
"Flooding the market with low-quality early uses" is a great summary of the problem with AI as it currently is. You don't need to be sceptical about the potential of the technology to observe that to date it has been terrible at many of the things it claims to do.
Flooding the market with low-quality early uses, relying on political patronage (the deals various companies have made with Trump) to bypass the normal checks and balances companies are held to, all while talking like they are science fiction characters....
at the 'stack, I argue for the indispensability of the human in the loop backofmind.substack.com/p/because-i-...
because I say so, that’s why
a loop cannot close itself
backofmind.substack.com
The problem can be summed up with one question: how many people who can name Jason Arday can name the plagiarising professor at the same institution who is still in post? How many people can name the VC criticised by name in an employment tribunal this week?
Ok, some thoughts, as an olde tyme academic, on the Arday tragedy. First, allegations about causes. We don't know what was in his mind, but many people are keen to blame him, or woke lefties/academia, or the media. Some people are absolutely certain that the media are faultless. 1/
Or much of a drop in public support. Fascinating example of elite consensus switching despite popular opinion.
In two years Net Zero went from broad political consensus to politically toxic, without any particular scandal over the policy.
Meta AI Glasses but for parents to wear all day every day to document every interaction with their child, rate its difficulty and produce a comparable score to solve every argument ever about how the domestic workload is shared.
By far the most terrifying opening paragraph I've read in an FT piece giftarticle.ft.com/giftarticle/... omg @soumayakeynes.ft.com
NHSE strategies have neglected building the foundations of improvement, negating the ambitious goals of their strategies: www.hsj.co.uk/daily-insigh...
The mythbuster: Saying it is not doing it
Repeated cuts to NHS management, an explosion of central mandates, and a chronic shortage of digitally skilled staff have left local providers unable to drive the productivity gains the health service...
hsj.co.uk
It is an interesting argument but makes the mistake many do when opposing good tax reform: every tax has to be fair by itself and not create losers. this is a barrier to any change but misses the point...
In the same way that farts only smell bad when someone else has done them.
The hill I will die on: Radio 4’s Today programme has become really annoying since I left | John Humphrys
A case of mistaken identity, I fear. Count Binface has a bin on his face, not a face on his bin. This is the family of Count Facebin.
Just when you think they can't stoop any lower, sky news has doorstepped Count Binface's family
A greatly expanded version of this thread, with interactive charts, and links to all sources/references is here: buff.ly/YvWBJlK
37 ways Andy Burnham could raise £4.7bn
Andy Burnham may need to find £4.7bn for defence. Here are 37 possible tax rises and reforms.
buff.ly
Spotting the opportunity to make the argument that if you are willing to pay that much for the worst AI company, surely you will pay just as much for the best one?
Breaking news: The AI company has confidentially filed for a $1tn-plus initial public offering, setting a three-way race with OpenAI and SpaceX for potentially three of the biggest listings of all time. ft.trib.al/o8iOlMa
IMHO possibly the most interesting question in finance and economics right now is the extent to which future AI dominance can retain any monopoly characteristics. People assume because it's "Silicon Valley" that those explosive network effects will necessarily pertain. They really might not... 1/
Without claiming any expertise in the problem in question, I think it's notable that OpenAI says that it was solved by "a new general-purpose reasoning model" i.e. not by an LLM. This makes sense as I do not think LLMs are even in principle able to solve novel problems.
OpenAI's claim that this is a central conjecture in discrete geometry is not an exaggeration. This will I think be looked back on as the first time that AI solved a major mathematics problem (defined as a problem that all experts in some subfield had thought about). openai.com/index/model-...
I have been thinking about this point a lot over the past couple of days. Leadership is hard and you may need to compromise on some issues for the sake of the bigger picture. But having the judgement to know when you have triangulated yourself into unacceptable positions seems important.
I will never understand how Keir reconciles his self-conception as a good man doing a good job as prime minister with “my premiership had to be brought to the brink of death for me to say this”.
I will never understand how Keir reconciles his self-conception as a good man doing a good job as prime minister with “my premiership had to be brought to the brink of death for me to say this”.
Prime Minister public message ahead of the Unite the Kingdom rally in London today, organised by Tommy Robinson and others.
Dear Prime Minister, I am great and therefore should remain in my job. However, I no longer think you are great and therefore I cannot. Yours sincerely, Wes
BREAKING: Wes Streeting quits as health secretary
Social care reform, immediately. Bon the commission, give task force 3 months to propose final recommendations building from Dilnot proposals, implement immediately with penny on basic and higher rates of income tax to pay for it. Immediately transforms local govt fortunes too…
Can we start compiling a list of every policy you'd like to see a new Labour leader make? Big or small, what would actually start to make the country feel better for people?
The defence of the FPTP political system has always been, in effect, that it is unstable. An increase in vote share can have a non-linear effect on the share of seats, so a swing to one party can quickly snowball into a large majority. This is apparently good because it leads to strong government.
Power works best when it's not exercised. In the abstract, we equate "the strongest army in the world" with absolute power and act accordingly. When put into practice, the limits of that power become clear and it seems somewhat less impressive.
1/ We got 50,000 articles about how weak and impotent soft power has proven in the 21st century. Meanwhile, the "1st army in the world" can't bend Iran to its will and the (self-proclaimed) "2nd army in the world" is over a thousand days into its 3 day special military operation.
I can't remember whether @dsquareddigest.bsky.social covered this exact angle in his book, but there's a kind of accountability sink that's created in law by the government to protect people and organisations from the consequences of their actions, in order to encourage risk-taking and innovation.
TW: Not new news but I reported this blatant hateful and anti-Muslim propaganda on Facebook. Facebook’s response today…nah, ok by us. Disgraceful.
Headline measures of wealth inequality haven't meaningfully changed since the 1980s, but it's harder to work and save your way up the ladder than it used to be. Meanwhile, inheritances are becoming ever more important. Me for this weekend's Observer: observer.co.uk/news/capital...
This is confusing symptom with cause. Labour is repeating the mistakes of the Conservatives in trying to win voters from the far right by chasing anti-immigration headlines. That's only serving as promotional material for Reform and is (part of) the reason the leaderships is under question.
Heidi Alexander on Sky: "Those people who think that we should have a leadership election now, and repeat the mistakes that the Conservative government made in churning through prime ministers, probably do need to give their head, a gentle wobble." A not uncommon Labour sentiment*.
This works on multiple levels, because even the bit that RFK presumably thinks is right is actually wrong. $100 rising to $600 is a 500% increase. Just great Content all round.
RFK Jr: "A Democratic senator claimed it's mathematically impossible to have a drug drop by 600%. I said, 'Well, if the drug was $100 and it raises to $600, that would be a 600% rise. If it drops from $600 to $100, that's a 600% savings.'" Trump: "Right"
This is a very insightful comment about how we shape our careers. It seems obvious to me that you'd want to have a career that sees you do the things you are good at and avoid the things you are bad at. But maybe this is why I will never be PM (or e.g. CEO of a large company).
Think an underrated part of why Starmer is Like That is that like most very successful people, he has had a long career of “I can reshape my work load to avoid my areas of weakness”, and you cannot be an effective prime minister on that basis.