abe94

Its a bit more complicated than that quote implies

At posthog we see if a customer is using an llm, they use more than 1 model. The typical pattern is frontier models for a small percentage of 'harder' tasks and then one of these chinese models for more standardized procedures. As you get better at standardizing procedures you are able to use the chinese models for more and more work so token usage goes up, but the $ spend on top models has still been growing

I've been seeing a lot of usage of the word `real`

from recent fable sessions:

- That gap is the real story: 3,873 flows

- the real conversion filter types are:

- I'll update the breakdown query to include a column for each real type

An aside but there is a wonderful article on how the swiss watchmaking industry won the "brand age" here - https://www.worksinprogress.news/p/why-we-still-have-mechani...

related discussion here - https://news.ycombinator.com/item?id=47008607

here is the government report - https://assets.gov.ie/static/documents/b87d2659/20250929_BIA...

The cost benefit analysis includes a euro value to attribute to better wellbeing, using the WELLBY framework and apply £13,000 per WELLBY

Very cool - I've wanted something like this for a while. I currently use a patchwork of site specific extensions, so will definitely give this a go

Something in a similar vein that I would love would be a single feed you have control over, powered by an extension. You can specify in plain english what the algorithm should exclude / include - that pulls from your fb/ig/gmail/tiktok feeds.

You can pick and choose what you see from the audit (though in no circumstances is it an astounding success).

The general points i'm trying to get across is

- Responder safety/back-up needs mean you can’t fully “swap out” police. This program still needed to bring in police most of the time. - Coverage and scale are hard in an actually big city, like NYC. (also why denver's success in a tiny city is sort of a useless comparison)

So why not just equip police to better handle mental health cases instead of creating a different task force which doesn't have any of structure the police already has? This isn't rhetorical - the reason is idealogical stubbornness, there are better solutions for achieving mamadanis goals.

Which is an issue.

Mamdani and politicians with similar 'progressive' positions are those advocating for reducing them (either through budget cuts, or moving resources to worse initiatives), while the most robust studies on crime have shown beat cops to be pretty effective for reducing crime (see the Philadelphia Foot Patrol Experiment, and other meta analyses)

Here's an account from the NYC government itself on how it hasn't worked well - https://comptroller.nyc.gov/reports/audit-of-the-behavioral-...

A liquor store and a grocery store aren't the same thing, they were set up for different reasons, have different policy goals and are run differently. Also if the goal is helping with "food deserts" - which seems like a tenuous claim already for a city with a ton of bodegas there are better less costly solutions. When comparing actual state run grocery store pilots (in the US) they have been a disaster - see the kansas city example here (https://www.kcur.org/news/2025-08-12/kansas-city-grocery-sto...)

Denver is the best example in the US that saw limited success, and one of the very few - in most places similar approaches were tried (in the US), we were worse off then just keeping police as the primary responders. In NYC specifically we tried a pilot recently, and it was (unsurprisingly) ineffectual - where police had to be called as backup 65% anyway(https://comptroller.nyc.gov/reports/audit-of-the-behavioral-...). So no in NY at least police don't make the situation worse thats a myth.

Maybe the subsidized transit is normal in europe, and universal childcare in some parts of europe (definetly not all of western europe), but this article is stretching when claiming state run grocery stores are normal.

Its also conveniently leaving out the policy ideas on reducing policing, and introducing mental health crisis workers which have been tried in the US (SF) and worked disastrously.

The new dan brown book uses this as a central concept in the story

I went to a middle eastern wedding recently and they gave everyone these phone pouches to keep their phones in that were locked for the event's duration.

Honestly made the whole event better

https://skudinsurf.com/ may have something for you. They also do beginner sessions in a mall - https://skudinsurfamericandream.com/

Great read,

One question i've had is if its given that we care about making humans space faring, why focus on getting to new planets and colonisation and not instead on building the massive megastructures mentioned in the article as an end in itself?

i don't know - I agree we haven't seen changes to our built environment, but as for an "explosion of new start-ups, products" we sort of are seeing that?

I see new AI assisted products everyday, and a lot of them have real usage. Beyond the code-assistants/gen companies which are very real examples, here's an anecdote.

I was thinking of writing a new story, and found http://sudowrite.com/ via an ad, an ai assistant for helping you write, its already used by a ton of journalists and serious writers, and am trying it out.

Then i wanted to plan a trip - tried google but saw nothing useful, and then asked chatgpt and now have a clear plan

very cool - what was the hardest part of building this?

I keep forgetting this is the same Mark Carney that ran the bank of England

"It did not disclose the reason behind the error."

is this a right sizing move - correcting over hiring in the pandemic, or google reducing investment in this area to increase investment in their AI efforts?

I'd hope the writer or editor used lunacy for this reason haha

not so sure about that after the last few Daniel Craig movies, can't get much worse, they just need a coherent plot line, and need to stop doing multi movie arcs

Not just for vietnam, its a risky proposition to have a foreign company supplying the US military becoming a core communication provider in your country.

The race for national clusters of chips continues -

I wonder how these investments will play out (ie what the ROI will be) if open source models keep getting released that are close to SOTA

a moby dick sequel in the making

can coming into contact with a single transformer disrupt power to the entire island?

archive link?

This is a bit of clickbaity article, the tweet in question was comparing the unreleased model's performance in a negative light against competitors.

Also a number of other tech companies have similar policies in place for speaking about unreleased products

yeah pretty cool to see how they can better use a by-product of lithium extraction

Expanding on this, from what i understand there aren't too many labs (understandably) that are able to grow extremophiles in lab conditions.

It would be great to run a bunch of experiments at various different conditions and see minimum and maximum on a variety of variables like salinity, acidity, heat, etc