lifeisstillgood

Wait - how? I thought the whole reason the darn things are lower cost than equivalent PC monitors is the smart mode tracks viewing habits and seeks that to data brokers?

I think at some point we have to give up the “we are creating AGI” and admit it was “a better search engine than Google, we can cash grab their ad revenue”

The 10 most famous castles in the world

Palace of Versailles France

Forbidden City People's Republic of China

Neuschwanstein Castle Germany

Buckingham Palace England

Alhambra Spain

Tower of London England

Windsor Castle England

Kremlin Russia

Red Fort India

Topkapı Palace Turkey

So I could name recognise the top ten, and yeah I won’t argue too hard with the list, just apart from Neuschwanstein - but the second I saw it I realised I did not know the name but “Bavarian fairy castle” clearly deserves its ranking

Meh. That’s like saying we have a parking shortage in cities. No we have more car journies than we need.

We have societies designed by the default choice. If that choice is walkable streets, low cost electric buses, dense neighbourhoods (mostly I mean you can walk for miles along streets and parks and shops without crossing a car park)

Then you get far less car use. People aren’t stupid, but living in downtown Houston means you have far less choice about driving everywhere than living in suburban Amsterdam

It’s all choices

We just are making bad ones mostly

The choices now are would you like ads or more ads. Shall I waste compute seeing if the web page you clicked on can be summarised ?

The simple answer to AI is to charge it at cost - not subsidised. Then the market will start to shake out.

It might take the US stock market with it …

>> Have a strong anti-counterfeit strategy. Don’t overlook this.

Can you expand? What counterfeits do you see (literally someone selling boxes labelled jamcorder? How do you handle that? Lots of scanning amazon and sending cease desist letter?

It’s a bit “follow for more”, but I do like the implications for industrial policy - a focus on long term, a means to support eco-systems not specific companies and ability to focus on B2B less B2C

I get that people object - but I don’t agree. I walk around with my face and my fingerprints and I shop and I use public services. I don’t go masked, and I don’t jump over the turnstiles to use the services for free.

Having my face “online” should be no different.

It’s just we don’t trust Amazon or Twitter or 1000 random data brokers. And you know what - we should not trust them. We should regulate them.

And on a purely national security level, a passport to participate in online discussions is reasonable - look at metrics around Scottish independence postings online - they dropped 25% the second the 12 day bombing began last year and then cam back on line slowly.

I will support to my last breath the right of a Glaswegian to say the English should fuck off, but I object to Iranian or Russian bot farms doing it. And poor sods in Indonesian bot farms should not be able to scare Glaswegian Grannies into handing over their bank details.

So, yes loss of privacy is bad, but privacy is not secrecy - my neighbour knows my “secrets”, he is just not trying to sell me something on the back of it.

I struggle with this.

Wasting political capital on “ID cards” in Britain is a sensible choice for and incoming PM

But a government digital ID is a fantastic basic utility every citizen will need. I would think “just” adding secureenclaves to passports - that seems a low impact approach but have not got a clear plan (any pointers gratefully recvd)

As in they can get out of the obligation to build the DCs? It seems the obvious way out …

I don’t think there is anything to disagree on - my take is “coders are the new managers”.

DanLuu recently published something about “manually written automated tests” being phased out and LLMs just writing huge numbers of tests (he mentions 1000 servers as size of test cluster).

I have no idea how to get there but definitely feel I am slowing down the test writing process by being part of it !

My back of the envelope maths:

Suns energy at ISS is about 1.4KW/m2 Solar panels about 35% efficient but let’s say 50% for fun

700w/m2, or about one H100 worth per sq metre (hey, I could run my own H100 off a roof top panel !!)

We want a small 70MW data centre - which is 100,000 times the size so 100,000 m2 or an array 316mx316m or 15 football pitches

Then as it’s energy in and energy out you need radiators on dark side of same size

The ISS is ~ 2000m2, so that’s fifty ISSes

I mean it’s physically possible. But the engineering, the space launch costs they are staggering. And the upside is … Im not sure

All the win seems to be is free sun energy, but a data centre in Texas or Nigeria just needs about twice the solar panels and some big ass batteries.

Im not costing that out but, honestly it seems like a marketing pitch or a really obscure need to put compute beyond the reach of governments.

Almost 5 billion humans have smartphones - as a species wide achievement its utterly incredible. And yet there are two major manufacturers and not even ten with 100M plus handsets (apple, Samsung, xiaomi, oppo, vivo, huawei)

This is a strategic risk right up there with AI ans starlink - and while we don’t want it to stay this way, it’s even harder to imagine how to fix it.

we are descending into a balkanised world of trade wars and threats. Imagine huawei, or apple being told by their respective governments to turn off security services for phones in europe, for example.

It’s not just an AI arms race.

(My tentative solution is governments start to handout devices that provide NFC digital IDs and start growing from there… but that’s a long way from “as good as apple”

So my very first reaction was Inwoukd rather write python. My second reaction was, can this make an awesome cucumber like test scenario language?

I think my issues are twofold

1. You still have a DSL (“give back” > “return”) so it’s not English - it’s a DSL that looks more like English than python

2. Where there is English language like flexibility (Set tempo to {{ a tempo appropriate for a slow blues }}.) we are still passing out to an LLM during the compile process - so it’s not entirely deterministic

There feels like there is something here - in the area of test definitions and subject matter experts and Fitnesse tests. But I just have never seen cucumber like tests work right - in the end it’s another DSL and the SMEs end up describing the problem to devs who write the cucumber test and the SME read it

First the problem and s a lack of software literacy - if everyone could code we would just write unit tests

And you cannot get past that. No matter what we do an SME who is English illiterate is not going to write down their knowledge and same goes for software literacy.

I think there might be more traction in having LLMs help the BA / SME write comprehensive docs upfront … but that has other problems

But love that there is willingness to try and explore - keep going

This is really a banking / credit card thing. CC suppliers put a lot of effort into packaging recurring subscriptions for their businesses because … people forget to cancel. So the real leverage here is not go to each individual website to one click cancel, but just cancel the direct debit / subscription from in your banking or credit card app.

>> Grant me the serenity to accept the bad code i shouldn't fix, the courage to change the code I can, and the wisdom to know the difference.

Fantastic

Isn’t this a question of how much the “terrorised survivors” spend on tokens to make the output “canonised”?

I think the main argument that the billions spent are not going to be recouped is accurate, but I strongly suspect the cost of producing high quality code will remain the same -just being produced faster (speed, cost, quality - you still only get to pick two)

If one Steve Yegge can burn tokens in “Gas Town” that cost as much as me and ten others then you have saved my salary but spent it on Steve’s token use for roughly the same code quality as me steve and ten others would have produced in three months - just it took steve three days

Same price, faster delivery. Is that a win ? I suspect that facebooks recent announcement (“we cannot think of enough things to do with software so who wants our GPUs?” Might suggest that it’s a business model problem more than a software probkem

Ok, so I have to try that.

How are you specifiying the graphs? Is this on github (I am still trying to move from concept to how to actually do it (plus Inhave only just woken up and need coffee :-)

So you are using a “cloud” provider and at 1c per million tokens …

Love to hear more about how you structure the orchestrator etc

> just aren't good for end to end agentic workflows.

This is (apparently) the conceit of SteveYegge / GasTown - no model can cope unassisted so chunk it up, run it and if it falls over remember the exact place and restart, merging it all in

But that’s not my point.

I believe that software is a new form of literacy and just as all Companies and societies are literate now, in the future (tm) companies will run exclusively on software - AI developed software and those who go all out will have the sort of advantages the Catholic Church had over .. guilds?

Anyhow, that’s me being AI optimist. But writing the code is going to be a small part of that transition - almost everything to do with LLMs that is claimed amazing (Computer vision is something else) - almost everything people say we need an LLM is stuff you could have done three years ago but your internal politics just would not let you. Oh look we can see if our policies are being met (you could have written the policies in code and solved the whole problem)

Im struggling to get it out but - almost everything AI is proposed for is stuff a well run engineering firm coukd have taken on. A software literate firm could have done without AI is where firms are hoping AI will Get them

Imagine how far ahead real software literate firms will be - as long as they don’t burn their runway in tokens

Which is why, the right play imo is still buy in-house as much as possible, engineer around the problems and explore the phase space at marginal Cost.

Then and only then think frontier models.

My advice to any CEO / individual - throw your hands in the air and bring it in-house. Yeah the performance can dip depending on what GPUs you can salvage these days but the uncertainty over price is almost nothing compared to the uncertainty over the effective use of AI. It’s not just coding (do I go partly agentic or all out Steve Yegge). This is all over the enterprise - do we parse every email, rewrite PowerPoints? Or just stop using PowerPoints at all. Do we throw LLMs at the mess of wikis and word docs, do we pretend that the policies no-one has ever read actually are how the LLM should think or is it how the work actually gets done - barely documented

The uncertainty of how to use this vastly vastly outweighs the price in a data centre - so buckle up, buy enoughbGPUs to experiment at a known cost and one day you will find the approach that gives you 10x returns - at that point pay any price per token but not till then

I hold an unpopular opinion here, but well regulated digital IDs might not be a benefit to us all.

Almost all big tech giants are frankly paper tigers (can I survive without facebook, sure, without next day delivery, harder, without banking or secure ways to talk to my utility providers, waaay harder.

Having a way to get rid of bots that sour our Online discussions, lovely. To reduce billion in bank fraud, sounds great.

I feel as privacy advocates we need to be clear on the difference between secrecy (that’s gone unless you stay offline.) and privacy (your neighbour knows you are having sex in the afternoons but does not say anything.

So I’ve been wondering about “one or two levels back” chip design. If I understand it, 28nm chips (pre EUV) is just about suitable to run (not train just inference) frontier models.

And so if I was a mid-level State would it be worth while to take my nascent chip industry and push it out to build a 28nm foundry and supporting eco-system.

The models will come but the real challenge of the future is having enough compute power for every one and every use. Even if LLMs don’t become AGI they will still be incredible tools - and as OpenAI seems to spend 8000 for each 200 monthly subscription building one’s own data centres seems sensible

>> For now I have not moved past the point of comprehension being important to me.

I see software as new form of literacy, even in the AI world, so yeah in my world view, comprehension will be something we always cling to.

I might comprehend some code the way I comprehend the newspaper article on the second page, others I comprehend like a Dylan Thomas poem. My attention might be different but I still need to understand it.

I think the recent study was that 200 Claude subscription burnt through on average 8000 worth of datacenter and GPU. There is a reason OpenAPI and all are losing a fortune - they are chucking H100s at anyone who types in a query or subscribed to GitHub. It’s a totally unsustainable business model - price things at cost and we will find value in these services - but I won’t pay a fortune for a search result. Even a better search result.

In this case - let the market decide

Being told you are carrying a scorpion across a river is not a lot of help halfway across.

I think however this mostly speaks about how hard it is to run anything even vaguely securely. There are so many independant CPUs and OSes on one motherboard plus HDD these days …

I think this is going to lead towards companies and governments starting to build their own data centres for this kind of thing (reviewing millions of emails to find the smoking gun is what the cops will want to do as will both sides in a civil case). The thing is no-one (probably not OpenAI) wants to send kidnapping footage to data centre in Texas for a Manchester case.

The thing we are all finding now (software devs) is that AI is great but boy is it pricey if we want anything useful …

And walking through millions of pages of digital evidence is going to cost, and I think governments would rather buy a rack of H100s and spread the cost.

So is the national AI drive one to be able to build frontier models, or one to just build data centre or one to build a chips capability a few years behind ?

In the early years of X-rays, doctors found all sorts of patients with major organ displacement, and performed surgery to, for example, hold the liver or kidneys “up”.

It took a while to realise that textbooks since Leonardos time had drawn and based anatomy on (dead) patients lying on a slab. But X-rays were taken with (alive) patients standing up. So of course there was a lot of “your kidney has slipped!”

I fully support and applaud this kind of medical innovation (even if … why midjourney?) but we need to be careful of the medical term VOMIT (victim of modern imaging technology). At some point we need a human doctor to say “calm down, live your life, eat right, exercise right, and accept that somethings don’t need to be panicked over yet - come back in six months”

>> you could get stuck in a recovery loop and softbrick your device.

Your car …

>> devcontainers with podman compose using the Debian container and nix feature, and devenv.

Can you expand on that please?

So to try and understand your position - you are hired by a small? Company in which sector? And you built an app that does what? And I think most importantly - how did you find the gig? Were they explicitly hiring a AI capable person to do X?