fsloth

Yeah, this is the difference.

2000's tech bubble was caused among other things over-investment to infrastructure and technology that had no users yet.

Totally different setup.

Does not mean AI boom will not turn to bust, but weak analogues generally don't help with understanding complex systems.

I'm paying $20 for Codex and $90 for the Claude Max plan. They are a "pry from my cold dead fingers" product for me.

IMO if someone tried this tech last time 6 months ago, or their only exposure is eg. via MS copilot, they do have a rational reason for skepticism. No technology of this complexity has improved this rapidly in my memory (well, ok, we had the CPU speed races from 90's to early 2000's).

"I've always been surprised that Microsoft didn't do a full operating system refactor and provide a compatibility layer for running old binaries"

Just keeping a legacy system in working order is different skillset than writing a new system from scratch.

So you need a new team. Nothing from Windows maintenance transfers.

Maybe would require hiring someone who knows how to design an OS.

It would be a major undertaking, needing protection by CEO (and if it would not succeed CEO would loose a lot of prestige).

I'm not saying MS does not have the existing talent base. I don't _know_.

But I've been inside a house maintaining a monstrous legacy codebase.

I can tell you - it requires surprisingly little deep understanding just to keep an existing system going.

Sure but different target market.

CRUD apps are non-trivial.

If Unity were to ship platform native replacement for WPF equivalent (hell or even winforms) it would become a really enticing app development platform.

I didn't say backcompat isn't important.

There are lots of other ways to run old binaries than at your main OS level.

There are tons of other platforms that precede the current ones.

I would not like the requirements from those platforms to hamper the current gen os.

I do think it's valuable to be able to run the programs from those platforms.

I think brands that build trust will be the key to sustain income.

People want to outsource some things. Thats what markets are for.

The internet will become even more scammy and noisy.

Hence trusted vendors will be like beacons, and not drown.

Agree - there are variations to how much tweaking Windows needs.

Enterprise Windows config that comes eg in Thinkpads is more ready out of the box than the consumer OEM configss.

IMHO - disagree but it depends on point of view so this is not ”you are wrong” but ”in my view it’s not like that”.

I think it’s the role of the software vendor to offer a package for a modern platform.

Not the role of OS vendor to support infinite legacy tail.

I don’t personally ever need generational program binary compatibility. What I generally want is data compatibility.

I don’t want to operate on my data with decades old packages.

My point of view is either you innovate or offer backward compatibility. I much prefer forward thinking innovation with clear data migration path rather than having binary compatibility.

If I want 100% reproducible computing I think viable options are open source or super stable vendors - and in the latter case one can license the latest build. Or using Windows which mostly _does_ support backward binaries and I agree it is not a useless feature.

Exactly.

I want to have a computer with stable vendor supported OS so _I can do my stuff_ not tweak some os level configs.

I _don’t_ want to spend my time playing an os systems programmer.

OS is a _component_. Like the wifi driver. I think it’s great some people love developing wifi drivers but personally I just want network that-just-works because there are billion other cool things you can do with a computer.

Similarly I want an OS that just works! Without asking me to do a anything! Because _i don’t really care_. (I mean i care it works but i expect the engineers actually developing an os offering to have a far better idea than myself what is a good stable default config for the system)

Ok.

Look.

I guess we all care about software business here.

And computer? It’s what consumers buy from store. Preferably in cybermonday or similar sale.

To run the software they ran on their previous computer.

They hope slightly faster. But honestly? They couldnt tell. Anyway the new computer is shinier.

OS? What’s that? (They honestly could not care less)

They dont buy apple for the os. They buy it for the brand.

Don’t undersell it, it’s perfect actually.

Open source is a supply chain specific issue and consumers don’t care about supply chain.

Anyone with any illusions about this name quickly the top vendor for the third item in the materials itinerary of the first thing with a materials itinerary you get your hands on (for me it’s usually food. Who is the main vendor for citric acid? Or sugar. Or that red dye that causes adhd. I have no clue)

General consumers could not care less about open source.

It’s component.

Not a product.

FWIW Sony's wh-1000xm3 and later are suberb.

Windows and bluetooth is a really difficult combination. The problem is that for some reason if Windows detects a microphone on the Bluetooth headphone, it switches to a transport mode that a) allows the mic through b) makes the sound sound horrible. So disabling mics sometimes helps there.

Well, if you buy only one pair it does work really nicely with all Apple kit. So you get really nice cinematic sound from Apple TV (for my non-prosumer ears), and effortlessly can switch between phone, laptop etc. The sound is really good for video calls.

They just work.

I mean there are other pieces of kit that probably just work as well but with these you don't need to do market research.

It's surprising how non-trivial even _adequate_ sound is still in 2026 and that's what these are guaranteed to give in any situation IMHO.

If you have only one Apple device probably no selling point as such.

I’m also in the camp ”I respect the authors but here am not impressed”

Regarding wider adoption ”One Battle Another” is incomprehensible mess unless you are familiar with US. I understand the white supremacist conspiracy but nothing else makes sense _at all_. I trust when people say it’s super intelligent but I just don’t have any of the context.

Sinners is a cool music video that is about 5x too long - the few dances are really nice, some of the characters have very strong presence and few dialogues really are strong but … I would not categorize it as industry leading in anything unless the industry reference points are super low

They don't have understanding but if you follow the research literature they obviously have a tendency to produce a token stream, the result of which humans could fairly call "entity with nefarious agency".

Why? Nobody knows.

My bet is that they are just larping all the hostile AI:s in popular culture because that's part of the context they were trained in.

Linear is pretty nice IMO otoh have not experienced it at megacorp scale.

I would have expected MS has charged some fee for Windows but honestly no clue.

I'm sure there are readers with actual insight here :)

As OS matters less that’s probably not going to entrench Microsoft less. Their relationship with OEM:s is not market based but a two way relationship.

At this point I would not be surprised if MS started to subvent the PC manufacturers to favour Windows over Linux if that ever comes to that.

No, that’s not how LLM:s work. Single prompt does not make it any better. Please focus on interesting humans comments.

If you feel like it sure chat with claude to build your insight. Then write what you think _yourself_.

If you want to introduce references use urls to non-ai generated contexts.

I means as a HN protocol.

HN is supposed to be interesting.

LLM output specifically is not interesting because everyone else can generate roughly the same output.

"It's not that different from pasting a quote from Wikipedia"

Claude's output it _totally different_ from pasting a quote from Wikipedia.

The latter has the potential to be edited and reviewed by global subject experts.

Claude's output totally depends on what priors you gave it and while you can have high confidence in the context no third party should have.

This.

Ofc a huge chunk of that is in companies but I'm fairly sure there are at least two windows 11 machines per one mac in consumer segment as well.

I use a git gui :)

So the only time I need terminal, it’s for something non-obvious.

”There are tons of resources”

This is not a standard curriculum as such though.

I’ve tried to come to terms with posix for 25 years and am so happy I don’t need to anymore. That’s just me!

You do realize it does not help _me_ at all if _you_ have found your perfect custom setup.

Because it’s custom there is no standard curriculum you could point me to etc.

So it’s great you’ve found a setup that works for you but I hope you realize it’s silly to become idignant I don’t share it.

TBH I’ve been here a while, never felt what the point with the above is but do feel LLM:s are a new valuable affordance in computer use.

I mean I don’t have to remember the horrible git command line anymore which already improves my exprience as a dev 50%.

It’s not all hype bs this time.

Operation Entebbe comes to mind of an insane sounding stunt that worked unreasonably well.

No need to be sorry - you raise an excellent point!

Note my critique was labeling all of us LLM enthusiasts by association ”incompetents” which I believe is an incorrect assumption.

The point raised that more people can now code I think was a correct one though. I think that’s a net benefit.

Let me be brief. There are two topics here - CAD & AI and AI & society which I think the underlying point we are discussing.

I appreciate you made a domain specific example, but like _all_ AI workflows - it does not really hold up unless one is extremely specific what the workflow is.

First of all if someone is making a CAD tool for drawings that’s really not a segment. All 3D design tools target a specific content workflow, with specific domain model. Drawings are one possible output from this domain model - just like the on-screen 3D presentation or a 3MF file you get for export.

What ever LLM competency level is it does not come with it’s own domain model. Real people want to configure the models they create. This means there needs to be a domain model you hook up to the LLM to have stable model with specific editable components.

So if you are prompting a model, you are still better off if you prompt the domain model in a real cad package.

So I don’t think CAD packages will die.

Second - I’m mainly trying to serve _my_ need (which I believe is shared by others). My need is that I want to design 3D models with minimum effort, in an enviroment that has perfect undo, perfect boolean, versioning, snaphshotting and intuitive parametricity. This package did not exist in the market before.

Will it have traction? I would expect there are lot of human users that want to create models themselves. Computer chess did not kill chess etc.

To be super specific, there is a clear wedge in the market between Tinkercad and Fusion360 for an affordable desktop offering with the above features.

I do realize my market thesis is just a hypothesis at this point. Which is fine - it’s a passion project. I hope it will be usefull for others, but if not, at least I will have the tool I want.

I’m mainly excited about the possibility of being able to ship to test my market hypothesis.

Without LLM tools I would not be able to ship.

Regarding society:

I believe we are discussing a normal destructive phaze of innovation cycle. Machine looms, weavers, luddites, new forms of labour etc.

Regarding living standards the main worry is - can ”normal” people exist above poverty?

I guess the markets will want to have consumers in the future so either there will be new jobs or some form of basic income.

It’s possible I’m wrong as well.

I have no idea if democracies will survive.

I am sorry you feel that way but I feel professionally strongly insulted by your statement.

Specifically the implication high LLM affinity implies low professional competence.

"My experience is that people who weren't very good at writing software are the ones now "most excited" to "create" with a LLM."

Strong disagree.

I've earned my wings. 5 years realtime rendering in world class teams. 13 years in AEC CAD developing software to build the world around us. In the past two years I designed and architected a complex modeling component, plus led the initial productization and rendering efforts, to my employers map offering.

Now I've managed to build in my freetime the easy-to-use consumer/hobbyist CAD application I always wanted - in two years[0].

The hard parts, that are novel and value adding are specific, complex and hand written. But the amount on ungodly boilerplate needed to implement the vision would have taken either a) team and funding or b) 10 years.

It's still raw and alpha and it's coming together. Would have been totally impossible without Claude, Codex and Cursor.

I do agree I'm not an expert in several of the non-core technologies used - webview2 for .net for example, or xaml. But I don't have to be. They are commodity components, architected to their specific slot, replaceable and rewritable as needed.

As an example of component I _had_ professional competence 15 years ago - OpenGL - I don't need to re-learn. I can just spec quickly the renderpasses, stencil states, shader techniques etc etc and have the LLM generate most of that code in place. If you select old, decades old technlogies and techniques and know what you want the output is very usable most of the time (20 year old realtime rendering is practically already timeless and good enough for many, many things).

[0] https://www.adashape.com/

"Interesting, this is a trap that I've seen multiple senior hires fall into."

Definetly it's a trap. If you are a purist it's nigh impossible. But if you ruthlessly 80/20 it most stakeholders will be pleasantly surprised.

I have no clue why I end up in these situations but I sure do like them.

I do realize this would sound more of a perpetual "not invented here syndrome" but technical implementation of modeling aspects for 3D and computational geometry is such a scarce talent you actually get to do novel stuff for your business.

The last time this happened I designed & implemented the core modeling architecture and led the implementation effort for our new map feature[0]

[0] See section "Stunning new building facades add practical value" in https://www.mapbox.com/blog/detailed-architecture-and-new-de...

FWIW this is the best kind of job ever for me as well:

"when you started you didn't even know if there was a solution."

Regardless what the problem is - as long as I know _nobody knows if there is a solution_ it's an instant sugar rush.

You are free to bang your head against a stone wall for months trying to crack the damn thing.

OFC you need to deliver in the end. And this requires then ruthless "worse is better" mentality - ship something - anything. Preferably a the smallest package that can act as mvp that you know you can extend _if this is the thing_ what people want.

Because that's the other side of the coin - if the solution is not known - people are also not aware if the solution has true value or if it is a guess.

So in any case you have to rush to the mvp.

Such joy!

Of course the mvp must land, and must be extensible.

But these type of MVP:s are not a slam dunk.

The combined requirement of a) must ship within limited time b) nobody knows what _does_ require a certain mindset.