[tempnow987]

Google was doing Fuchsia. SUPER cool... but what is the business case there?

They are also ham strung by just not listening to customers it feels like - so I think that impacts traction. You get an 80% solution with simple glaring edges. Thousands ask - please fix this edge, but whoever is driving the product is onto something else it feels like sometime.

Apple's main product lines all seem to make money, have positive margin. Their R&D is speculative as well, but at least too me seem to have a clearer path to product. I would love an apple interface in my car vs whatever the OEM ships in most cases. Even Tesla, with most advanced interface just can't keep it current / smooth / familiar etc the way apple does (music / playlist / etc integrations alone).

[karlkatzke]

A bunch of Google’s lack of focus is their internal management culture and nothing will change (and google/alphabet will not be worth being a customer of or being investing in) until the culture problem is fixed.

Google only promotes engineers when they are part of a program launch. No credit is ever given for feature launch or program maintenance. If you are not a part of a program launch within a few years at google/alphabet you are managed out.

This doesn’t create a company that I’d find it worth investing my limited engineering hours in long term. A GCE sales person called me last week and I couldn’t help but laughing when I told them that I’d just agreed to five year terms on an AWS private pricing agreement the previous week. Google’s recent actions made me unsure GCE would still be a product in five years.

[ElFitz]

> Google’s recent actions made me unsure GCE would still be a product in five years.

Out of curiosity, which actions are you referring to?

[mattclarkdotnet]

These ones? https://killedbygoogle.com/

[ElFitz]

Ah! Yes, ok. I wondered if there was anything out of the ordinary that I may have missed.

Thanks!

[olliej]

Fuchsia is (was?) a long term investment, so would be expected to not be directly profitable for a long time.

The core problem I guess is that it suffers from an inability to partially migrate to it. This is a common problem that many people fail to understand actually matters. For example, we know rust is a safer language than C and C++, and so people say "all should be rewritten in rust", but rust is designed with no intent or plan to support partial migration. What that means is that even Mozilla - who created rust specifically for gecko - has been unable to replace the bulk of their code with rust: they can only replace entire large subsections at a time, which is harder, riskier, and as a result slower. Fuchsia's (I guess Zircon's) adoption is hindered because the objectively superior security model also meant that none of the infrastructure around the linux kernel say could work with it, so feature parity required at best significant porting, but frequently complete rewrites of large swathes of the surround infra. e.g. porting android to zircon (the fuchsia kernel) would require significant rewriting of the existing code bases, and the time spent doing that is not time spent adding shiny new features and animations (or chat programs?). Apple is generally much more pragmatic which is why Swift had such good objc and C interop from day 1, and why real C++ interop is something it's working on: if fixing C++ projects can only be done by all-at-once complete rewrites you basically ensure the C++ will live forever - rewriting a file at a time however is much more achievable, which is personally what I want (as for Swift vs rust I'm on the fence I like different parts of both, and also c++ :D).

[surajrmal]

This is very false. Fuchsia is built around the idea of making it easy to port software. It just doesn't attempt to achieve it by providing a POSIX interface. POSIX doesn't specify the vast majority of interfaces a modern OS needs to define anyways. For instance you won't see POSIX interfaces that tell you that provide you a signal for when memory on the system is low and you should free up memory.

Instead Fuchsia tries to target porting runtimes which applications are built against. For instance, porting chrome as a runtime unlocks web applications. Porting flutter unlocks flutter applications. Porting Android as a runtime is challenging but still achievable. It is not necessary to rewrite all of Android to accomplish this either.

At the extreme, it's possible to simply implement runtime support by implementing various virtio interfaces and running a full OS in a VM, similar to how other OS like ChromeOS and Windows achieve Android app support today. Fuchsia is written in a way that you don't need to do that, but it's always an option as well.

[microtherion]

I think it might be worth pointing out that Apple is less inclined than other companies to talk about pre-product R&D, so it may not be as evident when certain lines of investigation don't make money.

[azinman2]

My understanding with Fuchsia was to do better than Linux/Android in terms of security, scalability, and adaptability across devices. Why wouldn't they want to own a great OS full-stack? It's part of their bread & butter.

[warkdarrior]

> My understanding with Fuchsia was to do better than Linux/Android in terms of security, scalability, and adaptability across devices.

That's the theory and the marketing copy. Since Fuchsia has only been tried on a few closed-world embedded devices (Nest hub, Nest camera), it is really too soon to declare it more secure, more scalable, or more adaptable. [By "closed-world" I mean the devices run only a limited set of apps, with functionality known ahead of time and from known developers.]

[dmitriid]

> Why wouldn't they want to own a great OS full-stack? It's part of their bread & butter.

It's not their bread and butter. Their bread and butter is web advertisement. 80-90% of all money they make comes from web advertisement.

Someone in the company said "we need vertical integration like Apple has and we don't like Android", and for a while they managed to run with it. Now they lost the internal power struggle. Oh well. It never had any tangible value to Google as a company anyway.

[azinman2]

Android is a hugely important piece to Google. It’s pretty clearly the case for a wide variety of reasons, some of which is because they want to control the browser/app experience on mobile as much as Desktop in order to fuel those ads.

Android itself of course makes billions alone, which is enough to justify Fuschia. They need a better setup to compete with Apple specifically on decoupling the OS from the drivers for updates on 3rd party manufacturers, in addition to security. I think there’s even some thought it could run in the data center - that drives everything Google. They already employ entire teams that just work on the Linux kernel, compilers, etc. All of that is bread and butter.

[dmitriid]

> Android itself of course makes billions alone, which is enough to justify Fuschia.

Ads are 80% of Google's revenue. Literally nothing else is Google's bread an butter. There's even speculation that Google makes more money from the search deal with Apple than from the entire Android ecosystem.

Android is a vehicle for Google's ads, and suffers from the company's rollercoaster of "interested/not interested" over the years.

Fuchsia by this time is just a 6-year-old money sink with no revenue story.

[KerrAvon]

Sure, but they’re not treating it that way. They simply don’t seem to care about it much.

[azinman2]

Thing is that it’s up to culture to change that. In Japan it wouldn’t go this way, for example.

[surajrmal]

Your phrasing makes it sound like Fuchsia was canceled. It is still being developed.

[worthless-trash]

I would just love to be able to run my own OS on the car.. I'd write the shit out of a better interface.

[astrange]

I think if you asked me to write a car OS the ethical thing to do would be to quit.

[worthless-trash]

The entertainment systems are really quite dull, go checkout toyotas default interface in 2021.

[astrange]

Oh, I mean I'd probably end up killing someone. It sounds hard and I don't know how you become an expert in it.

[worthless-trash]

Just to be clear "car os" != engine control system.