Hacker News new | past | comments | ask | show | jobs | submit login
Google Fuchsia: A system built for 'Instant Apps' on steroids (9to5google.com)
46 points by datguacdoh on March 18, 2018 | hide | past | favorite | 29 comments



AI-driven search, always-on, cloud-focused, centralized, over-engineered for maximum tracking/profiling, decisions left to a single authority... I didn't think Google could outdo itself but along comes Fuchsia.

This is my knee-jerk reaction and I admit I'm looking ahead as Fuchsia evolves further. Trying to see the benefits of this in the long-term but currently failing. Is this another nail in the coffin to conventional desktop OSes as we possibly evolve (or de-evolve) in the decades to come?

I sometimes imagine what systems like those in 'Star Trek', '2001 - A Space Odyssey' (HAL) and others might look/feel like. Is Fuschia an attempt to get us there (cynical flipside: albeit with authority figures bent on reestablishing control of a narrative they have lost over the last few years).

How much control do we give up and... can we trust the gatekeepers?


I watched an episode of Star Trek TNG this morning where they disconnect the microphones in a room so the computer can't hear them.

(the one where Barclay gets zapped with the brain light)

On Star Trek though the computer runs on free energy and is just a tool. Google has to make money.


> On Star Trek though the computer runs on free energy and is just a tool. Google has to make money.

Right. As I think ahead, though, the only tools we're essentially getting are from commercial companies. Even our own governments (worldwide) are running these tools from the same commercial sources.

I'm not aware of any solutions being built now for the kind of interconnected future (smart homes, smart cars, smart "news", etc) that isn't centralized and also coming from commercial sources (Google, Amazon, etal).

The Star Trek/2001 analogy is about an OS system that isn't too dissimilar to our own connected future. Instead of handling a Star Ship, these systems would be running smart homes, managing our lives. That's a lot of profiling in the hands of a few companies (and their "partners") who may not have the best intentions.


This all assumes constant, good and cheap data network coverage tho - is that an assumption the whole OS is making? :-( if so.


Yeah, open some previously-unused (or even not-recently-used) feature of an app and suddenly bam, there goes 100mb of data.


Ad networks do this shit all the time.


Author here. Manual installation will certainly always be possible. The instant/cloud parts of this feature are designed so that things can "just work" under optimal network conditions. Suboptimal and offline conditions are always to be expected and planned for. This is shown by Ledger's offline-first design.


This is basically how the web works - it pulls content in as needed and caches it.

All you need to get the offline experience with an App in this situation is having an Install button that prevents cache eviction.

It also mentions P2P retrieval, which would further avoid using an expensive data connection.


Yes, but with the web as a User I have a certain degree of control. I expect data to be used when I browse the web. And I totally expect data not to be used when I'm not browsing the web.

It sounds like with Fuchsia any predicability goes out the window. I could do a task one day that is fine, the next day it causes issue. My question is, does the user get any predicability and control? Eg: you mention an "install" button - will there be one? The article doesn't mention that. In fact, it's the reverse: "The most important thing though, is that these processes will be completely transparent."

(Admittedly knowledge of this kind of thing does require some technical skill on the part of the user. Maybe Google have decided to remove that to make a more user friendly experience. That might be a good choice to make, but seems worth commenting on at least.)


Instant apps are an interesting concept. I believe my Android 8 phone supports them but I really didn't notice any of them running. Maybe it's how they are designed. Anyway, does anybody know the answers to these questions?

1) Are they going to ask the same permissions every time an instant app is downloaded or are they caching the answers of the user?

2) How do they pick the instant app to download when multiple apps can do the same thing?

3) As a consequence, how to know which company to contact for assistance and/or ask for our personal data (think GDPR)?


Great questions. Specifically for #2 it's hard to say at this point. At present (in it's very early state), I believe they just pick the first result. But the resolution process returns a "confidence" value. This is purely my speculation and I need evidence before I publish it in an article, but I believe that confidence here indicates the intent to use ML.

It can probably be assumed that you would be able to set a preference for one module over another, but I haven't found any proof of that yet.


Try Vimeo. It is the only instant app I have seen yet. It seems to lay around and be more instant if you have already launched it.


I went to vimeo.com with the browser of my phone (Firefox). Started to watch a video, the video plays inside the browser. I checked if I had instant apps enabled in Settings (Apps, Advanced, Default apps, Opening links) and I did. Maybe it's because I'm using Firefox? Would it work with Chrome? Oh well, it really doesn't matter. If I want an app I'm installing it, otherwise it's perfectly fine not to have it. Actually, I prefer not to have too many things going on without me knowing.


I believe you need to use Chrome


Also NYT Crossword puzzle app shows up as an instant app when you Google search for it I think.


I prefer PWAs more than this. Instant apps is highly proprietary. PWAs have the potential to work across platforms.


With a lower UI/UX quality than any native UI, but hey it is portable.

For example, WebGL based games still don't run properly on devices that barely sweat with OpenGL 3.x native apps.


What happened to Escher? The repo is deleted and I can't seem to find anything on it.


I'm not sure why Escher started in a separate repo (possibly to make testing on Linux easier), but it's been moved to its rightful place inside of the Garnet repo.


i hope they don't overengineer this OS by trying to reinvent the wheel. Just a regular mobile OS, working fast on cheap hardware, with an elegant SDK and the possibility to develop apps using cross-platform tools would look appealing enough to me. IE : take iOS or android, fix the parts that are broken, and remove useless feature ( because you don't care about backward compatibility anymore), use a modern and safe language for the sdk, make it open source enough that people can have experiment with it, and you've got a winner.


This reminds me of Apple's OpenDoc[1]

[1] https://en.wikipedia.org/wiki/OpenDoc


Progressive installation is great

I'm getting a bit apprehensive about the feed centric view though. What are the odds Google is going to require full browsing/identity/use/location/voice/video/file permissions for everything, by mixing it all together into one 'experience', like they do with the Google assistant

I suppose at least it's open source


Based on what happened with Android and Open Source, I'd be wary. Over time some of the Open Source core apps and bits of the O.S. (I think?) were replaced with closed source things via Google Play. There was good technical motivation for doing so - it bypassed the slow carriers and allowed users to get the latest updates quicker - but the end result was the same.

I'm also wary of the feed thing to. As well as the data access thing you mention, how does it decide what the "best" app for a task is and do I get any say? Also, I wonder if it will end up with a situation that if you want your content to be available to these users you have to make it available with no control over how it's presented? (You could argue that on the web currently most content producers have messed that up with trackers, ads, etc which is a fair point but I don't know if this is the answer.) If so, whether it's Open Source or not makes little difference to most people trying to publish stuff.


Perhaps I'm missing something, but it seems hugely wasteful to require apps to be downloaded for every use.


they would use caching heavily, they just don't delete it until you actually need some free space.


Perhaps it's part of a transition to web apps?


Ephemerality sounds like JIT but for apps. Seems the necessary stuff to run the app will be downloaded dynamically and cached locally to run later. Though, why cant be this done in the current setup? What are the limitations that fuschia overcomes in this regards?


So... like web pages? https://xkcd.com/1367/

(except worse because it's a walled garden)


"Apps" do not imply walled garden.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: