I don't see the point of comparing HN apps. HN is a website and doesn't do anything meaningful as an app.
Android 5.x has less than 10% marketshare, and 4.x seems quite resilient in low-end devices. It's going to take a while until you can build hybrid apps exclusively for Lollipop.
If you don't think mobile browsers have performance issues, you must use high-end devices exclusively.
Twitter is list of stories with notifications. For me its more like HN app. HN has list of stories, Twitter has 140 characters tweet. When you click on the HN story, you get list of comments or article. When you click on the tweet, you get replies and retweets and so on.
I didn't start with Android 5.x. I started my app with Android 4.x. I don't think you use low end Android device. Even native apps on Android has lot of issues related to performance.
On my Nexus 5.0/Android 5.x, Facebook scroll is not smooth when there are pictures. I only get non-important notifications from my twitter app.
I test my app in first generation of iPad 2 (My iphone 4s broke) and galaxy s3.
---
Now show me 5 meaningful 100% native apps (Non-game) that are not developed my a startup with a huge capital or big company, available on iOS and Android, and has 4.x star ratings.
Well, lets see, I use Droidlight but that's by Motorola LLC, so maybe ColorLights. ConnectBot (ssh client). Barcode Scanner (I'm not actually sure if this qualifies as 100% native -- but I think it does). Open Camera. K9 (email - there are others, with different strength/use cases). ChatSecure.
Now, these are all Android apps, and few (if any) have iOS versions. But they all have iOS equivalents -- and I don't think most of them would've been as good, as non-native apps.
Note the absence of stuff that works well as a website, like facebook or twitter. For games, I think the choice between native/html should probably be governed by the answer to:"Is it as much fun as html/webview?". This is a good yardstick for other "apps" as well.
For eg hn, there's no need for an app, but of course hn is a pretty terrible web page. There are a number of easy fixes that would make it more usable, especially on small screens, but the maintainers don't care. And that's fine -- it's not accidentally bad, it's intentionally bad.
One could make a similar site that didn't break voting, threading, screenlayout on small screens quite easily. And it should probably be a web site.
One can make a "web app" to schedule meetings (see doodle.com) -- and one could augment that with something native[1]. Or one could make native apps like the ones mentioned above.
Could one make a web-rtc proxy for ssh and allow logging in to edit firewall rules via web browser? Absolutely. I'm not sure I'd want that though. Not as long as web browser security refuses to learn from office macros: unsigned js code, all or nothing execution etc.
There is some middleground, as google docs demonstrates. Personally I prefer content creation/editing to be local, possible to do off-line (with sync) -- and I'd like competing apps to be able to easily share data (via eg: the file system).
Now, that twitter can't be bothered to make a decent web-site isn't really an argument against web-sites. It's an argument against poor web-sites.
> Android 5.x has less than 10% marketshare, and 4.x seems quite resilient in low-end devices.
Folks were saying the same exact thing about 2.x when 4.x came out. By the time 6.x comes out, everyone will be yammering on about how 5.x marketshare is still strong and "seems quite resilient in low-end devices", blissfully ignoring the point in time when 4.x - like 2.x - fades out of view as it's increasingly ignored by app developers.
Meanwhile, in the actual low-end space, you have things like FirefoxOS where there's zero difference between "web" and "native" because "native" apps are just web apps with special hooks for things like cameras and sensors, or you have things like "feature" phones where the closest thing to a standardized platform you have is Java ME.
In the real world, Firefox OS has zero marketshare at the low-end. Mozilla announced just a day or two ago that they're abandoning the low-end strategy.
Featurephones are largely gone. Microsoft bought Nokia's featurephone business, shut down further development and has been converting it to low-end Lumias as fast as they can, but Android is nibbling away most of that market.
Mozilla only announced that they're abandoning efforts to build the lowest-end stuff. IIRC, most FirefoxOS handsets are still pretty low-end, and have been doing reasonably well.
And feature phones are still dominant in places like Africa and South America, particularly due to their lower cost and power consumption (smartphones are still unable to reach the battery lives of even dumbphones from a decade ago, which is an important consideration in environments with limited/inconsistent electricity). Even in "developed" countries, feature phones are popular with the elderly and disabled, since they usually feature physical buttons that are easier to work with (better tactile feedback, easier to find with poor vision, etc.) and have simpler interfaces.
Android 5.x has less than 10% marketshare, and 4.x seems quite resilient in low-end devices. It's going to take a while until you can build hybrid apps exclusively for Lollipop.
If you don't think mobile browsers have performance issues, you must use high-end devices exclusively.