What are the other 310?
Again, a lot of these companies have also been around so if my app has a ton of bloat from years of work, we're bringing in frameworks from other companies who have been around for years as well and probably have a similar amount of crap in their code. Plus they probably want to track everything we're doing so they're bringing in their own third party libraries. It's kind of ridiculous but that's where we're at.
Bigger companies won't have as many third party libraries as they can afford to do it in house but they're still bringing in similar sized libraries to do the same things.
You can see why bug bounties get rewarded well. Though mindful, money is not what drives everyone. Then there are the greedy, in which such exploits value on the black market can be higher. Not forgetting government agencies level.
I wonder which email client will break the 1GB mark, and when we will see a resurgence in reducing bloat. I'm sure that phase will come, did for Microsoft once.
Woof. This is just so wild if one ever stops to think about it. ~300mb of tracking for a single app is bigger than the entire hard drive of a fully internet-capable desktop computer in the mid '90s.
It's not clear if compute can be cheaper than storage still today.
On one hand you can afford to use less storage but you have to use GPU power everytime to draw graphics, if the chip can support the compute requirement you can save on storage, but you pay with higher power draw at every interraction.
On the other hand you can just put more storage, chip assest that are rendered for the device they'll be used on and be ok.
Outside of crisis like now, storage should be cheaper in the long term I think. I doubt there is that much benefit in having assets being able to resize to any arbitrary resolution. The definition used in phones isn't that far away than what is used in laptops, monitors and now even TVs.
Something to think about is that icons/assets often need to change shape slightly as they become smaller or bigger for optical reasons. So even if you manage a fully vector scaled UI, you might still need to have difference depending on DPI to reading distance ratio. Rasterized assests might still be the real answer for a very long time.
Considering how bad is the iOS 26 release on performance, because of its dynamically computed interface, I'm not sure it's worth pursuing vector UIs, it doesn't make a lot of sense to make a more powerfull chip just to draw prettier or more "pure" interfaces...
You can of course override this behavior and redraw your vector every 8.3 ms if you want, but I promise you that this is not faster. For sparse pyramid-tiled vector images like Google/Apple maps, this is a two step process using the latter method followed by the former.
Besides that, there's just a lot of garbage that gets added by various people with different interests. An unoptimized version of the app I'm currently working on has a ~15mb binary, the core app not including all the "extras" people have asked for. It has about 75mb of assets, probably 10-15% are unused but I have no idea. The download size is about 400mb.