And if you keep starting new apps you can’t keep iterating/updating/maintaining all the old ones too without spreading your engineering talent too thinly.
Google’s internal culture, at least from the outside, doesn’t seem to be one that often releases a bad product and iterates it into a good one - sometimes they do, but with most either it gets lots of interest/users from day 1 of general release or it’s quickly abandoned.
But I would contrast it with Apple (who seem to spend much longer in the pre-release baking stage + have a more mature user feedback cycle post-release [for hardware...]) and Amazon (spray and pray + iterate like crazy + a more mature deprecation and support story) and Microsoft (developers, developers, developers [and especially enterprise developers]).
An engineer gets promotion from "having impact." Improving an existing app is usually not enough impact. Replacing it with some shiny new thing is. Once it's out there, you ideally transfer to something else where you can also have impact, or you quit & join an outside company.
As many people have pointed out, your own manager cannot promote you; only the Promotion Committee can (although your manager can certainly help).
I feel like Microsoft is a worse offender at this. As soon as they get some traction with one product they start overloading the name. Think of all the unrelated products crammed together under the umbrellas: ".*365", "Visual Studio .*", "Surface .*"
They probably just measured how much people with new devices are using the phone version + projected 3-4yrs into the future and saw it declining even further (it'd probably take about ~2yrs for the release to hit mainstream).
Then evaluated whether it was worth slowing down product development and new releases to support two UI versions.
It could be better for customers in the end as the number of used cars without a screen dwindles each year and they can improve the product quicker.
The other question is what Apple is doing...