I think the tweet author was expressing frustration with folks treating discussions around MVP/MVI/MVC like fashion, dismissing or promoting one pattern based on what is a sense for the community pick. I think a few responses in this channel might jive with that.
The author is at Slack and they probably have a large team. My guess is most of their code is structured so you have FeatureView, FeaturePresenter, and WidgetRepository, and that some folks are pushing to use some MVI variant. I can imagine his stance being that if you designed your app such that presenters survive configuration changes, and your goal is to move towards something MVI-ish you can implement a unidirectional flow of data within your presenter by just changing the contract to something akin to what we see in ViewModel's, the only difference would then be the class name. And you can similarly outsource work to repositories and datafetchers so your number of lines of code (LOC) don't become unwieldy.
Agreed there is a lot of wiring/scaffolding necessary for setting up a redux-like pattern. I started off by copying the pattern in
https://github.com/kaushikgopal/movies-usf-android and nearly abandoned that approach a few times out of frustration for how much wiring it required. I am glad I gave it a try. I used to work on Keepsafe, an app for encrypted document store. For a few of the trickier bugs around syncing I went through some time-consuming cycles of instrumenting breadcrumbs, releasing the app, realizing I wanted to track more state and adding in more breadcrumbs. Now I am more open to approaches that require upfront wiring but carry some benefits for more quickly understanding how a specific state is tied to a bug and possibly making it easier to write a test for generating that state.
Reading through some questions here and elsewhere about how to handle scrolling or implementing a text editor with MVVM/MVI has me questioning if maintaining duplicate state is always a good idea. I haven't dug into jetpack compose yet either.
I abandoned instant run after a few bad experiences. Seems like a lot of folks did. Maybe it can make a turnaround.