The problem is that for certain things you can not get away from an app. With MySpace you could visit another website.

For example I make an IoT device (https://www.stationdisplay.com/) that gets configured via Bluetooth LE. I can not access BLE via mobile browser because most don't support it (https://developer.mozilla.org/en-US/docs/Web/API/Web_Bluetoo...).

Of course there are alternative methods like starting a wifi access point and having users connect to that but this makes it a lot more complicated for the end user. That results in un-happy customers and a lot of extra support costs.

I totally agree. My software is essentially a digital asset manager, though with a bit of extra functionality and project management built-in. I don't see how a web app could efficiently index millions of files on the local file system.

However, based on all the horror stories I've heard, I'm seriously thinking about re-considering my approach and settling for a different, simpler web-only product. It's frustrating to think that the majority of development time is not going to be spent on the app & cool technology but rather on bureaucracy, platform distribution,integration into app stores, and setting up cross-platform CI.

This. I built a cross platform app. You pay the yearly dev fee to apple and you can sign your apps for macos in and out of the store. MS do let you do signing for their store, but for outside distribution you need EV certs that are very expensive and a pain to manage. For a small business it's probably fine but for a solo dev it's really shit. And yeah, it flagged one version of my installer as a virus leaving me scrambling and anxious thinking my build could have been compromised and then oh no it's just based on predictive ml and it's actually fine.