I feel like google maps is a great example. Yes, you could technically make that a server-rendered page with a touch JS. But most of the benefit of google map

I feel like google maps is a great example. Yes, you could technically make that a server-rendered page with a touch JS. But most of the benefit of google maps derives from its interactivity. If I were creating a new startup to build google maps, I'd have to consider: the best version of my product is the one with lots of heavyweight JS. 99.9% of my users will be able to use it. 80% will get a decent experience. It'd take me 60% longer to build a version that degrades gracefully. My competitors are breathing down my neck- is it worth spending that time? Most apps aren't as clear-cut as a video game or a blog. One company I worked at was a 2-sided e-commerce site (think "Craigslist for X"). It started server-side. We started adding more and more functionality to the main view, a grid of items. Quick-look popups and complex interdependent filters, and as-you-type search are what stick out in my mind, although there were others. We ended up rebuilding that page, and the decision was made to skip progressive enhancement so we could focus our limited resource on improving the javascript-heavy experience. We already had a mobile app and a mobile-specific site, so the only people that would be bothered by this would be the people with JS disabled on their desktops (a vanishingly small minority) or those with slow desktop connections (we only supported US sellers/buyers at the time, so that skewed us more towards fast internet). Another company was building an online version of a well-studied mental health technique (cognitive behavioral therapy). You can get this from a therapist, and it works quite well (measurably so), but it's quite expensive. You can get it from a $20 book, and it works well (measurably so), but completion rates are like 20% of when you do it with a therapist. Turns out that self-improvement is hard and people don't like doing hard things. So this company was building an online program that measurably worked. However, one of the key value propositions is that people should complete it at a higher rate than a book. Ideally at the same rate as a therapist (but cheaper and more convenient). To do that, the company needed every possible UX technique they could get their hands on. They needed every bit of interactivity at their disposal. In that case, it seems like a pretty good argument for focusing effort on a js-only app, even though the underlying content is largely a text-book.