Another very late response: you’re confusing Googlebot (the crawler) with Caffeine (Google’s indexing system). Yes, Google can and does index JavaScript content. The indexer renders the page and extracts the content. But the crawler doesn’t do that.

What happens when you use React without server-side rendering is that the crawler halts on the very first page because it can’t see any hyperlinks to follow. It sends the page to the indexer, which then has to render the page and extracts the hyperlinks, which will then be added to the crawler’s queue. Then the crawler will eventually crawl the next set of pages, and again will stop there because all the links are invisible until the JavaScript is rendered. So it has to wait for the indexer to come back with a new set of URLs to crawl. Etc.

It makes the crawl process incredibly slow and inefficient. And that is why websites built on React (and similar JavaScript platforms) perform worse in Google than websites that primarily serve plain HTML to the crawler. The latter type of website will be crawled much more efficiently; plain HTML websites can be crawled very efficiently, newly added and changed content will be crawled and indexed much quicker, and Google is much better able to evaluate the crawl priority of individual pages on such websites.

Whereas JavaScript-based websites, while they can be indexed, are incredibly tough to crawl. Kudos to Google for trying to do so anyway, despite the enormous overhead this causes in their data centres.