Don't worry, AMP is just an interim step on the path to Google's ultimate vision, which is to answer all search queries directly with a fact, passage, or knowledge graph datasheet, using information it extracted from your content and without providing any url at all.
Basically the whole web has no single point of failure, with the exception of DNS, and even that is decentralized. So what do you do with that?-- obviously introduce a single point of failure in the form of a private service.
Is there a more recently updated analysis on why AMP is/isn't bad for the internet? Using my Google-fu to look for criticisms of AMP seems to yield little to no results
In an attempt to do my part in killing AMP, I just disabled it on my Wordpress-hosted site. If you've got a Wordpress blog hosted on their servers, you can go to Settings > Performance and disable AMP completely.
NoScript breaks some sites but makes many run faster. Everything defaults to blocked, but I've marked some known ad servers and Google/Facebook trackers as specifically blocked so I don't accidentally unblock them.
Also, setting up custom searches is very nice: when editing a bookmark, attach a keyword to it and put "%s" in the URL. If you type the keyword first, the rest of the search string will replace the %s. I use it for the usual Wikipedia search, but also I've set "@r tf2" to take me directly to the subreddit called tf2, and @i immediately throws the query at DDG's image search.
My personal basics are Adnauseaum (uBlock origin based, makes exceptions for non-tracking ads, and clicks all tracking ones to pollute data), Decentraleyes, HTTPS Everywhere, and multi-touch zoom if I'm on a 2-in-1
Please add in a (2017) to indicate this article is quite dated at this point