"Becoming true "User Agents," in other words"
I remember the early browsers as smaller programs that were oriented around retrieving content for further viewing/interaction using other applications. As I understood it, this was one of the ways that MIME was useful.
The problems with "modern" web browsing that annoy users probably have less to do with what, back then, would have been separate programs. They have more to do with the retrieval step. User have little control over 1. the content the browser requests and retrieves and 2. whether and how that content is processed (i.e. the "default settings" of the browser decide and to my knowledge few users today manually adjust MIME settings or whatever today's equivalent is). Because users today expect the browser to do everything. It is more than a user agent i.e. http client, it is an "all-in-one" program that can seemingly do anything. No auxiliary programs are needed. This comes at a cost. I would argue that cost includes many hassles that users experience, including "speed".
The point is that this blog post is talking about #2, the processing of retrieved content. But is that really what causes most of the user annoyances? Is that what ultimately affects "speed"?
Or could it be, at least in part, that some web page is controlling the web browser to have it make dozens upon dozens of DNS and HTTP requests, not only for tracking, but in some cases for "content" the user does not want.
The first task of the user agent, #1, to retrieve content, is not nearly as complex as the second, #2. It probably has not changed much since the early days of the www. (I still use original netcat from 1996 as well as other 1990's tcp clients to great effect.) Retrieval of content can be user controlled. The complexity is manageable. The "user agent" can be small, open source and easily compiled by the user. And with todays memory and networking retrieval can be very fast.
When I think of a "user agent" I think of a tcp or http client, a small program that can make TCP connections and HTTP requests, not one that is also a graphics rendering engine, a video player, a programming language interpreter or a PDF viewer, etc.
It is the disappearance of the MIME idea in favor of a highly complex, all-in-one application called a "modern web browser" that is at the source of every known problem I can think of that is associated with "modern web browsers" today, whether it is browser "speed" or privacy or security or whatever. It is the price one must pay for not having separate programs with separate privileges to accomplish separate tasks.