"I just loaded the New York Times front page. It was 6.6mb."

   ftp -4o 1.htm https://www.nytimes.com

   du -h 1.htm

   206K

For the author, 206K somehow grew to 6.6M.

Could it have anything to do with the browser he is using?

Does it automatically load resources specified by someone other than the user, without any user input?

Above I specified www.nytimes.com. I did not specify any other sources. I got what I wanted: text/html. It came from the domain I specified. (I can use a client that does not do redirects.)

But what if I used a popular web browser to download the front page?

What would I get then? Maybe I would get more than just text, more than 206K and perhaps more from sources I did not specify.

If the user wants application/json instead of text/html, NYTimes has feeds for each section:

    curl  https://static01.nyt.com/services/json/sectionfronts/$1/index.jsonp

where $1 is the section name, e.g., "world".

The user can use the json to create the html page she wants, client-side. Or she can let a browser javascript engine use it to construct the page that someone else wants, probably constructing it in a way that benefits advertisers.