It's right in the section called "How to Play." The DM describes the environment. The players describe what they want to do. (The DM might call for a check at this point or might not.) The DM narrates the results of the adventurers' actions. Repeat, for every single kind of scene in the game. It doesn't say to call for a check before the players describe what they want to do because maybe the DM might describe the environment differently then.
Nobody has to play that way, but I do. Perhaps not surprisingly, the game runs very smoothly as a result.
Well yes, that is essentially a "how to play"... basically any role-playing game, really. Except maybe the really experimental ones, I guess. Of course, while it does not explicitly say to call for a roll, it also doesn't explicitly say
not to, which seems to be the contention. I'd say it's ideal, especially with ideal players, but there have been times, many times even, where I've felt it necessary to at least prod a player into declaring an action, if not actually saying "why don't you make a History check to see if you know more about it?"
I DM for new players in one-shots, too. I just ask them to tell me what they want to do (per How to Play). I'll take care of the rest.
I don't know if it's a style difference or difference between one-shots and long-running campaigns, or just a difference in players, such is the nature of anecdotal evidence.
To be clear, calling for checks when I haven't even described an action for my character is the deal breaker. It's bad in my view because not only does it mean the DM has not read or internalized the section on "How to Play," which seems pretty fundamental to playing a game of any sort, it means the DM is not even aware what calling for an ability check necessarily establishes - that my character has done something with an uncertain outcome and a meaningful consequence of failure. Something that I haven't been given an opportunity to describe or otherwise establish, which means the DM is effectively playing my character for me.
I think that's the significant difference and the source of the misunderstanding then. Part of it is the awkward way in which D&D handles non-active skills. Part of it is a style difference, and in the lack of appreciating (or simply just not liking) passive skill checks, and the way nobody at WotC figured that that might be a good mechanic for knowledge-type skills also. Especially since, in many cases, ability checks that are usually active (particularly in the case of tools, but others as well) might call for a passive check related to knowledge of the craft/field. I personally don't like withholding information that characters should know (or at least have a chance to know) but their players aren't able to make the connections on their character sheets.
But there's a pretty big difference between calling for a roll that is essentially passive and declaring a player's action for them, which I think most of us can agree is far from ideal, at best.
To be even clearer, my position is not some principle that I apply to all games as Celebrim has suggested. In fact, such a suggestion completely misunderstands the underlying notion that each game should be, in my view, be played differently according to their rules and procedures. This is simply what is called for in D&D 5e. Nothing more, nothing less.
You see, to me it would make much more sense if it were more of a principle than a quibble with how the rules are presented in one specific game. The former is all about personal preference and playstyle, while the latter just seems... pedantic. But then I guess that too can be part of a personal preference.