By "boxed text," I mean the design principle of including, in an adventure module, a section at the beginning of each keyed area which describes what the characters see -- before, then, going into details for the DM about the reality of the situation. Boxed text is written in second person: "You" see this, as opposed to "the characters" see this.
In most cases, the boxed text is used as read-aloud material -- the DM simply reads the description to the players.
As I run through what I think the pros and cons of this approach are, the beginning point is to look at what the text is intended to achieve, and WHY it's intended to achieve that. In other words, when we talk about why, there's an underlying assumption about how a module is supposed to be played -- or, at least, what compromises have been decided upon by the authors of the module in terms of how the module plays most effectively.
So, if I'm going to make any sense with this little article, I have to step up and mention how I think a module ought to be played. That's a bit more than subjective in my case (as editor, publisher, and writer), because you can't write a module without having some of these assumptions in play. A module can't be written if you're not keeping an eye on how to make it work best, and there are choices. If you try to please everyone, you're writing by committee. This has to do with several features of a module that might not be immediately apparent to a reader, including even such things as the size of the font being used.
How I think a module plays best
The assumption I work from is that the strongest gaming moments are when the direct communication between the players and the DM is at its highest level. There's eye contact. The DM is able to answer questions and announce results based on his/her own mental picture of the situation and the environment, without stopping to read something. There is the immediacy of a back-and-forth conversation rather than the sudden gear-shifts involved in changing from direct speech into reading aloud or referencing something. I'm not offering this as a "rule for DMs," I'm just making the observation that -- whatever level it reaches with a particular DM or group of players -- these tend to be when the action is fast and the gaming is strongest.
The upsides and downsides to using this approach as a design rule (and there are downsides to playing this way) are what drive the following opinions.
In Favor: Point #1
The first point to be made about boxed text is, I think, very simple and clear. In the older style of modules (mainly the G and D series, but also B1, etc), boxed text didn't appear. As a result, when the module is actually in play at the gaming table, the DM either has to remember the salient details of the encounter area, or else stop the game temporarily to skim the material. I happen to be an insanely fast reader, and even I have trouble doing this without pausing, or else I lose focus and might miss something a player is saying. Here, I mean when I'm skimming the next room while players are still only approaching it. So here, we avoid one gear-shift. If there's boxed text, the DM doesn't have to suddenly stop, fall silent for a couple of moments, and read (or skim) the room's text.
In Favor: Point #2
This is very much related to point #1. It's about missing an important detail. The faster you remind yourself about what's in that room, the higher the chance that you'll forget to mention something vitally important about the room's description - something the players need to know is in there. The risk is exponential -- it's the complex rooms that require the most reading-through because the room description is longer, which means that they will get skimmed faster by the DM right before playing it, and yet these are the rooms where a subtle detail is more likely to show up -- they are the complex rooms. Boxed text eliminates this risk if it is read aloud.
Against: Point #1
Stopping to read aloud is obviously a shift of gears. It probably creates different levels of shifting, or different levels of problems, depending on the DM and less so on the players. If a group of players simply has a high tolerance for the shift from extempore to reading, it's not a problem at all. And if the DM is superb at reading aloud, it's less of a problem. On the other hand, if the DM has a quiet voice, or reads in monotone, the gear-shift is more pronounced. Whether it's a problem for an individual group, I think it's incontrovertible that the shift in gears exists, and from my perspective that has the potential to reduce the quality of the module in play.
In Favor: Point #3
Since we're talking about a DM who reads in a monotone, we also need to realize that some of us really suck at describing things off the top of our heads. I think I'm really good at this, which is why my initial assumption is that extempore descriptions are better. But many DMs, I think, buy modules precisely because their ability to turn their own visions into rich description isn't very strong. People buy a module for a reason. I don't think this is the most common reason, but it's definitely there -- not everyone is a good communicator. A DM who can read aloud like a silver-tongued devil, but just can't manage to think on his feet about mentioning little details like damp walls or echoing sounds, might definitely prefer to use the boxed text as his tool for setting the atmospheric tone, allowing him to focus on his other skills like monster tactics, or whatever.
Against: Point #2
And THAT brings us to the next negative point, which is that boxed text can force the DM, consciously or unconsciously, into atmospheric descriptions that aren't exactly what he wants to portray. Okay, that's a really, really minor point. Indeed, it's something I totally disregard when writing a module. I think a lot more people, even the ones who don't like boxed text, are still buying the module to see the atmospheric feel of it. Nobody wants to read a bland module, a module that has an inconsistent tone, or a module that keeps telling the DM "you can do this any way you want." If the DM is good enough to distinguish and portray nuances of atmosphere, that DM is almost certainly going to be able to handle the "conversion" from one tone to another.
Notice that there is a shift of gears whether you read the text aloud, or whether you skim it quickly to remind yourself of the important details before giving your own extemporaneous descriptions. You either shift to reading (ie, a change in the tone of how things are working at the table) or you pause for a few seconds to refresh your memory (a shift in the pace). A DM who prepares modules with great thoroughness might be able to play without refresher pauses. A DM with less prep-time can't do this, though. This is one of the stark choices to make as an author: who is my audience? I make the decision that more people are buying modules precisely because they have less prep-time available. Hence, that suggests that the modules should have boxed text.
The above conclusion, which mentions prep-time, brings up another important point. Everything I addressed above had to do with how the module plays at the game table. But it's also important that the module is easy for the DM to assimilate in the first place. Boxed text isn't just a tool for telling players what they see. It also puts the DM into the shoes of the players while reading. Boxed text can make reading and preparing the module more fun and easier to read and assimilate.
Boxed text can be written badly. Text descriptions of a room can be written in a disorganized fashion and fail to convey atmosphere. A badly written module can't be saved by minor questions like whether they should include boxed text or not.
There are some ways to mitigate those gear-shift changes I've mentioned as being negatives. For example, boxed text can be kept short, so that the gaming reverts quickly back to the extempore conversation across the top of the DM screen. Long, long flowery descriptions will, at a certain point, turn stale and boring. But if they're short, sweet, and evocative, then they will keep things moving as fast as possible, and provide the best compromise in terms of whether you're going to break the pace by reading aloud, break the pace by reading silently for a second, or (possibly the worst) break the pace by discovering that you botched some major detail in the room description.
So I come down in favor of short, boxed text (or italicized, or whatever). The ultimate question is really whether the module is well written, but assuming it's well written, I think this is the best way to make that well-written module function best at the gaming table as a tool for actual gaming.