Too many DMs, despite speaking so highly of the importance of stakes and how choices have consequences etc. seem rather precious about making sure that every fight goes according to plan, such that they'd rather deceive their players than allow a fight to be an accidental curbstomp (whether for or against the players.)
I don't think the normative language ("too many") is necessary. It seems more productive to recognise that you have a different priority from them - their priority is that the desire "story beats" occur and be experienced by the players; your priority is a type of "integrity of the challenge" that would be disrupted by the kinds of in-the-moment changes/fudging that are being discussed.
I explicitly referred to doing it in play ("once battle is joined," which I said twice) so it feels more than a little unfair to skewer me on "but clearly this can happen BEFORE play!" Statistics that haven't entered play yet are not inviolate. I have no problem with that. I am specifically and exclusively talking about modifying a creature's statistics or rolls during combat.
Here's an excellent example of why normative language is best avoided. Your willingness to allow changes of statistics before entering play works for you. But there are approaches - some of the classic approaches that I mentioned upthread and
@Thomas Shey commented on - where
what is in the GM's notes, and particularly the dungeon map and key,
are expected to be inviolate, because part of the point of play is for the players to engage with and work out that hidden information, and they can't do that if it is subject to change.
To be clear: I am a very poor practitioner of that classic style, very rarely attempt it (unless you count my recent Torchbearer sessions) and am not advocating for it in this thread. I'm just using it to illustrate the point that different groups of RPGers have different play priorities, and those different priorities make different techniques - including changing stats during or before encounters - more or less useful. And so it seems more helpful, in discussion, to bring out those features of different techniques and their utility for different approaches, rather than to adopt normative orientations towards them.
I have never--not one single time--needed to fudge a single roll or secretly alter a single creature's stats, in four years of gaming (with a fair number of "we need a week off" breaks now and then, but definitely not a full year's worth of them). It is not necessary to do this
I mean, the words "needed" and "necessary" here really mean
wanted and
desirable. It's about preference and play priorities.
In my Traveller campaign, I have tended to make up the geography - be it local, or interstellar - "just in time", often in front of my players with them aware that I'm making it up, because it is no part of the play of that game to uncover hidden geographic information. But when I recently GMed a session of White Plume Mountain, and running Torchbearer over the past several weeks, I have religiously stuck to the prep as best I can, because part of the point of those RPG experiences is for the players to cleverly engage with the pre-established geography of the setting.
it is absolutely deceptive to do it. Coleville himself openly said he will fake dice rolls so he can "prove" that the die "really" rolled what he said it did, even though it didn't. That is actively deceptive and, thus, cheating.
There are all sorts of deception that are not cheating, from parents telling kids about Santa, to teachers fudging answers to questions about what will be on the exam, to white lies told to smooth over socially awkward moments, to the performances of stage magicians.
Clearly you do not place the same priority on
GM as deliverer of a performance and
RPG play as experience of a tightly pre-scripted story as Coleville apparently does. That's fine. As it happens, nor do I. That doesn't mean that Coleville is
cheating. The inference from preference to norms, here, is unsound.
he (and a lot of other DMs) design a lot of monsters and encounters and playtest them in their main campaigns, which they don't want to completely destroy by a single encounter.
If you accidentally homebrew an overpowered monster for your players to fight or end up adding too many enemies to a single encounter (which could lead to a TPK), it is perfectly valid to adjust their stats mid-combat and fudge die rolls to make them a more appropriate challenge for the party's specific level and class composition.
Speaking a bit abstractly - thinking across a range of RPGs beyond just 5e - there are other possibilities here. For instance, in Burning Wheel if a monster introduces a new creature, a player is entitled to ask about how many "points" of build it has (there's a bit more to it than that, but I hope I've conveyed the idea). In Marvel Heroic RP, the stat block for an opponent is normally much more public than in D&D. In many systems, there are approaches to consequence narration which mean that
harder than expected doesn't entail TPK, let alone "game over".
My reason for making this point is to further illustrate why I think talking about play priorities, and how different available techniques - some of which are system-dependent - support or undermine them, is a more profitable approach to discussion than a strongly normative orientation.