I'm always somewhat hesitant to dig into armchair analysis of such play examples, especially for improv-based games, because the analysts are able to pick at nits while missing out on the emotional table-state and known player qualities and not suffering the time and performance pressures. But here we go.
You're fine. Doesn't bother me unless I'm fairly certain the involved party isn't sincere. I know you're sincere, so pick away. Besides, I invited it and I think this is good (and topical) conversation.
Absolutely! and the best way to mechanically address this leverage would be through a bonus to the roll! Oh wait, that's the wrong type of rule mechanics for this game. Although not forbidden, such probability adjustments are discouraged. Follow the story where the dice go and if the situation is trivial, don't bother with a roll is closer to its credo.
So based on your description I would expect no roll would be necessary to appropriately persuade the dog to take the food. If a roll is called for, then obviously there are other factors at play which will be revealed on a 9-. Like the fact the dog isn't a dog, isn't in the condition presented, isn't alone, will attack regardless, or perhaps the dog will flee to its death rather than deal with a human.
My thoughts here:
1) As you mentioned, the relevance of the Leverage
can be an input to action resolution (take +1 or take -1). However, as you also mentioned, it is discouraged as being a mere input to action resolution. Therefore, we're left with it being an impact on the output of action resolution.
In this case, that output would be "what happens" on a 7-9 or a 6-. So, for consideration I've got (a) the holistic approach of DW's Agendas/Principles, (b) the overarching fictional positioning to account for, and (c) the specific fictional positioning of the leverage in question. On (c), my sense of it as an output for action resolution is to serve to mitigate failure prospects specific to that Parley move. That is, like
Instincts in Burning Wheel, this Leverage (along with all of (a) and (b) above) contracts my prospects for for framing the fiction post-move resolution.
Now that doesn't mean that I can't make things suck post-resolution. It just means that I feel constrained such that I shouldn't make things suck post-resolution in a specific way (the dog going full aggro/hostile). I aslo think part of what may be happening here is maybe you're showing a certain predisposition toward tight coupling of causal logic in action resolution that I don't share (our conversations have gone in this direction in the past). In this case:
Parley = Task input to action resolution (this has primacy...not context, not intent, and/or not genre logic)
Fortune = Failure
Thefore...
Output must = Consequence of failure should mean adverse NPC attitude toward PC (possibly manifesting hostility, but either way, closing off social prospects).
In my case, I take the position that the the components in parentheses should have primacy here. There can still be conflict in Dungeon World even if the dog is predisposed toward wanting companionship, security, food. It can just be less apt to do so given the overarching context and its present circumstances. Therefore, the danger in Parley doesn't have to be hostility of the NPC. It could be any number of dangers; in this case the wilds, the weather, and the gathering dusk that the creature fled to. Dungeon World is predicated upon this kind of stuff snowballing into orthogonal content generation/danger introduction.
So "say yes" wouldn't need (and I say shouldn't) apply here. I don't think the situation was trivial at all. The situation was definitely fraught with peril and there was danger to be had with no guarantee of gaining the asset (despite the predisposition and the canny Leverage the PC deployed).
Finally, Blades in the Dark has introduced further formalizing of GM adjudication that you may like (I think its helpful). Each situation's fictional positioning is declared as "controlled", "risky", or "desperate". The complications/costs/results of action resolution become more punitive as you move from the left to the right. For instance, "controlled" would be when you're on your turf and holding most (but not all) of the cards. My opinion going into that play moment would definitely have been that the situation was either controlled or somewhere between controlled teetering on risky.
2) As I re-examine this moment of play, the only real prospect of removing the asset from the fiction (because I don't believe hostility would apply and running headlong into the reindeer herd would be absurd) would be a predation from below. The Ranger (and high Wis Ranger at that) shouldn't have a hard move introduced where an avian or land-based creature (specifically on an open glacial expanse that isn't under fulldark) swoops/sneaks in right on top of the PC and takes the dog. That would be akin to violating an
Instinct in BW. So the only option for outright removing the asset would have been a subterranean predation. Problem is, I wasn't even aware of the Remorhaz at that point (I introduced evidence of the creaturet the next afternoon I believe). So it didn't even occur to me.
If I had thought of a Remorhaz attack (the ground beneath begins to subtly rumble, geysers explode in steam all around, etc) and it consumed the dog as it exploded from the earth (then potentially threatening the PC), I'm assuming you would have found this more palatable (not from a process-sim perspective, but from a "locked the asset out" perspective)? I may have done that if I would have thought of it. Sometimes you're limited by the best bit of dangerous fiction your brain can offer you in the moment (even though I still think the route I went was best all things considered).
And this is my primary complaint with the system. It places undue reliance on the GM to adhere to the game's principles and provides few checks the players can use to detect or correct for variance. A GM can trivally insert soft moves and moves coloured with his own expectations to guide players around by the nose with similar effect to an outcome-based game like D&D's DM using illusionism and fudging. This is made worse in some ways because there is no secondary check method (such as rolling in the open) that can be used to constrain the behaviour. Quite often, the GM would need to drift the situation quite far to get off genre and thus become detectable.
I think holistically the system is considerably more coherent, transparent (the machinery of action and resolution should be apparent for players), and wholly easier to "play to find out" than you give it credit for here. Trying to apply force seems to be much more arduous and flat difficult than just playing straight.
That being said, Blades in the Dark has introduced "Clocks" in a different way than the Threat Clocks of Apocalypse World. These serve as a physical, player-facing countdown to failure/escalation as a conflict is introduced. When the clock ticks, things get more and more pear-shaped until all the segments are filled or the scene has been resolved positively for the players.
50 mph for a couple hundred meters, tops. Typical running speed is half that and typical herd travel speed is half that again. I'd expect the tag to shift from far to near or at worst close next round which makes them a threat but not an immediate one.
Well, on this we get into the abstraction of the Near and Far tags. Near is "can see the whites of your eyes" and Far is "shouting distance". To firm up the fictional positioning when its required for players to make informed action declarations, I typically default to 10-15 meters for Near and beyond that out to 100 meters for Far. Real world hunters typically stalk and kill between 30 and 40 meters. Proficient archers can hit targets with compound bows at about 50 to 60 meters. 60 meters is well inside the limits of being able to understand someone shouting. All told then, I think 100 meters would apply and is more than fair, IMO, for fantasy genre hunters.
Our fastest speed in the 100 is about 28 MPH. You're talking inside 10 seconds. Move that up to a 50 MPH sprint and you're effectively talking in your face in short order. The prospects of escaping by foot-speed seem rather grim. You're almost assured of being trampled by those 10 reindeers that can move about double your speed. If a player would have declared that they were going to attempt to outrun them, I would have declared "allright, but (a) you can't get a 10+ and (b) you take -1 (this would definitely be a situation that would apply)."
So, I suppose here my sense of "immediate" may be slightly different than your own. My sense is that you're Defying Danger right now if you don't have some other method of eggress (more on that below).
I think it'd apply here if the player had rolled a 7-9 total not a 6-. Partial success being of course, you have gained the dog's trust and it is open to you; how do you plan on saving its (and your own) life?
I feel a 6- should remain a failure. "The dog backpadels from you baring its teeth; the beast has obviously had bad interactions with humans in the past. Gaining its trust is going to take much longer if it is at all possible. <Cue herd of reindeer> what do you do?"
There is very limited material differences between presented result and a theoretical 7-9 result assuming the same threat introduction move was made and the player got the primary goal: the dog became accessible. The failure was short-circuited. It coupld be because the first thing that came to mind and the GM needs to keep the momentum going. It could be the GM is an animal-lover and would prefer a good outcome. It could be because the GM knows the player is an animal-lover and would take a bad outcome more poorly than is desirable. It could also be the GM has future plans for the dog (and the alien egg growing inside it bwah haha) and doesn't want to see that opportunity lost. Why doesn't matter much. The failure was short-circuited.
If she would have gotten a 7-9, my response would have likely have been:
"Alright, you give the dog some food (mark off 1 Ration). It scarfs it down and looks at you for more. Sating its appetite and earning its trust will definitely require more than a single Ration."
So you're looking at a situation where the player gives up a 2nd Ration. Most D&D groups don't know what its like to be "Ration-stressed." Dungeon World games exist in that state perpetually so this isn't a small resource expenditure. Further still, this group was inevitably going to be Ration-stressed beyond normal (limited areas to resupply and in a wasteland bereft of foraging and hunting opportunities. So every extra Ration spent is costly. Exacerbating things further still is taking on the extra mouth of the dog. Further still, the Ration-stressed situation would be exacerbated the next day as the group took on several more mouths to feed! Coin (which they were also desperately short on, so they would then become "Coin-stressed") and Resupplying (in Earthmaw) would become a desparate thing.
Finally, the ultimate cost of the whole ordeal was:
1) 3 Ammo
After a quick look around at her options, Saerie determined that the reindeer were probably maddened and indeed making a bee-line for her. Further, the snow-drift was likely her only real chance to escape. She grabbed the dog, dove into the partial burrow and collapsed the roof on she and the dog. She rolled a 9 on her Defy Danger. Consequently, in the doing, the frantic dog squirmed and kicked and upended her quiver. Out spilled nearly all her arrows (only 1 Ammo left..so now "Ammo-stressed"...for a primary archer Ranger). She could snag them and face the danger of the herd or sacrifice them and stay put. She chose to sacrifice them.
Now this could have easily snowballed very badly if Saerie would have failed her Discern Realities or Defy Danger effort. So the danger/threat level of this situation was not only not insignificant, it was positively deadly.
2) 2 Rations
She had to pay the equivalent of the 7-9 Rations for the dog anyway.
In the doing, she gained an old, deaf dog as a Cohort/Hireling which she would have to regularly pay its Cost to keep up its Loyalty. The dog could help her on Hunt and Track (take +1 but you can only get a max 7-9) or grant her a +1 to Defy Danger (with 7-9 max result) where the fictional positioning warrants it...but it would put the dog in peril.
I don't believe she may have used the first feature once and the second perhaps twice.
Further, she would get some minor intel (mostly independent confirmation on what they found in journals/records in the village or what they had surmised/deduced) on what transpired in World's End Bluff.
She would form a bond with the old dog though, so when he died it was a big deal (hence her Spirit Realm journey to commune with his spirit).
So I guess what I'm getting at with the above is:
* I don't think "say yes" was the right approach here despite the situation being the equivalent of "Controlled" fictional positioning in Blades in the Dark.
* I think Leverage as an output (constraining or opening outcomes contingent upon Leverage deployed) to action resolution rather than an input (eg take +1 or -1) is the way to go.
* The move certainly felt much more of a Hard Move to me than it seems it was to you.
* The ceiling cost (high likelihood of death or at least considerable HP damage or Debility) of the move made was even higher than the cost that was ultimately paid (which was significant from a resource perspective).
* I think a Dungeon World that uses Blades in the Dark's Clocks as conflict resolution would likely be much more palatable to you as a gaming engine (or hell, just use all of BitD's machinery rather than DW).