A good point, and it's what makes me wish the GDC and the like had more insight into MMO development, specifically, because I can't seem to find enough samples to draw real examples of what good prototyping and bad prototyping each look like. Thus, I'll have to fall back on hypotheticals and abstracts for now, which is... far from ideal.
Let's say we're building something like Eureka in the next year, but with a lot more features and a more compelling, less grindy experience intended by its apex.
"Good" prototyping:
Player's attentions shouldn't be drawn to new modes of play until those modes are fleshed out well enough to be compelling. You want to avoid complaints by adding the modular components you wish to test prior to the new mode's release into existing modes that aren't flush with criticisms. (The only exception would be if there's a heavily criticized mode into which the components can fit and would fix some of the issues perceived in that mode, in which case you start there instead.) In that way, they seem like an unexpected treat in something players are already used to. (Or, it seems slow progress towards polishing the existing game, rather than abandoning (n) for (n+1), which tends to prompt increasingly soured or more critical perceptions of each new mode.)
Regardless of overall design philosophy, feedback tapers off with age, so any flaws in a new mode are going to have an even larger impact on its player impressions of competence than will the positive effects of the mode's novelty (and that's assuming, perhaps wrongly, it is even truly unique or novel), so it needs to feel like a complete experience upon release. This does not mean that every new feature needs to be in place, but there should be no obvious issues onto which players can latch and the components should play out newly enough not to be met with complaints that the content is just more of the same or repackaged mechanics, etc.
So, let's say wanted to make a compelling experience of Anemos for the start. A few more features would likely be necessary. We'd want to include these features in a way that doesn't necessarily lead players to expect that new content is on the way (as a side-effect, to be clear, of having a reasonable fit and purpose among existing modes).
To take a spitball approach for now, let's say we start by making a big deal out of the upcoming Monster Hunter crossover; we can claim to be preparing the way for it or just leave our motives implied at best. At any rate, we start by revamping our hunt system somewhat. It reintegrates B, A, and S ranks with the general hunt board and Hunt Logs, introduces a tracking system for A ranks that works a bit like a miniature quest chain wherein your party hunts down these mobs, and scales hunt difficulty and rewards when tracked to the item level of your party, moving away from hunt trains and towards scaled, less zerg-ish party content.(Reader insert here: would be great if this ultimately culminated into a guildhest as a great way to re-invigorate that content that seems to collect dust) If possible, we also try to include fixes to load priorities so players can more easily see their own party and the boss when in massive fights. To the players, this already seems like an answer to common complaints about the hunt system. In truth, it does, but it also paves the way for a more interesting experience in Anemos.
In our next preparatory patch, we introduce our Monster Hunter bosses as sort of the apex experiences for the previously given system -- an S-rank expansion of the tracking system that now involves multiple interwoven party tracking quests. One party prevents Rathalos from grabbing and running off with... giant buffalo or what have you... for food. Another besieges its nest, and a third finds a way to lock it within once it's drawn back, etc., etc. Now, that's more resource-expensive than simply making a standard Extreme fight out of it, but it gives you opportunities for Anemos and beyond. Further, make Rathalos drop actual crafting ingrediants to be used in making his stuff; anyone can talk to the hunters, one of which now has a work station by some forge in Kugane, to have the stuff the made for them for a fee, but players with an especially well-geared crafter can use fewer materials to guarantee an HQ Rathalos item or some similar reward. (Again, Eureka prep.) On the more casual side of things, you increase mob kill EXP in the open world and bring back Behests, which now allow for a bit more free-form patrols and forays and allow successfully cleared encounters to spawn stronger monsters, up to the point where parties can enjoyably EXP grind in some areas (albeit still not as efficiently as dungeon spamming, but often made up for if the mobs carry useful crafting ingredients). On the surface it's sort of a change towards "Play as you like" and better utilization of the open world, but it's also pivotal for making Anemos mob-slaughter feel more intrinsically rewarding.
So by now we have just enough to make Anemos truly compelling. All we need now is to not alienate broad sections of players (e.g. those with friends who play [Eureka] more often or less often than they do and design the map and mobs in such a way as to make it feel more about danger than grind. Needless to say, you squash the total number of levels, and/or the leveling curve, and allow for level sync to one's party. You tune mobs more for risk-reward than grind-efficiency by giving them powerful enough stats, skills, and maybe even some AI or obligatory links (pulling one pulls the pack) for variety. You use the spawns of more powerful mobs, prior to the NMs, to vary the grind experience in a given mob zone. You include tracks system triggers and clues to nudge players out of their comfort zones and into more or less populated spaces as useful to preserve the intended gameplay. Etc., etc.
Though, this also means picking the right threads to start from and work towards over the Eureka series. If you bother with a system like the elemental wheel, for instance, you must make it do more than just punish fighting in zones where you're likely to pick up multiple mob types or where tanks have to excessively compromise between enmity (as a consequence of damage) and defense. If you want to make the prior work, make it a real threat that can't just be toggled about. Otherwise, skip it, and spend that time instead on, say... turning Anemos more into the Land of the Winds through some unique mechanic (though you can certainly have it be less of an impact in your introductory zone than in later ones) and developing craftable gear that makes use of the Anemos NMs as per what the Monster Hunter crossover (and Rathalos) started.
You can then guess where Pagos will go: more monsters directly usable for crafting, and perhaps more horizontal progression or the like, such as to set up camps to deal with the harsh blizzards (not everyone needs a Cloak of Winter or the like, but someone will at least need a Winter's Shroud tent which they can later disassemble into personal items). Rather than just grinding your way through everything or only doing your challenge logs once per week you'd have things to target. I want to kill a Dragon because... Heart of Winter and that really awesome Dragoon armor. We need Wendigo furs for... a community base-camp up in the northeast. We want to kill the Spriggin Queen to steal her consumables for cloaking ourselves in the snow or to turn into a craftable which, when equipped, gives us access to that skill via a CD. Etc., etc. Rather than just throwing in systems for the sake of systems or artificial longevity, we try to engage more with risk-reward than mere gather-grind efficiency and give each mob a deeper impact on the surrounding mode, e.g. through the skills it reveals to you (which higher rank versions are likely to repeat with even more danger and less warning), what it helps spawn, and what you can craft from it.
To an extent, generalization, such as what the Eurekan kettle did for us, can be a good thing in that it allows anyone to join a party anywhere in the zone -- just as a squished leveling curve would have helped in Anemos, especially in the absence of leveling sync. Relic gear is of course iconic and generally liked. But there needs to at least be something to vary that experience, rather than just '95% literally any mob' and '5% Pazuzu', or the like. Everyone should make use of the top NMs in about the same proportion, but there are all sorts of things you can do to maintain accessibility while providing variety. For instance, what about custom crafting? Maybe I can use more Wind NM ingredients and get different Relic armor effects (at least within Eureka), similar to the HW "build your own Relic('s secondary stats)" system. (Of course, personally, I'd prefer to skip the Relic armor in favor of Eureka-specific Relic items usable by all for their ilvl, but adding to the Eureka experience, especially as we change what all can be done with the Eureka zones over the course of the series, but I get that for many content merely equates to a way to most efficiently gear up or to most efficiently gear up when broke and unwilling to run remotely difficult content, so... idk).
For contrast, "bad" (or at least mediocre) prototyping:
What we got.