All analysis is based on the following high-level assumptions:
Core loop = quest -> explore -> combat -> improve (stats)
Short-term engagement = consistent, measurable power rewards
Mid-term engagement = difficulty modes that push the player back into the short-term cycle
Long-term engagement = create multiple characters
Primary player behavior motivator: Achievement through power
Mission delivery: The implementation of the map-based mission delivery system, in conjunction with unlocking missions by using boards in those areas keeps the sense of large scale while reducing the overwhelming feeling of a map exploding with things to do. The waypoint system leading through the open world to the objective is clear and useful, keeping the player from feeling lost in the streets of Manhattan.
Co-op roles: Each skill line provides clear and familiar role specialization into Tank, Support, and DPS, which makes co-op more rewarding and necessary.
Base: The base is a fantastic and motivating way to track the player's progression. They get to not only watch a meter fill, but also see the changes reflected in the world. This serves to immediately immerse the player in the narrative and provides a method of progression outside of power that is satisfying.
Dark Zone: MMO logic dictates that the most satisfying reward is showing off your progression to other players. The Dark Zone is a unique way to express and show off your power.
Base: From what I've been able to gather, while a successful mechanism for expressing the player's progression, I think this feature was under-developed. Because it was mostly used as a location to spend skill points, it ended up being more of a hassle to have to return to then the centralized motivator it could've been. If you compare this base implementation to the X-Com HQ or Dragon Age: Inquisition's keep it is pretty shallow and superfluous.
New player difficulty: The Division is a pretty unforgiving tactical cover-based third person shooter. Based on this basic barrier to entry, I think they failed to provide enough starting options to players of various skill levels to feel successful.
Power curve: In its current implementation, with the type of gameplay the game has, which is highly skill-based, I would argue that power progression is a mismatch and should be removed. The moment to moment challenge focuses on managing positioning in relation to the AI, not the configuration of your gear. All the stats make the effect of you being good at the game better, but do not compensate for you being bad at the game. In my opinion, if you want your game to be about getting items, those items need to be a significant portion of the method of your success with skill taking very close second place. Borderlands does this expertly. This mismatch is one of the reasons the game gets complaints about the AI being bullet sponges.
These suggestions assume that the design intended for power-based loot to be the core reward mechanism.
Power curve: For me, starting player power is skewed too low and scales too slowly. This causes the early pacing of the game to feel slow and a bit punishing. I'd skew in favor of the first 5-10 hours being relatively easy and ramp difficulty up by decreasing the power ramp over time, while introducing new mechanics or stats to make up for the decrease.
Resists and damage types: One of the mechanics I'd add over time is damage types and enemy resists. These don't necessarily need to be elemental and fantastic. They could put in a simple system, like Mass Effect 2 & 3, where there are 3 resist types and weapons and abilities counter those. This would allow some forgiveness for the bullet sponge feeling since there would be context and a counter for it.
Enemy durability expectations: With a game that feels as skill-centric as this one with human enemies, I think they needed to reinforce the time to kill through the visuals of the enemies. When I see an enemy wearing a hoodie and jeans, I expect two or three bullets and definitely a headshot to take him out. If it takes multiple headshots, he should at least be wearing a helmet.
Increase weapon types & mechanical variety: Their weapon selection seems limited. I'd add another category of weapons, like rocket launchers, crossbows, or energy weapons to give more playstyle options, feel variety, and spectacle. I'd also move out of the realistic and into weapon mechanics that are fun and interesting, not as far as Borderlands, but maybe a sniper rifle that charges up.
Perk, Talent, and Ability upgrading: To increase the tail on progression, I'd add upgrade levels to the player progression systems. Upgrading these would require the same currencies used to unlock them in the base.
Base: I think the base could've been made more integral to the game by including it in content. If, for example, the base was under threat of attack, then you could add defensive and offensive progression leveled up by items and currency and have events where enemies attack it either synchronous or asynchronous. They could also add more flavor touches and customization options that sink currency. Basically, extend the functionality to tie it deeper into the reward loop. I'd also evaluate if there should be duplicate functionality outside of the base, like merchants in other places in the world, to make going back feel useful.