TLDR is in one of my previous comments and is the following:
'More difficult modes obviously should give more rewards' is misconception. Some developers have made such a mistake in the past, they gave better rewards and more character experience on higher difficulty levels. The results were actually not surprising: higher difficulty modes we exploited so much that became easier as better rewards were outweighing higher difficulty.
The opposite is actually true: players who are good enough to beat higher difficulties do not need better rewards that would widen the gap even more. They need handicaps.
I think it is better to give more explanations. What is difficulty? When someone is talking about difficulty, whether it is easy content or non easy content, he is talking about relative difficulty. He is not talking (and likely doesn't care) about absolute difficulty. That means that he is talking about difficulty for him. When someone wants more difficulty, he wants more difficulty for him (because now it is too easy for him). If someone wants less difficulty, he wants less difficulty for him (because he struggles through the content). That relative difficulty is best described as effort.
What is effort? Effort can be described as a difference between absolute difficulty and character effectiveness. At the same time character effectiveness consists of two parts: mechanical effectiveness (character stats and everything that modifies them including equipment, food, portions…) and player skill.
The first part of the following text will contain the main principle and an illustrative example without direct relation to ESO. The second part will describe how to use it in ESO case.
So,
first.
We have mechanically weak character at the beginning (almost zero mechanical effectiveness) and a new player (almost zero player skill). We have a period of time that will be named 'game' but it can be scaled to parts of game with precautions. We will see how later. During that period 'game' mechanical effectiveness will rise. At the same time player skill will rise too. We will have something like fig. 1.

Fig. 1
Some notes: first, don't ask the numbers. They are for illustration only. Second, it is definitely not a straight line in any real progression. We need this line as a base line as all others will be shown relative to this line. It can be a curve – and likely will be a curve – but we know that it is an increasing curve. As such the most simple and easy to understand is straight line.
That is character effectiveness, that changes from the beginning of the 'game' till the end. There is also some desired effort for normal difficulty. The normal difficulty is red on fig. 2.

Fig. 2
Effort for normal difficulty is a difference. Usually this effort is approximately constant through the 'game'. Sometimes effort is lower at the end (for example, fight with Molag Bal is easier than fight with Mannimarco). Sometimes effort is higher at the end (for example, fight with the last boss in the dungeon is usually harder than previous fights). Both situations can be achieved either by changing absolute difficulty or by changing character effectiveness (mechanical effectiveness to be more precise). Since changing of player skill during the course of the 'game' can be different for different players, that will also affect the perceived value of effort.
Those are some of the examples. In future figures I will use only fig. 2 as a reference, but any of those can also be used instead.
Now we imagine not a new player, but a player who already has skill. He may have skill from other similar games or from playing this game once or more. That will place his starting point somewhere above the starting point of the new player. That will give us also a starting point of the new difficulty with the effort likely bigger than the effort of normal difficulty. Our desire then to get approximately the constant effort through the game. We know how absolute difficulty will change. For example, fig. 6.

Fig. 6
Green line is new difficulty. Yellow line is new desired character effectiveness. The main problem begins here. This is only a desired effectiveness. Everything can go wrong.
For example, lets imagine that difficulty was increased by increasing of health and damage of the enemies and we use learn-by-doing system for leveling up skill experience after every successful skill usage (that is not an exact reproduction of ESO of Skyrim system, it is more simple model). Since health of the enemies is now bigger and damage of the enemies is bigger too that means that player will need to use attack skills and defense skills more. That means that usage of skills increases. That means that skills level up faster. Now we have two different situations. In the first situation character level depends on character experience. In the second situation character level depends on skill levels. That means that in the first situation character level will increase at the same level as in normal difficulty if we give the same experience (sit. 1a). That also means that in the second situation character level will increase much faster than in normal difficulty (sit. 2). The same will be if in the first situation we will give more experience in harder difficulty (sit. 1b). If we have some kind of level scaling that means that in 1a situation we will rapidly overlevel enemies (skill effectiveness increases faster that character level), fig. 7.

Fig. 7
As you can see, effort on higher difficulty decreases, becomes zero and becomes negative.
In 1b and 2 situations with level scaling we will get that, first, an absolute difficulty increases faster than expected (fig. 8), and second, that at some time we will hit the level cap. If we need five times more skill uses to defeat an enemy then we we get level cap five times faster.

Fig. 8
Now what if we don't have level scaling? Then in situation 1a character effectiveness rises faster than we expected (due to increased skill level ups) while initial changing of the difficulty was rather simple and automated. Now we need to adjust that changing to find a balance and likely change something else that will change character effectiveness line that will force us to change difficulty again… We are now in the iterative process much like the same as for normal difficulty that needs to be done separately. In situations 1b and 2 we have both problems at once: first individual tuning, then level cap.
That's not all. The discussed problems were the result of mechanical effectiveness changes. Now remember that we also have player skill. And since player needs more time to kill an enemy, he needs more skill usage, he will learn faster. Not five times faster if five times more usages, but it is still an increase that cannot be ignored.
Fig. 7 will be something like this

Fig. 9
Fig. 8 will be something like this

Fig. 10
The situations without level scaling will not become better too. They will become worse. As I said earlier, changing of player skill during the course of the 'game' can be different for different players. With increased difficulty it is much more visible. Even if there is balance with good constant effort for some medium rate of player skill changing, there will be those who are not so fast in learning a game and will suffer from increased effort (much more increased than from the same problem on normal difficulty) and there will be those who are faster than medium and will say 'easy' (even earlier than with the same problem on normal difficulty).
Note that only character experience (to keep the same or to increase) was discussed as an impact factor. Add 'better rewards' to the harder difficulty and you will have even worse results.
So, the
second. What about ESO? ESO has two discussed problems, overland difficulty and dungeon story presentation. What do we know about overland? That new player can start from any story (chapter) and will not suffer from difficulty. All of them are level scaled to the new player. That is unlikely to change due to marketing reasons. That means that an ideal player will get something like this (fig. 11)

Fig. 11
Where blue is 1st story (chapter), red is 2nd story (chapter), yellow is 3rd story (chapter) and green is 4th story (chapter) in the order of playing selected by player, upper lines are absolute difficulty. If we level scale without taking into account changing of player skill (in addition to mechanical effectiveness), we will get slightly increasing lower lines that may or may not cross upper lines. If we take changing of player skill into account and player is not ideal (has another changing of the player skill), then lower lines may go either up or down. That is for new player and new character.
What if player is not new? What is character is not new? The problem may be described by fig. 12.

Fig. 12
Here player is not completely new so he starts upper, yet still below the proposed difficulty. Then he goes for dungeons – trials – PvP and returns to 2nd story (chapter), after that he again goes for dungeons – trials – PvP and returns… and again. Character effectiveness increased. You say, 'but level scaling', right? That will not help much. Level scaling will try to balance mechanical effectiveness (based on such a simple thing as character level). It is not almighty already. But there is also player skill. And you will likely get something like fig. 13.

Fig. 13
And if gaining player skill through stories is not taken into account in level scaling, you will have something like this.

Fig. 14
That is because level scaling doesn't take into account gaining player skill through other (potentially endless) activities.
Another problem are dungeons. The problem of difficulty of dungeons is even worse as it is repeatable content (that's why we can use the first part for reference but should remember that player skill changes every run while absolute difficulty is not). Not only player can do dungeons in any order (just like chapters), player can also repeat them. The first result is player skill that again is not taken into account (it is assumed by character level that is far from correct). The second result in unexpected mechanical effectiveness. Unexpected (or out of control) mechanical effectiveness is the result of the better experience and better rewards. Lets remember, base game normal dungeons are expected to (have absolute difficulty) to be done with crafted or at most overland equipment. Base game veteran dungeons are expected to be done with the addition of set equipment from normal dungeons. Then we have 'better rewards'. We have an addition of monster helmets. We have significantly increased player skill from numerous runs of veteran dungeons (to get right monster helm and get keys for right monster shoulder) and significantly increased mechanical effectiveness. The results are 'its too easy' (it needs no effort, the character effectiveness is higher than absolute difficulty) and DLC dungeons as an answer. With even better rewards. Then even better 'better' trial rewards are used in the same dungeons, even more increasing character effectiveness and even more increasing need for raising absolute difficulty of new dungeons. That is power creep as a book example!
And then – surprise – the same over-the-top absolute difficulty dungeons became part of the story of the same chapters that are at the very bottom and are also prequels to those chapters! Wrathstone to be done before Elsweyr that is done by new player new character. Harrowstorm to be done before Greymoor that is to be done by new player new character. And that is not working good.
So,
third. Conclusions. Results.
Equality of stories (chapters) and level scaling are extremely unlikely to change. Any solutions should take that into account. Absolute difficulty change from normal to hard or vice versa should be done without iterations that appear when change of effort leads to change of character effectiveness. Mechanical effectiveness is the part that can be controlled the most. Change of mechanical effectiveness because of change of effort can be minimised or removed (speaking of the example above, 5 time more usages of skills on higher difficulty should not make skills level up faster than on normal). Influence of effort on player skill (and vice versa) almost cannot be removed (even with V.A.T.S.-like system that will not work with ESO there are choices in using sets and using skills that are a result of player skill).
If players are advocating changing difficulty because they don't like the effort they should advocate level scaling done better. Level scaling that takes into account player skill. Level scaling that takes into account not only one character but all characters of that player. Level scaling that knows the difference between 810CP from trials and 810CP from crafting. One of possible ways to do this is taking into account all corresponding achievements account wide.
Then, since level scaling changes player character stats and not enemy stats, players should be ready that level scaling may not only increase but also decrease player character stats to match needed effort.
Then, since extra character experience and better rewards need even more level scaling, players should be ready for 'various but not necessary better' rewards.
Last, this should be optional.