I am doing my very best to be concise, but there is a lot of things to discuss.
I would appreciate if you do try to keep responses to this post factual and correctly quote the part you agree/disagree with and why.
The current state of WvW is atrocious. I will go through all (might miss some) of it's problems as someone who plays in a small very organized group of 2-5 people (depending on the day), while also touching problems with running solo and what is currently wrong with zerging.
I will also briefly discuss what would fix these issues. I will try to avoid singling one class out as the cause of all these problems and try to provide as general information as possible. I will try my very best to avoid class balance as this is a subject likely to derail the thread and this thread is not about specific issues but about global game-wide issues, even if some of these issues are partly or fully limited to a subset of the classes.
I will also be listing the main offending classes (sometimes specifying build) for each category, note this is not a call to nerf any particular class, just me pointing out that they are some of the main offenders for one or multiple parts of what makes WvW bad. These could be missing some classes/builds as I might forget them, but the examples are mainly there to illustrate a point.
1.1 Builds with access to incredibly high mobility.
Some builds have so much mobility that trying to kill them on most other classes is an excercise in futility, you can make them flee/reset, but you will never actually get the kill, this means that they will keep resetting the fight until they eventually win, be it the 3rd or the 33rd try. This is both frustrating and very poor gameplay as your best option becomes finding somewhere relative safe (generally water/a tower/somewhere with LoS) and wait for them to get bored. These issues are further compounded on by that they are generally the same ones with access to stealths which will be discussed in subsection (1.2) and very high burst (which will be discussed in subsection 1.3)
My Solution: Make a lot more of the movement skills require a target so you can't use them to reset. Nerf ones where this is not applicable (1-2 movement options at most w/o a target seems reasonable/class)
Main Offender(s): Thieves (all specs), Mirage, Soulbeast, Druid, Warrior (all specs), Revenant (Shiro)
1.2 Builds with good access to stealth.
There is very little counterplay to stealth, you can try to guess where they go, but the odds of that are generally about as good as me picking 3 numbers between 1 and 18 and you trying to guess all 3 (360 degrees to go in, your margin of error is at best 20 degrees, they can change the direction hence the 3 times). Even if you do manage to predict where they go, the builds with access to stealth are generally not slow builds w/o any mobility skills.
My Solution: Make stealth give you an unpurgeable version of cripple that prevents you from using movement skills. This would make actually finding the person in stealth rewarding while making it so that when you don't they have at least cleared a lot less distance.
Main Offender(s): Thief (all specs), Mirage, Soulbeast, Druid
1.3 Builds with very high burst damage and strong "defenses".
This point is basically a combination of subsection (1.1) and (1.2), but extended to include all active defenses (such as Endure Pain for example). When builds can go for their full burst on you (which can 1 shot you if you don't play counter accordingly (and sometimes even then depending on what build you're running and what build is bursting you). Now if you after surviving their full burst could properly retaliate this would be far less of a problem, but this burst is generally followed up by chaining several mobility skills together with using invulnerabilities (or stealth or similar). Meaning that the sequence goes something like Burst>You survive>Reset fight>10 sec later>Burst>You Survive>Reset Fight>Wait 10 sec>Repeat. This is less of an issue when you yourself play the kind of class/build that can do this, but for those of us who don't enjoy burst>reset gameplay there are very few viable options.
My Solution: Implement my suggestions from subsection (1.1) and (1.2) and reduce the damage of the burstiest skills across the board (especially ones without a clear tell)
Main Offender(s): Thieves (all specs), Mirages, Soulbeasts, Warriors, Holosmith, Fresh Air Ele variants
1.4 Builds that combine some previous points with long range.
Once again the issue here is counterplay, the builds that can combine these things also generally do the bulk of their burst with unblockable, which removes yet another counterplay option, leaving LoS as the only real counterplay option. These builds are in my opinion not more of a problem in even numbered fights than similar builds w/o range but their range makes very oppressive when outnumbering the enemy.
My Solution: Ranged attacks should never be unblockable (or unblockable shouldn't affect projectile destruction/reflection). Ranged attacks should either be similar to that of staff elementalist (slow and telegraphed) or do much lower damage, there should be no 15k unblockable attacks.
Main Offender(s): Deadeyes, Soulbeasts, Revenant (Hammer+Shiro, so to a lesser degree than the others)
1.5 Overall Damage Proliferation
This point isn't targeted at any specific class or issue but rather at the fact that the game has seen overall damage go up, meaning fights become more burst, less attrition. In my opinion attrition based or "grindy" fights are more healthy for a game as they give each player more possibilities to play good/bad meaning that they measure "skill" in a better way.
My Solution: This problem is often due to a lot of stacked +x% damage modifiers, reducing the number of these in the game would be a very good start
Main Offender(s): Game-wide issue
The mechanics that currently reward this is:
2.1 The AoE cap on damaging skills
AoE skills are supposed to be the counterplay to multiple stacked enemies, yet the optimal counterplay to AoE effects is to stack more players in tighter blobs, which in turn makes single target effects all but useless (due to heals being AoE and "smart")(especially ones that can be bodyblocked). Outnumbering the enemy is it's own benefit, such a strategy does not also need a 60-90% blanket damage/cc reduction.
My solution is simple, just remove the AoE cap from any offensive skills completely.
Main Offender(s): Everyone
2.2 The way the downstate and ressing works vastly promotes bigger groups.
Ressing is something which you can't do better (except for Mercy runes/ressing traits) except for just having more people do it, it is also in the current state way way too fast to ress 1 player, even without mercy runes/similar, but when there is several mercy runes using players the downstate becomes what you really fight against as everyone you down almost instantly gets ressed.
My Solution: for this would be to cut ressing speed by 50-75%, remove ressing traits from the game and remove mercy runes from the game and in addition to this add a debuff on people currently getting stomped that reduced their "healing gained" by 100%. Several/all downstate skills should be nerfed to roughly the level of the Necromancer/Revenant ones (easily counterable, only 1 interrupt, can't interrupt people with stab or interrupt through blind) or completely remove all downstate skills beyond 1 and 4.
Main Offender(s): Druids, Scrappers, Guardians, Mercy Runes
2.3 Builds with a large number of "double duty" skills.
A great example here would be Scourge (there are other too), which prior to February 06th Patch was my favorite class and main. After multiple excessive nerfs scourge is still amazing at zerging, but is now terrible for anything else. The reasons they are great at zerging is because all of their skills are both offensive and defensive at the same time, while many of their skills have a target cap of 10. The dual nature of their skills means that there is never a "bad" time to use them, meaning that while the average person in a zerg might not use the skills optimally, they will be used "good enough" to be effective (Back when Scourge wasn't trash outside of a zerg, I considered playing one optimally outside of a zerg to be one of the harder classes to master, while being mediocre/good enough to be one of the easiest classes(ie a bad Scourge player was still a threat, but there was a lot of room for improvement)).
My Solution: Make more of these skills have an option to be either supportive or damaging, for example picking up "Path of Corruption" on Scourge might make f2 corrupt 2 boons like it used to, but no longer convert any conditions (or only convert conditions on yourself). Basically more tradeoffs and less skills that are mediocre at a lot of stuff. This rewards good decision making.
Main Offender(s): Scourge, Firebrand, Warrior, Herald
2.4 The exponential power growth of multiple players due to supportive non-supports
A lot of optimal single target "buffs" or "defensives" have an equally strong AoE part, for example, "Healing Turret", "Sand Cascade", "Leap of Resolve" just to name a few skills from various classes are all abilities you would use while playing their respective class solo, that makes them very difficult to tune correctly as they must be strong enough to use without any allies, but while doing 5x as much as "usual" they become overpowered. AoE healing and support builds are not at all in themselves a problem, I personally find that they are part of what makes WvW interesting, what is a problem is when every "offensive" build runs with several supportive skills by "accident", ie skills they would be using even if they only affected the caster.
My Solution: The lines between support builds and non-support builds should be clearer, for example a nerf targeted at "Healing Turret" might include removing the personal heal on it (reducing it's self healing by 50%) or making the second part of the healing self-exclusive, now I am not saying "Healing Turret" in particular does or doesn't needs nerfing, it is just a good example of that kind of skill.
Main Offender(s): Scourge, Firebrand, Guardian, Herald, Renegade, Scrapper
I originally while writing this did not include this point. Mainly because it did not cross my mind. But after reading several replies I realize that this deserves to make the list. This boon can sometimes be doing 60% of the damage you take (according to my damage meter's incoming damage)(or more in extreme examples). There is no counterplay beyond "don't attack".
My Solution: Redesign the buff completely, don't even keep the concept.
Main Offenders: Guards and Chronomancers are the ones that give the boon out the most, but the sources of it aren't the problem.
This boon is poor design. I assume it was intended as a way of countering conditions in a similar vein that protection counters power damage. It however does so far too well, if specced for it, you can have a ridiculous uptime on this boon, especially in a group, blanket immunity to one entire type of damage does not seem like good design.
My Solution: Keep the immunity to impairing conditions and make it reduce condition damage by 33-50%, adjust availability accordingly.
Main Offender(s): Firebrands, Mallyx Revenant
3.2 Access to condition cleanses
A large majority of the condi clears in this game are AoE, this means that the larger the enemy force is, the worse condition builds become compared to power builds. I have long been of the opinion that group condition clears are too strong, while for most classes, personal condition clears are too weak (this used to be a bigger problem and now doesn't affect as many classes).
My Solution: Nerf some of the more spammable AoE condi clears, while buffing personal condi clears.
Main Offender(s): (Too much Group): Firebrand, Scourge, Heal Scrapper,
Main Offender(s): (Too little personal). Necromancer (Mainly Power Variants), Revenant (Power Variants),
3.3 Excessive CC-spam
This is more evident when outnumbered, but can be seen in equal fights too. CC effects are generally needed for a game, however most games have some mechanic that prevents you from being CC-locked. Being CC-locked is a situation where you get caught in a loop of CC effects until you die, generally because your stunbreak is on cooldown, this is a bigger problem for some classes than others.
My Solution: Introduce a diminishing returns effect on CC, ie everytime you get CCed you either get a few seconds of immunity, or CC-effects applied to you afterwards have a reduced duration, how to make this work with the various lengths of CC effects is the hard part, but something along the lines of the longer your previous CC-effects were and the more of them was applied to you, the more powerful the diminishing returns. Adding more stunbreaks/sources of stability for problem classes is also an option, but imo a more shortsighted and more band-aidy one.
Main Offender(s): Spellbreakers, Holosmith, Chronomancer, Large Fights