When it comes to narrative forms of art (film, literature, comics), there’s often a great disconnect between works that are highly influential (and thus become ‘classics’) and works that are well-executed. I don’t think this is an accident.
Well-executed works (hereafter, “good”) don’t particularly need to be novel. Whether they are high art or low art doesn’t matter — someone working in a “low” genre (say, harem anime) can be skilled enough to produce a truly shining example of that genre. Like good design, good art disappears: it is a complete and flawless implementation of expectations, and only upon close examination does the skill and effort involved in executing the expected attributes of the genre become visible.
As an example of unambiguously “low” art being extremely well executed, consider Monster Musume — a harem show involving young women who are mythological creatures. It seeks to sate essentially puerile desires: the goal is to have sexually-charged slapstick humor involving attractive women who are part animal. It is not the first show to do this by a long shot, but it may be the best: nearly all elements outside those that add to that goal are invisible, and it reaches its goal admirably, but at the same time it has a good and well-planned justification for nearly everything that happens (by combining the idea of diplomatic relations, corrupt/lazy officials, and a largely hostile population, you get an interesting take on the fantasy harem genre that can be seen as a stealth satire of racial politics and international relations), and it furthermore satirizes its own genre admirably (the protagonist’s name is never spoken in the show and rarely in the source material, and his facial features are often left blank; in other words, he is an exaggerated form of the self-insert character).
While “good” art may be popular in the short term, it is rarely influential: mostly, it subsumes itself in its own influences. Iconic works are by definition influential; they often come from outsiders and break best practices. While iconic works are memorable, it is rare that an iconic work is superior to those works that copy it. Consider Dracula and Frankenstein — both extremely iconic works, both widely adapted, and both (in their original form) nearly laughably incompetently constructed by the standards of both our time and theirs. It is not the high standards of craftsmanship that make Dracula and Frankenstein iconic; it’s not their popularity, either, although both were popular. Instead, each brought into being a particular collection of ideas that fit into a missing slot in the culture; this undigested piece of mental matter was sized upon and as these works were adapted or influenced other works the core interesting idea was progressively isolated from its trappings, until the point at which it becomes fully assimilated and no longer numinous. Dracula is no longer numinous to us: we have a mental image of Bela Lugosi in a cape, and we forget that he was supposed to have hairy palms, and we forget about Doctor Von Hellsing being a blood expert, and we forget about Lucy being obsessed with wax cylinder audio recording; while all of those elements are more interesting to us now, the iconic elements of that story are sufficiently captured by Bela Lugosi in a cape, to the point that this view of vampires as nobility (which did not originate in Dracula but really had its purest representation in the 1933 Universal Studios adaptation of Dracula that has become the most iconic one) has become dominant. The idea of permanently young and beautiful blood-sucking aristocrats afraid of the sun is one that resonated with the early twentieth century American culture very strongly, even as Max Schreck’s portrayal of Orlock in Nosferatu is a more accurate representation of how Dracula was portrayed in the book. Likewise: Frankenstein’s Monster originally looked far more human, and was highly intelligent (speaking several languages); we have made an icon out of an ugly Frankenstein’s Monster incapable of speech or complex thought and a Victor Frankenstein with a god complex and boundless ambition, rather than a beautiful but slightly unnerving monster and a Dr Frankenstein who wouldn’t be out of place at a My Chemical Romance concert because the former was a closer fit for exactly what was (and no longer is) unnerving about the story. Iconic works allow us to identify the unheimlich and integrate it into ourselves and our society in a disarmed form; or, to be more cynical: the Spectacle uses iconic works as an early warning system telling it what to consume next. Of course, as highly iconic figures become fully integrated, they cease to have the impact they otherwise would: recent Godzilla and King Kong movies flopped for the same reason that new viewers wouldn’t watch the originals, which is to say that these monsters have been integrated and are no longer monstrous.
Iconic works don’t need to be bad in order to be iconic, but even if they display technical excellence, they will face initial rejection. Consider Neon Genesis Evangelion: certainly iconic, and hardly poorly made, this show garnered very little interest during its initial run. Part of the reason is that it aired on an incorrect slot: this show, with its dense references to media from the 70s and its complex character dynamics and psychosexual undertones, was airing in a time slot that normally was geared toward ten year old boys. But, it’s more than that: Evangelion remains highly divisive, and remains relevant more than twenty years after it first aired, because its iconic elements have not yet been fully assimilated. Evangelion didn’t introduce the unwilling soldier (in fact, this element is part of why Gundam was iconic); it didn’t introduce the idea of a complex and incestuous conspiracy between a private high-tech defense organization and various government and religious authorities. But, Evangelion took upon itself the task of examining attributes of the mech genre realistically, and did so by taking a bunch of characters who border on archetypal and spending a great deal of effort trying to make their behavior and characterization realistic. Evangelion is still relevant because we haven’t figured out what makes it relevant yet: every post-Evangelion mech show is in some way a response to Evangelion in the same way that every post-Dracula vampire novel was a response to Dracula, yet even as some individual creators have done several generations of responses, none are a sufficient substitute for the original. For instance, Yoji Enokido, after working on Evangelion, went on to work on Utena (a non-mech show with certain very visible Evangelion references), Rah Xephon (a mech show that has been seen as an Evangelion clone), Star Driver (a mech show very similar to Rah Xephon with a more typical mecha protagonist), and Captain Earth (another mech show with a more typical mecha protagonist, which has some scenes directly lifted from Rah Xephon); while his vision has diverged from Evangelion proper, he’s still chewing a piece of the same cud.