IMDb ratings in general follow a four point scale (http://tvtropes.org/pmwiki/pmwiki.php/Main/FourPointScale) and are also skewed toward the internet-savvy demographic. It's a good example of a homogeneous population and the problems that can cause for data analysis.
I really like Firefly, but the cancellation meant they didn't have the opportunity to fuck it up. Every other Whedon series has had poor seasons.
I can't really speak to Buffy and Angel, but with regards to Firefly and Dollhouse... The indication is that Whedon wouldn't so much fuck it up, but requirements by Fox would force him to fuck it up. Compare "Train Job" to the original first episode of Firefly. Compare early episodes of Dollhouse to later ones, particularly the encore episode.
The idea of using the Big Red Reset Button as a continuous plot element, and telling the arc story through seepage across resets was a brilliant way of giving the finger to almost every other series.
And it turned into a young adults' barbie show.
Although it's better today, with Netflix and the free reign a lot of producers get, I think there is an art of show business maneuvering that shaped so much TV and kind of added to it's character. The classic "medium with a limitation" theory.
Personally I think shows should get three seasons by default, and should need a _really_ compelling reason to go beyond that.
I wish there were more shows that just were OK with planning an ending. I like stories with endings. Just know your limits, don't try to make a show that can go on for a million seasons, end it when its still good, and then use the goodwill garnered from making a masterpiece to build the next creative thing.
I'm not sure why this model seems to be more successful in, say, Anime as opposed to western TV. Budget, maybe?
It's mostly a US thing. The UK is fine with making shows that have short seasons and that only last two or three seasons.
Furthermore I'm not convinced UK's shorter season format is a good thing. I actually think it's a symptom of much smaller budgets.
Walking Dead -- eh, it's getting a little old. I thought they were going to explore some primitive economics, and maybe they will, but the cliffhanger gag last season reminded me too much of "Who Shot JR?" Not sure I care. And the spinoff went too heavy on the social commentary, couldn't even finish Season Two.
Mr. Robot is more promising, but I haven't seen one yet that comes close to the satisfaction I got from the clear arc of the story in Breaking Bad. Sons of Anarchy tried, but fell short somehow.
"Oh, we're walking or driving through the woods" "Cool, we found a place that looks cool and has cool people" "Oh, the cool people were actually horrible for some reason" "Time to blow it up I guess."
Budget and the revenue from a hit.
Making a hit is such a crap shoot that depends on a lot of things including what basically boils down t luck, i.e. it grabs the limelight for whatever reason. So when something does become a hit, there's a lot of pressure to just keep things going.
- The network likes that they have a predictable money-making show.
- A lot of fans will continue watching out of habit/duty.
- Many actors and others associated with a hit show may think (often for good reason) that they'll likely never see as good a payday again.
It's also worth noting that shows like Breaking Bad and Game of Thrones still had relatively niche audiences compared to a hit on a major network. Probably makes it easier to plan for a fixed length.
You do see some of the same effect in, say, genre fiction though. There are a lot of interminable series because they have a built-in readership.
Advertiser preferences are far more relevant. One could argue that Advertisers usually want to advertise in shows that people like to watch, but there are nuances to that and IMDB ratings aren't broken down by demo, so we can't even know in hindsight.
The reality is that Fox's decision was likely based on being able to charge more for ads on some new show, which makes it correct both at the time and in hindsight.
Firefly had decent ratings when it first aired, so a lot of people were really confused that they canceled it. Those people apparently failed to understand television as a business, as in addition to those decent ratings, it was also expensive to produce, far more expensive than was justified by the decent viewership it was getting (there's a lot of shows now spending millions of dollars per episode on big name stars, but they have a lot more viewers than Firefly was getting...).
Much more interesting (and true to the title used here) would be finding the regressions between subsequent cult-classic status and data points that either were, or might've been, or would be (had it been produced today) available at the time.
(I'd expect that Netflix is correlating vastly more data points about audience behaviour than we know about. Short- and long-term responses to plot elements & twists, pacing strategies, classifications of dialogue and modes of storytelling, presence of archetypes &c &c, should all be on their agenda.)
Such an article would also have to look at the costs to make each show, which it does not. (I expect Firefly to be more expensive than Veronica Mars due to things like special effects, and Joey expensive due to Matt LeBlanc.)
And it would have to consider monetization models for each, where shows with ads have to concern themselves more about advertiser friendly demographics vs. premium pay shows have to find a demographic that is willing to pay directly for TV.