You may recall that when "The Killing" finale aired, I didn't mince words in my review, saying that "This will be the last review I write of 'The Killing,' because this will be the last time I watch 'The Killing.'"

The problem with making such a definitive statement — particularly in the flush of the frustration that I felt over how badly that show spun out of control — is that when time passes, and the immediate emotion goes away, what seemed definite at the time becomes a bit murkier. So a few months ago, I decided I was at least going to look at the season 2 premiere, not expecting much but at least allowing for the possibility that Veena Sud and company learned something from the reaction to the first season — not just to the cliffhanger ending, but to the thin characterization, over-reliance on red herrings, and other things that I and many of you complained about for many episodes before we got to "Orpheus Descending." (As I've said many times, it's a false narrative to say people were only annoyed because we didn't find out who killed Rosie; we were annoyed for lots of reasons, and were looking to a resolution to the mystery as some kind of compensation for the time spent: Okay, so they've screwed up this, this and this, but I'll feel like less mad at myself for wasting so many hours if I at least find out who killed her.)

As I told New York Times Magazine writer Adam Sternbergh, who wrote a story about the role of the "superviewer" for a show like this, sometimes shows that had problems learn from their mistakes and get better. "Parks and Recreation" is the most obvious recent example, but it happens enough that I try not to completely write off shows if time allows. So I'll watch the premiere when it comes, and if it's better, I'll say that. And if it's not, I'll say that, too and move on.

But I bring all this up because Sternbergh and I actually talked for quite a while for the story, and as happens with virtually any interview (particularly for a multi-sourced piece designed to run in a print publication with a set space), only a fraction of what we discussed made the final version. And there are a couple of points that I think are worth discussing, not only about "The Killing," but how social media — blogs like mine, Twitter, message boards, etc. — impact many shows.

(A word of warning: I'll be discussing significant plot developments of several non-"Killing" shows up ahead. I'll put the names of the shows in bold as a warning for those who are behind.)

First of all, there's the problem of doing a whodunnit, or any kind of mystery-based series, in the age of Twitter. When you're telling a mystery to millions of viewers, odds are at least a few of them are going to figure out what the answer is ahead of time. That's not always the case — I'm not sure any "Lost" fan would have been able to predict the pool of golden light back in season 1 — but a handful of people at minimum almost certainly will. Thirty years ago, that's not a big deal at all, because at most the guy who figured out who shot J.R. would only be able to share the prediction with friends and family, or possibly to the readers of their "Dallas" newsletter. Even 20 years ago, anyone who might have figured out who killed Laura Palmer might have been able to share the theory with the relative handful of people on Usenet or primitive online bulletin boards.

But now everyone's online, and this stuff spreads like wildfire. When "Mad Men" introduced Don Draper to a rich party guest referring to himself as "Connie," a reader of my old blog passed along a guess on the Television Without Pity message boards that the man was, in fact, Conrad Hilton. When Connie returned several episodes later and stunned Don with his true identity, anyone who had read that theory online wasn't quite so surprised. In the week between the final two episodes of the most recent season of "Breaking Bad," someone (I wish I could find the link right now) published a detailed blog post predicting the exact method of Brock's poisoning, how and why Walt would have done it, etc. And that got picked up in several places (including the comments on my review of the penultimate episode), and anyone who read it wasn't as surprised.

Or take what happened to the poor "Dexter" writers this season. They introduced Colin Hanks and Edward James Olmos as what appeared to be the two big bads of the season, with Hanks apprenticing under a more veteran serial killer. Only a few people started guessing on Twitter within an episode or two that Olmos was actually playing a figment of Hanks' imagination, appearing to him the way Dexter's father Harry appears to him. I'll confess that I hadn't guessed it at the time I read one of those tweets, but the instant I did, it seemed obvious that this was the case. And that theory kept being retweeted, discussed on blogs and message boards and elsewhere, and soon pretty much any "Dexter" fan who spent any time at all discussing the show online had to be aware of this theory — which became a huge problem given that the season was structured around the audience being taken completely by surprise when the truth was revealed three-quarters of the way into things. What was supposed to make everyone's jaws dropped instead inspired a collective shrug.

Obviously, not everyone is online obsessing over this stuff. But enough people are that these things disseminate even to the non-superviewers, who may not spend a lot of time on Twitter or blogs like this, but now have friends or relatives who do. So I think it becomes harder and harder for any show that's either based around a mystery, like "The Killing" or "Awake" (where I'm sure someone online has already guessed exactly what the explanation for Detective Britten's situation is) or a show that introduces a smaller mystery that carries over for more than one episode, to actually succeed in impressing their audience with the solution. By the time Detective Linden finds out who did, in fact, kill Rosie Larsen, there's no way it won't be someone who's been fingered by plenty of superviewers, some of them actually explaining exactly how and why they did it.

Shows that do a good job with the non-mystery portions of their storytelling will be okay — even people who read the "Breaking Bad" guess/spoiler could still appreciate what the show did with Walt and Jesse in that final episode — but I think shows that are primarily plot-driven may be in some bit of trouble. ("The Killing" is, in theory, supposed to be a character-driven mystery; I just think the show fell down on the job badly in making most of its characters interesting.)

The other social media issue we discussed was what responsibility, if any, showrunners have to keep track of online feedback to their shows. And even though I didn't like most of "The Killing" season 1 and think Veena Sud came off in many interviews (including the one with me) as oblivious to both the show's problems and the audience's reaction to it, I genuinely don't think it's a necessity — nor, at times, even a good idea — for creators to spend time reading what online viewers have to say about their work.

Sometimes, that can work out well, admittedly. Dan Harmon mentioned again at PaleyFest that so much of what they've done on "Community" has been inspired by the "feedback loop" the show's writers have with their audience. The way they recalibrated Britta once they realized viewers didn't see the character the way the writers did was a particularly fine example of how that two-way relationship can be valuable.

But I also know plenty of showrunners who get too hung up on reading online reaction — and, worse, who then go against their own storytelling instincts to try to appease the fans and wind up making things worse. (I've heard the phrase "I made a New Year's resolution to stop reading all the blogs and message boards" more than once.) I don't believe in creativity by committee. Many of the best shows in the history of this medium were made in a vacuum, either in a time long before there was easy online feedback, or were made by creators who aren't very interested in what others have to say about their work. "The Sopranos" was the exact show that David Chase wanted to make, and he had no compunctions about doing things that frustrated his audience. David Simon reads blogs sometimes, but he never compromises what he does based on whether viewers like some characters more than others. Even if Matt Weiner has a secret Twitter account to keep track of what people are saying about each "Mad Men" episode, each season is written and largely produced before we can react to any of it. Etc.

Feedback can be valuable at times, sure, but more often than not I think shows are better off with a strong creative voice that doesn't put too much trust in what a loud but probably statistically insignificant segment of the audience(*) is saying. The problem is that, based on "The Killing" season 1, Veena Sud doesn't appear to be as talented as those other producers I mentioned, even though she's been given similar autonomy. And if that's the case, I don't know how much listening to us yell and complain is going to improve the show, anyway. Either the characterization deepens, or it doesn't, you know?

(*) The ratings for "The Killing" season 2 should be an interesting test about how representative the online audience is. Numbers for season 1 were pretty consistent throughout, so either the bulk of the viewers were pleased with what they were seeing, or else they were, like me and some others, annoyed but sticking around just to learn whodunnit. If the ratings remain consistent for the second season (or, as is often the pattern with cable shows, go up), then it doesn't matter how loudly we yell, because the silent majority has made it a success for AMC. If the numbers are notably down, then, yes, everyone horribly miscalculated with that finale.  

I'm going to watch "The Killing" season 2 premiere in hopes that the show is better. If not, I walk away. Making any show great is hard, but I do think that social media has made it even harder to make certain types of shows.

Alan Sepinwall may be reached at sepinwall@hitfix.com