Heroes season 3 has really been all over the place. It got off to a good start, covering more ground in a couple of episodes than you’d sometimes get in half a season. But although they quickened the pace, this just meant it got sillier quicker, getting bogged down in formulas and double-crosses and undeveloped ideas. But I’ve kept watching because it has enough moments to remind you how good it can be, and when it’s bad, it’s entertaingly bad.

But what would I do to improve the show? One of my favourite Heroes pisodes is Company Man from series 1, in which Noah Bennet attempts to protect his family as his involvement with the sinister Company comes unravelling all around him. One reason it was such a good episode was that it focused on one set of characters, and told a complete story (though one that formed part of the continuing narrative), rather than half-a-dozen bits of stories strung together. I think Heroes would work much better if the writers focused more episodes on just one or two sets of characters rather than jumping between them.

Let’s take season 2 as an example, and look at how it could be restructured. It’s a good idea to reintroduce all the characters in the first episode, so having lots of characters in episode one isn’t a problem. But over the following episodes, I’d focus on one set of characters at a time, before beginning to weave the threads together, like this:
Episode 2 – Hiro in Japan
3 – Matt and Nathan’s investigation
4 – Maya, Alejandro and Sylar’s journey
5 – Nikki, Micah and Monica
6 – Peter and Caitlin in Ireland
7 – Claire and West/HRG and Suresh
From then on, I’d then bring the different storylines together, moving back to the usual style of Heroes storytelling as the volume finale approached and everything became much more interlinked.

The advantage of individual episodes is that the characters’ storyline has to be stronger. For weeks on end, we had Maya and Alejandro trying to get to America, getting in a spot of trouble because of Maya’s power, and getting out of trouble again – rinse and repeat, and in later episodes add Sylar glowering menacingly.

You can almost get away with this kind of repetition when it’s spread out in weekly chunks across different episodes. But if you edited it together into an episode, it would be deadly dull. You’d have to actually develop the relationship between the two twins, develop more depth to the story than capture/escape/capture. You could spend less time recapping, and more time on the characters.

The other thing I’d do if I was writing Heroes would be to destroy the Company, or at least push it away into the background. The reason I’d do this would be to completely get away from secret organisation politics and to refocus on ordinary people learning to live with superpowers. The trouble with the Company and Pinehurst is that it is completely removed from any normal situation most of us face, and it has no inherent emotional resonance. Those kind of secret organisations can work when you see their effect on ordinary people, on their lives and families (as Company Man does brilliantly). But shadowy organisations work best when you see them from the outside.

The other thing to do is to give characters consistent motivations and actually make them develop in a believable and interesting way, rather than lurching from one random change of allegiance to another for Reasons of Plot. Changing allegiance every five minutes does not create moral ambiguity. Moral ambiguity requires that we can understand and sympathise with a character’s motivations, while those motivations lead them to both good actions and bad.

I recently saw that season one writer Bryan Fuller is returning to the show now that Pushing Daisies has been cancelled. He wrote Company Man and in this interview that I just found he seems to have a sensible idea of the direction the show needs to go in. But the Heroes writers have cried wolf too many times with their claim that “It’s better this time, honest!” for me to be convinced the show will improve – only time will tell, not Tim Kring.