Jumping the shark: the data

‘Jumping the shark’ refers to the point at which a long-running TV programme stops being good. It’s defined on Wikipedia as:

“the moment in the evolution of a television show when it begins a decline in quality, which is usually a particular scene, episode, or aspect of a show in which the writers use some type of “gimmick” in an attempt to keep viewers’ interest.”

The website TV Tropes explains it further:

“The moment when an established TV show changes in a significant manner in an attempt to stay fresh. Ironically, that moment makes the viewers realize that the show’s finally run out of ideas. It’s reached its peak, it’ll never be the same again, and from now on it’s all downhill.”

The term is named after the scene from Happy Days, in the episode ‘Hollywood: Part 3’, in which Fonzie, on water-skis, literally jumps over a shark. However, it’s now used more generally, not just for the introduction of gimmicks which signal the drying up of ideas, but to other changes which signficantly detract from a programme’s quality: executive meddling, or the departure of key cast or writers, for example.

The Graph TV tool, created and published by Kevin Wu, lets us examine jumping the shark moments properly. The tool lets you enter the name of any TV programme, and automatically plots the individual episode ratings from IMDb, sorted by seasons and with trend lines. Now, for any series, we can easily see what the consensus of opinion is on whether and when it jumped the shark.



First, let’s look at the original: Happy Days.

Happy Days, plotted on Graph TV. The “jumping the shark” episode (S03E03) is indicated by the slightly larger circle to the bottom right of the panel.

From this, it does look like there was a slight decline in quality, with seasons 5 to 11 rated slightly lower than 1 to 4. The shark-jump moment itself (season 5, episode 3) occurred at the tipping point between those two phases.

(More specifically, seasons 1 to 3 were generally scored above 7, averaging around 7.4, until a decline to around 7.0 at the end of season 3 and into season 4.

Season 5 averages about 6.8-6.9, has some episodes in the low 6s but others in the 7s, and one is among the best rated of all time. Hollywood: Part 3 gets one of the lower scores, at 6.3.

Seasons thereafter remain around or just below 7.0.)

So, there was a slight decline in quality just before the relevant episode, that season was patchy and the episode was symptomatic of the worst ones. On the other hand, the series didn’t really worsen at all from that point; if anything, it improved slightly and held together consistently to the end.

So, while it’s understandable that Hollywood: Part 3 would be remembered as a particularly notable example of Happy Days’ already-completed fall from its earlier heyday, it’s also a bit of a myth that it marks the start of a major drop in quality, or that Happy Days was significantly worse after it than before.

By toggling Graph TV to show the full y-axis from 0 to 10, it’s clear that Happy Days was, like Friends or Cheers, of a pretty consistent quality throughout its run. Though it’s great that it contributed such a memorable phrase to the cultural lexicon, there are in fact a number of programmes which had far more dramatic popularity crashes / “jumping the shark” points.



I love Babylon 5: it’s one of my all-time favourite programmes. But I have to admit there was a bit of a shark-jump moment with season 5. That moment, though, was associated with executive meddling, not with a creative failure.

It’s interesting to compare it here, because it has an interesting progression of ratings, and it shows an even more sudden and significant drop in popularity than Happy Days – a full 1.0 in average rating between seasons 4 and 5 – followed by a final season recovery from its shaky beginnings.

Babylon 5’s popularity arc starts at an average 6.9 at the beginning of the first season, which has episodes varying widely in quality from 5.4 to 8.7. It steadily rises, through a much tighter second season, to reach its peak in seasons 3 and 4, where it averaged above 8.0.

This is a good example of a related phenomenon, the opposite of jumping the shark: growing the beard. Named for the noticeable improvement in quality in the second season of Star Trek: The Next Generation, which coincided with the appearance of a trimmed goatee on the face of Jonathan Frakes, who played Commander William Riker, it refers to the way programmes tend to get better over the course of their first few seasons: as they find their feet, establish popular characters, start to develop story arcs, and so on.

Babylon 5 always had a planned 5-season story arc from the beginning. The first season started slowly, as intended, with a lot of character introductions, and not much seemingly going on. However, the first pieces of the puzzle were being laid, with plot elements, questions and clues being included which were to have big pay-offs later on. It was in the second season that the big story arc took a more obvious shape, and in the third and fourth that it reached its climax and conclusion, and you can see that progression in the Graph TV results.

So what happened in the fifth season? It was a catastrophic case of Screwed by the Network: its writer, J. Michael Straczynski, had always intended his story arc to last for five seasons. However, he also had a backup plan to compress the last two seasons into one, in case the fifth season was cancelled by the network. But he wasn’t quite prepared for what did happen: the fifth season was cancelled, and the story arc was compressed into the (brilliant) fourth season as planned. Then, after the whole show’s story arc was wrapped up and Babylon 5 was essentially ‘done’, the network changed its mind and commissioned a fifth season after all. JMS had to rapidly come up with new plot lines and conjure a new fifth season out of mid-air, following on from the natural narrative conclusion at the end of season 4.

Hence the sudden drop in user ratings from their peak in season 4 (the final showdown), to the start of season 5 (major anti-climax). However, JMS being the brilliant writer he is, he managed to create something worthwhile out of those disparate, initially disappointing plot strands, and by the end of season 5, he’d regained Babylon 5’s mojo.



Heroes is a better example of a programme doing well for a while, then running out of creative steam and taking a nose-dive.

The first season of Heroes was a planned, tightly-plotted story arc. Every character, with each of their nascent superpowers, was written with a specific purpose: to fit into the unfolding plot, providing a complex set of problems, then develop and climax together and logically resolve each other in a clever, satisfying conclusion. It wasn’t the greatest TV programme ever, but it was good piece of multi-stranded storytelling.

However, the problem began as the end credits rolled on the first season finale: as a success, it was already commissioned for another season, which would inevitably lose the focus of the first season, and end up a mess.

The writers had to corral the surviving characters (and bring back ones who died), throw in a few more, then contrive new situations for them to deal with. It didn’t have the same sense of character development: in the first season, the heroes were just discovering their powers, but in later ones, they already had them. The plot elements, the challenges they had to face, felt artificial and imposed, rather than emerging logically as a result of the characters themselves.

And the shark-jumping moment itself is easy to spot: at the very end of the first season’s final episode, the teaser scene for the next season showed one of the characters suddenly displaced into feudal Japan. The feeling of, ‘Oh god, that’s ridiculous, this show has just gone badly downhill,’ which it induced, must have been very similar to the feeling people had when they first saw Fonzie performing his water-ski stunt.



Red Dwarf follows the classic progression too: it started well, but really grew the beard in the second and third seasons with the arrival of Kryten, and the development of its distinctive character-driven humour. Then, in season 7, it badly jumped the shark.

Red Dwarf was written by ‘Grant Naylor‘, a single pseudonym for two individuals: Rob Grant and Doug Naylor. After the sixth season (the show’s peak according to the graph), Grant left the partnership to pursue other projects, leaving Naylor to write the seventh and eight seasons alone.

Grant Naylor’s division of responsibility was revealed: Naylor wrote the plots and Grant wrote the jokes. Naylor’s solo seasons were awful: too much time was spent developing non-humorous plot material and character storylines, while jokes were almost entirely absent. Naylor tried deriving humour purely from zany plot devices alone, such as an extended sequence where a miniaturised spaceship wedges itself into a rat’s anus and then flies around with it impaled on its bow.

Many fans still enjoyed it, but personally, I found it a poor comparison to the earlier scripts full of witty insult-trading and Blackadder-like verbal high jinks, and Graph TV seems to agree with me.

Coincidentally, I recently had a conversation with someone who tried to convince me that the tenth series was much better, a return to the show’s original form. That also seems to be borne out by the ratings.

Finally, two examples I found which, according to Graph TV, drastically jumped the shark, but which I know nothing about.



Dexter had high scores across most of its run, averaging around 9, and even the marginally weaker sixth season was still above 8. However, clearly something went badly wrong with the last season. I’m fascinated to know what could cause such a dramatic decline. Are there any Dexter fans out there who can enlighten me?



Again, I haven’t watched Dragon Ball Z, so I don’t know what happened here. But the user ratings – which are probably skewed somewhat by the ’10 or 0′ rating tendency of its younger viewers – are fascinating. Seasons 1 to 3 are rated unanimously awesome. Season 4, however, is utterly awful (except for some strange inconsistencies, like a few at the beginning, and one towards the end). Then it returns to a more varied, but generally better rated, pattern, for the remaining seasons.

Any thoughts?

4 thoughts on “Jumping the shark: the data

  1. Are those three pink dots on Red Dwarf the Back to Earth specials? My heart broke watching them. There wasn’t a single snigger to be wrung from me.

    I’m also shocked Lost didn’t do this at some point, but maybe that’s just because I am an outlier. I reached a fuck it point with that show and never watched another second once Locke’s dad appeared in a box, but it seems it was just me.

    • The three-episode season 9 which looks like a crudely drawn pink cock is indeed Back to Earth. I’d already given up before then and so my heart was spared. Season 10 looks promising though.

      Lost does show a slightly lower average for the final season and the last episode is among the lower rated ones, although you’re right, it’s surprising how well its ratings hold up despite the Chris Carter Effect kicking in at various points for many viewers as the series went on, and the final season’s above-8 average is astonishing given the vehement hatred for it which split the fandom in two. I guess as time goes on, more people come around to the “plot arc was unimportant, it was about the writing and characters” way of thinking.

  2. Do you think the results are slightly skewed by not taking into account how many people watched? i.e. I suspect lots of people will the first few seasons of a show, but once they spot a decline in quality (particularly if early in a new season, before they’re back in the habit of watching it weekly) they’ll drop out, and thus are less likely to rate following episodes, having not watched them to do so.

    • I’m sure the data is skewed in all sorts of ways. That’s one of them. It might mean some later seasons of shows which jumped the shark were even worse than indicated – as those submitting ratings were the ones who still liked it enough to keep watching. So possibly the effect is less obvious than it should be. It might also explain why later seasons tend to become more widely spread – as numbers of votes decrease, the data becomes less reliable.

      You can easily check the numbers of votes by hovering over each point on the Graph TV results. Look at Heroes – each episode gets 1000+ ratings in season 1, then episode 1 of season 2 gets 1400+ (great expectations) then numbers immediately drop below 1000 for the remainder (disappointed viewers ditching it en masse).

Leave a Reply

Your email address will not be published. Required fields are marked *