notes.husk.org/likes images.

toffeemilkshake:

Nice visualisation of TV show ratings with trend lines. via Phil Gyford.

1. Would be nice if the vertical axis always covered the full 0-10 domain (This, along with the tooltip styling, is a give away to me that Highcharts was used — confirmed with a quick view source).

2. The consensus seems to be that Buffy S-6 isn’t much worse than all the others — it is.

Fascinating. It shows the skill of most modern shows at doing series that broadly rise in “quality” (defined by: the extent to which there’s a consensus on episode X being good), and the way in which this is something clearly aimed for.

(As for WHY it’s a good idea to back-load your series in terms of perceived quality, I’m not sure - a natural artefact of arc-based drama, I guess, with the build ups in the first half and the pay-offs at the back. But it also maps onto the peak-end rule quite nicely: make your peak the finale (or have a standout earlier episode of equal quality to the finale) and that’s how people will remember the show.)

Of course I went and put in Doctor Who - http://graphtv.kevinformatics.com/tt0436992 - which is also very interesting, in terms of the difference between the Davies and Moffat eras (Davies: 3 out of 4 strongly rising arcs*, Moffat: 1 out of 3, and that not rising particularly sharply.) I don’t think this proves anything about either man’s approach to the show, though - the two Moffat seasons with falling arcs are the two that got broken up, and I suspect the trendlines for each half-season would be rising. So it’s an argument that broken-up Who seasons are a terrible idea if you’re then going to collect the seasons and present them as a whole, perhaps.

*The exception is series 2, which obeys the peak-end rule with twin highs in “Girl In The Fireplace” and “Doomsday”, but has a flat trendline thanks to “Fear Her” and “Love And Monsters” - this is the kind of season which makes ‘trendlines’ break down a bit though as perceived quality zig-zags SO extremely.