Last.fm listening clocks

Spotted on Wired, the developers responsible for Last.fm’s ‘VIP playground’ have announced the launch of a visualisation service that plots subscribers’ listening habits over a certain time period onto a 24 hour clock graphic. This follows other last.fm graphic developments such as the listening trends streamgraph (read more about streamgraphs here) and music universe.

I have previously praised radial plots when writing about the graphical history of the FTSE produced by Jeremy Christopher. However, I have also found reason to identify failings with this approach, like the GE ‘Cost of Getting Sick‘ chart (read more on EagerEyes).

Once again, I don’t think radial plot approach works particularly well as an approach to effectively communicate the information. This shortcoming is probably best explained by the apparent motivation behind these graphics:

A bit less than a year ago we launched the VIP zone on our Playground, with the promise that we would keep adding fancy visualizations to it as a special treat for our loyal subscribers.

To interpret the visualisation, the red and green bars represent weekday and weekend listening, respectively, and the longer the clock’s hands the more the listening was focused around the time to which they point.

To be fair it is clear where the clusters of peaked bars are in relation to the 24 hour clock. However, the overlayed plotting of weekday and weekend activity is messy and detracts the user from being able to intuitively judge the respective bar heights. Accuracy of reading is not a necessity here, its more about the patterns hence the lack of scale, but the lack of subtle concentric gridlines makes it difficult to compare values, should you wish, at either side of the clock face. The clock hands are an unnecessary and confusing addition presumably representing a peak of listening activity that could have been derived from the relevant bar trends anyway.

Unquestionably, last.fm have created a visualisation that is ‘fancy’ but it would be far more effective to plot both series onto a small multiple pair of bar charts or area charts. I’m sure I’ll be labelled a killjoy but if you’re going to produce something like this to help users explore and unearth insights from their listening behaviours, a potentially interesting dataset, I think it is disappointing to undermining this opportunity through inefficient visualisation approaches.

4 Comments

Andy CotgreaveSeptember 7th, 2010 at 1:39 pm

I agree with you. This viz has several main problems:
1. A 24 hour radial scale used within a 12-hr clockface metaphor. This means that I can’t use my knowledge of how clocks works to quickly relate the position on the dial to the relevant hour of the day.
2.Confusing overlays – the bigger bar is always under the smaller bar. This avoids occlusion but is visually confusing
3. No y-axis – ie, what does the length of the bar/spoke mean?

I like the fact that they are trying these things out, but the art is trumping the information.

Andy KirkSeptember 7th, 2010 at 2:05 pm

Spot on points Andy – if you’re going to use a metaphor it has to be consistently applied and a 24 hour clock fails to achieve this.

Richard KurschnerSeptember 8th, 2010 at 2:43 pm

In any analysis of usability, you need to remember the target audience. You may not be comfortable with 24 hour clocks, but a lot of us are because of our “round the clock” jobs. (No, I’m not in the military.) And I’ll bet that most station program managers, who would be the target for this data, are very comfortable with it. I actually like the choice: I see a lot of 24 hour data displayed on a linear chart and I’m always having to check when the starting time is (Midnight, 3AM, etc.). This makes that step unnecessary.

Because the intent seems to be pattern display, the lack of a radial scale doesn’t bother me nearly as much as the limited range of the plotted data. Why don’t the longest bars make full use of the face plate? That would allow for better discrimination between the bars.

I do agree though that the overlay of weekend and weekday data is very confusing as is the gratuitous use of the clock hands, especially since one is green and the other red…

Andy KirkSeptember 8th, 2010 at 4:27 pm

Many thanks for your comment Richard, you offer some valid observations about the utility of the 24 hour clock face, particularly in terms of the audience. I guess this continued axis also offers a neat solution for listening patterns that start at the end of one day and continue through the early hours – admittedly this would be slightly lost on a linear axis.