Almost without exception, whenever I have my laptop running on battery power, I have the Power Statistics window (gnome-power-statistics) open on one of the workspaces, showing me the battery charge history. It’s a great little chart that has magical powers to induce speedier working as I extrapolate that there’s only ten minutes of power left. But, with just a few small changes, I think it could be even better.
The main problem I have with it is the way the x-axis—showing time elapsed—is always divided into 10 sections. This is fine when it represents the last ten minutes, but it’s rather less appropriate when charting the last two hours, day, or week. Having a sequence of markers that reads:―
0, 16h48m, 1d09h, 2d02h, 2d19h, 3d12h, 4d04h, 4d21h, 5d14h, 6d07h, 7d
is just crazy. Dividing instead into 14 sections would give a more sensible 12-hour spacing. Similarly, the chart for the last day could be divided into 12 intervals of 2 hours. The chart for the last 2 hours also has the option of division by twelve (with 10-minute gaps) but I think 8 intervals, each a quarter-hour long, would be the common-sense choice.
That small change of altering the scale to more familiar units—quarter-hours, two hours, and half-days—would make a big difference. There is, however, another change that I believe could make when the peaks and troughs refer to instantly recognisable. I don’t tell people what I did five days ago, or 13 hours ago; instead I say what happened last Thursday, or after breakfast. For the most part, I define my history in absolute terms, not relative to an ever-changing now. This style could easily be applied to the charge history graph.
And so I present the following mock-ups, along with images of the current graphs. (They’re mostly mock-ups too, technically, as the battery hasn’t been used in the last week, so I had to create plots that would represent ordinary use.) In each mock-up, the right-hand edge no longer applies to now, but to the end of the current time interval. This allows each marker to be a sensible, easily comprehended label, such as
Wednesday, without having to move along the axis little-by-little over time.
The division into familiar blocks of time also allows functional colouring to be applied to the background, further highlighting the times data points represent. Rather than be a simple, statically striped background, the stripes mark particular blocks of time, wherever those blocks appear along the axis. In the 1-week graph, a stripe represents a day. For the chart over the last day, each represents two hours. Perhaps there is an argument to be made for 10-minute stripes in the 2-hour chart, or for changing the 10-minute chart to a 15-minute one, so that each stripe represents the full scale of the next chart, but I’ve only just thought of that so the 2-hour mock-ups below have 1-hour stripes.
All the mock-ups also feature sparser labels on the x-axis, appearing only for every other interval. I believe this is also of benefit, by limiting the potential for information overload. And with more familiar blocks of time as the intervals, the gaps are readily intuited, meaning that the kind of mental arithmetic currently needed for determining how long it took the battery to discharge would not be as necessary.
Finally, a third checkbox—Show absolute time—has been added which could be unticked by those folks who prefer the current, relative-to-now style.
Figure 2: The mock-up as it would appear on the same Tuesday afternoon as in Figure 1. Note that the plot ends at the current time—approximately 4 o’clock—which does not coincide with the right-hand edge.
Figure 3: The mock-up for the following morning, around 11 o’clock. The stripes have remained with the same blocks of time (whole days) they coloured in Figure 2, rather than forming a static background.
Figure 4: The chart for the last day, as it currently appears (at 1.15 p.m.). Determining when the battery started discharging means calculating the average of 16h48m and 19h12m—18h—and then taking that away from the current time.
Figure 5: A mock-up for the charge history over the twenty-four hours leading up to 2 o’clock. That the battery started discharging shortly after 7 o’clock the previous evening is made more obvious. The labelling scheme should follow the style used in the panel clock so users who dislike 24-hour labels can see 12-hour ones instead.
Opinions? Suggestions? News that future developments are already heading in a completely different direction? Leave me a comment below.