Hacker News Reader

Comment by mattparlane | Hacker News Reader

Can't believe that no one has mentioned the chart yet -- a non-zero baseline is highly misleading.

http://tctechcrunch2011.files.wordpress.com/2013/06/browser_...

There is at least a slight chance that the non-zero baseline represents power consumption at the desktop with no browser running, so that the heights on the chart depict just the power draw attributable to the browser (it may also be a baseline relative to showing about:blank). 15W+ even for simple stuff like Wikipedia seems pretty high to have the base load already subtracted out, unless this is just a measurement of the peak power draw during rendering (in which case we need to normalize by time to get Joules instead of Watts).

EDIT: reading the PDF, this seems to be the case: they claim an at-desktop baseline of 14.7W average for the six notebooks they tested. Actual baselines for the notebooks tested varied from 9.76W to 22.65W, and the most strenuous test adds 3.11W-6.92W depending on the machine (and oddly enough, it wasn't always YouTube that had the highest power draw - on one machine, it was Bing, if the data is to be believed).

The misleading Y-axes on two of the charts on the IE blog post were the first thing I noticed:

http://blogs.windows.com/ie/b/ie/archive/2013/06/05/internet...

Here's the top chart (left) and the same chart except with a less misleading Y-axis (right):

http://i.imgur.com/zI1UJeS.png

How is that misleading Y-axes? The intervals are same, only thing that will be different is the resolution, and going into that fine resolution amplifies the difference. I will not say it as misleading in a slightest bit.

Because it amplifies our perception of the difference, which misleads us to believe that the difference is more than it actually is.

The Y-axis should've started at 0 watts, not at (a rather arbitrary) 18 watts.

Look at the "Flash Video" and "HTML5 Video" bars in the original chart. The blue IE bars are roughly half the size of the green Chrome bars, which might lead you to believe that IE only consumes roughly half of the energy that Chrome consumes in those tests. But as the numbers show, the difference isn't nearly that great.

Starting the Y-axis at 18 watts only serves to make IE's advantage over the other browsers look greater than it actually is.

The y-axis should have kept the baselines that the study authors used in their graphs (14.7W for laptops, 37.8W for desktops), since those starting points are what gives you the power draw due to the browser. What the Microsoft blog did by moving the baselines up from there was unethical, but having the chart show power consumption relative to idle instead of powered down is perfectly reasonable.

Hah! Great find. When I took a technical communication course, we explicitly talked about the ethical implications of misleading the audience by manipulating information and the perception of information.

Needless to say, the authors did a poor job of following standard ethical practices when presenting information to support their claim.

Not to mention that the sites have no relationship to each other, so a line graph like this is pointless.

Here's the graph corrected for both zero-baseline and using a column graph: http://i.imgur.com/24UrJpB.png

Reply on news.ycombinator.com