Comment by so33
7 years ago
The NYT example is interesting.
During the US presidential election I interpreted the needle’s fluctuations (on what was a simulation of an analog gauge–a skeuomorphism) to indicate some form of uncertainty in the data. This would not be deceptive (at least according to my own mental model of how such a gauge would operate in real life).
However, according to this article many interpreted the movement as the needle reflecting actual changes in data. In this interpretation (the gauge reading reflects at all times the actual data) this would be deceptive.
I realized that I haven’t seen one of those gauges in real life for a very long time. Have people in general (or, inversely, the NYT) forgotten how analog gauges are supposed to work?
> according to this article many interpreted the movement as the needle reflecting actual changes in data.
I said more in another comment elsewhere in the thread, but I think outside of stem fields, things often need to be precise and black-or-white... the idea of quantifying the degree of uncertainty ("error bars") is foreign. That's one of the ideas I would try to give the rest of the world.
I've never seen an analog gauge move like this. The pressure gauge on our heating doesn't do that, the speedometer on my car doesn't tremble, and neither do analog Volt meters as far as I recall.
The speedometer on your car is almost certainly a digital system driving a needle. Unless you have a 40 year old car or something. So you shouldn't expect that to indicate how analog meters work.
My friend has a car manufactured in 1982 and the speedometer is practically useless below 30mph because of how bad the tremble is. He can be driving at a fairly constant 25mph and the needle will fluctuate rapidly between 30mph and bottomed out. Given the fact that the gauge bottoms out at 10mph, that behavior is probably by design.
Interesting, I didn’t know that.
Unless you're in Scientology, the gauge is supposed to be an analogue of whatever the current value it's measuring is. If the data are noisy, the gauge may tremble a bit, but if the data are steady the gauge should be steady.