Software Nerd

Thursday, January 31, 2008

Scale on "Global Warming" graphs

Since graphs are visual, we tend to measure them visually at first glance. At first glance, "steep" means visually steep. Consider these two graphs of company profits. Visually, one appears to show stagnating profits, while the other appears to show growth. Yet, the facts depicted are absolutely identical. The only difference is the y-axis scale. This is one of the oldest tricks in the book. Many company annual reports have stuff like this.

When Al Gore present his "global warming" graph, it looks something like this. the upward slope is visually steep. That's the message Gore wants his audience to take home. However. we aren't illiterate, so we take a second look. What's the scale? Is it appropriate? There appears to be an upward trend, but how do we judge if it is significant?

An expert may have an involved answer, but how do I -- a lay-person -- go from first glance to second-glance?

Here is how I approach the notion of an appropriate scale as a layperson: I try to relate the information to some concrete of the same type, which is familiar. If I have a familiar concrete of the same type, and overlay this information over that, I can better relate the visual to reality. So, here is the lay-person attempt:

1. This is a graph of temperature, so let me consider some familiar climate/weather temperature. Let me choose a city in the middle of the U.S. Let's say, I choose temperature variations in Topeka, Kansas.
1a. Google finds information on Topeka's averages for each month.
1b. I convert to Centigrade (because that's what the "global warming" graph uses)
1c. I express the mean of all months as "0 C", and express others as Centigrade above or below. (This does not change the scale, simply the naming of the Y-axis, to make it similar to Gore-style charts.)

2. That was a single "average" year. Next, I ask myself what Topeka's temperature would look like if it had the same "average-year" type of temperature year after year. So, instead of 12 months on the x-axis, I pack in about 30 years. Since I did this manually, it's not exactly even. Still, I would expect the actual year-after-year variations to be far more uneven. So, this is a pretty good depiction of what Topeka's temperatures would look like over 30 years is it is are basically unchanging.

Eyeing the tops and the bottoms, we can see what an almost zero-trend would look like.

3. Now, use the "zero-trend" as a background and overlay the "global warming" graph onto it, bringing the scale of that graph down to match the scale of this one. Here is what we get:

Are you scared now?

Compared to our flat background, the "global warming" graph does show an upward trend, but imagine Gore presenting this graph to his audience. Would it have the same impact?

It is the same data, after all.

I do concede that I'm presenting this as a lay-person. Maybe an expert will explain why this graph that looks insignificant from a lay-person's viewpoint is really significant. That's fine. However, to present this as if it is obviously a steep and significant slope is to play the charlatan.

Look at this correctly scaled graph and add in the fact that some experts question the data and say that it really isn't even as "steep" as depicted. Pardon me if I'm skeptical!

Friday, January 25, 2008

An Education Stimulus Package

Washington,D.C. (Jan 25, 2007): President Bush and House Speaker Nancy Pelosi today announced a bipartisan stimulus plan to improve higher education. Announcing the initiative on the White House lawn, the president said: "This package is urgent. We must boost falling grades with an immediate stimulus. I urge Congress to act soon."

The president had suggested increasing every student's GPA by 0.2 points, but Democrats objected, saying that help should be targetted to those who need it most. "The President wanted to use the failing grades of poor students as an excuse to raise grades for the good ones", speaker Pelosi commented, "while we wanted to help those who really need the help".

The compromise plan will add 0.4 to all students with a GPA under 3.0, but will add less for students scoring better, gradually phasing out to zero for anyone with a 3.8 or above.

On the campaign trail, Senator Obama criticized the deal, as a quick-fix, saying that grades were not enough; students who do well should be made to pay more, since they are obviously getting more from the system. The amounts raised could be used to pay for extra tuition for students who are failing. "We need a plan that creates hope", said the Senator, "we need creative solutions". GOP candidate Mike Huckabee, called for more structural changes and science programs that taught Intelligent Design.

Meanwhile, most students interview were happy with the new plan. "I can definitely use the help I can get", said Mitt Koplaski, who studies economics at the University of Oregon.

My question is this: if you can tell that this is spoof, explain why, in principle.

[Hat tip: "UMassHoops", posting on "The Motley Fool", for the kernel idea.]

Saturday, January 05, 2008

Trusting Computers?

Here are some funny stories:
First, "The Register" reports: "The driver of a £96k Mercedes SL500 had a lucky escape after her satnav directed her down a winding track and straight into the River Sence in Sheepy Magna, Leicestershire..."

Next, we have the 78-year old man who's GPS told him to "make a U-turn immediately", when he was on a highway (an 80 MPH speed limit type).

Now, we have a guy in upstate New York, who was driving across a railroad track when his GPS told him to turn right. So, he turned right, onto the tracks, making like he was a train! The car got stuck -- probably those damn horizontal sleepers and troublesome ballast stones, so car unfriendly! He got out and the train smashed his car. According to other reports, not only did he wreck the car, but this ended up holding up the line for 2 hours.

Any lessons from this? Are these just funny stories to be forgotten, or is there a lesson from them? I don't know, but I wonder. The apparent stupidity is some type of mistake. Is it "trusting computers too much"? For example, if another human was sitting beside them and had given them the same instructions, would they have followed blindly?

Why do people have this type of trust in computers? After all, some person put the data into the computer. Is there something about getting instructions from a device that lulls some people into forgetting what is metaphysical and what is man-made?