Computers slightly better at identifying sex than climate change

Computer models are often used instead of common sense because computers are faster and cheaper than finding experienced people with good judgment. Unfortunately computers are not much better at determining your sex from your writing than determining climate change from broken ground stations.

Example One: Gender Identification by Gender Genie

Those of you with inquiring minds and some spare time on your hands might want to submit a little of your work to the Gender Genie.
This a program that analyzes text and identifies the gender
of the writer based on a scan of key words and syntax.

Do NOT take this test if you are inclined to accept the output of
computer programs as valid evaluations of your self-worth, gender,
intelligence, or general value to humanity. I do not believe that the
results should be interpreted in any way but as an amusing diversion.

The program was developed by Moshe Koppel of Bar-Ilan University in
Ramat Gan, Israel. Koppel and colleagues trained their algorithm on a
few test cases to identify the most prevalent fingerprints of gender
and of fiction and non-fiction.

Gender Genie is claimed to be 80 percent accurate when the submitted
text is 500 words or more, but I question the validity of the results,
as they are based upon patterns in written works, not anything that
actually validates sexuality or gender.

Writing is such a fluid means of expression, with patterns that
change so markedly over time, that it is difficult to believe that
syntax is a reliable reference for gender. When I look at my current writing
style, I see little in common with my writing of a few years ago.

This is a fine way to spend a few minutes amusing yourself, but you would not want this process to have any control over your life. The algorithm is incomplete at best and is highly subject to the data you give it. At least it is claimed that it gives results that are right about 80% of the time and it has no preconceived bias built into it to skew the results..

Example Two: Climate Change Prediction

In this case, the most powerful computers available are being used to massage data from thousands of points on land and at sea to predict changes in Earth's climate.  Currently the computer results indicate that the seas are rising, the temperatures are rising catastrophically and we are all in for a disaster which can only be alleviated if we shut down essential industries, and buy carbon dispensations from Al Gore's company.

This prediction is being stalwartly supported by those who will profit from selling carbon credits and by those who feel that "green" industries do not need to be economically justified.

The unfortunate truth is that yearly temperatures have been falling for the past few years and that the majority of weather stations ( more than 70%) have been compromised by the encroachment of buildings, parking lots, incinerators and air conditioning units.

The computer model has been adjusted to produce a "hockey stick" output in the presence of data that is mostly white noise. The measurement errors have been known and published for several years, but the advocates of global alarmism are telling everyone to ignore the men furiously rushing about behind the curtain making adjustments to the data so the results come out right.

If the climate disaster predictors were half as good as the local weather predictions (30% – 50% right) they still would not measure up to the Gender Genie discussed above. Using such inaccurate output to control government policy is extremely unwise.

For a thorough discussion of the pros and cons of global warming, read Global Warming, the science & impacts of global warming.

My takeaway from this is that computer models have a long way to go before we should trust our future to their predictions.

Every solution has a cost and we should be solving real problems, not imaginary problems dreamed up by opportunists who are out to make a fortune from the credulous.

This entry was posted in Global Warming?, Technical Amusements and tagged , , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

9 + one =

This site uses Akismet to reduce spam. Learn how your comment data is processed.