Computer modeling is a phrase that seems to always draw respect.

This level of respect has always been puzzling to me. I have done a little computer modeling (in my time at a Tech University), and I learned early on that the "Computer" part is the last part of the process of Computer Modeling.

I also learned that the biggest weakness of Computer Modeling is in the model.
more...

Posted by: karrde at 19:59 | Comments (1) | Add Comment
Post contains 685 words, total size 4 kb.

Some time ago, a commenter asked my opinion (as a programmer) on the evolving mess that is known as ClimateGate.

My first thought was that the problem was not primarily an issue with bad programming (though there are questions about that), but an issue with the Data and the Model.

[NOTE: many of my links are to other bloggers who were skeptical about the causes and consequences of planet-wide rise in temperature before the Climategate story was broken. The discussion below should lead to its conclusion without any assumptions about what is causing the globe to warm, how much warming has occurred, and whether it can be expected to continue into the foreseeable future.]
more...

Posted by: karrde at 21:41 | No Comments | Add Comment
Post contains 573 words, total size 5 kb.

Some blogger tells us that the country is in the very best of hands. Specifically, the website designed to show how well the Recovery effort is working shows many jobs created in apparently-fictional Congressional Districts. Among the fictional districts in New Hampshire is district number "00".

The site also shows $1,471,518 going to New Hampshire’s 6th congressional district, $1,033,809 to the 4th congressional district and $124,774 to the 27th congressional district. In fact, New Hampshire only has two congressional districts; inviting confusion about where the money listed for the 00, 4th, 6th and 27th districts is going.

When I see this data, I see a different problem. Whoever designed and wrote the recovery.gov website left several gaping holes in the software logic on the input end, and in the logic on the storing/validation system. There are also possibly problems in the display system used to show the data to the general public.

Specifically, they should have checked that the inputs made sense relative to the data that was expected. There are ways to make sure that when a wide variety of people enter data into a system, the entries are checked against a range of expected values.

As an example: most websites that handle online-order-forms have an entry for putting an address in. If that entry is for an address in the U.S., there is a state-code to fill out. It is good programming practice to make sure that every state-code put in is a code for a state that exists. Often, the web-site provides a drop-down menu with selections, to make it harder for someone to enter "AK" when they want to ship to Arkansas, or to mistakenly enter "AM" for Alabama.

The problem of having people enter data for Cogressional districts in which Recovery-jobs have been created is similar. Each state has a specific number of Congressional districts. It is absurdly easy (from a software perspective) to turn this input form into a pair of drop-down boxes. Thus, the entry would not be left blank (likely the cause of district "00"). Also, it would be harder for the person entering the data to put a State Congressional District into a box labelled Federal Congressional District. (This might be the cause of districts 4, 6, and 27 in New Hampshire...)

Not only is the management of the Recovery program suspect, but their programming team (and the management team doing the oversight) published some really sloppy work. The website looks nice, but behaves like a first draft ready for testing, not a finished product.

UPDATE: after a careful perusal of each State's data as shown in Recovery.gov, I found that all states except Nebraska and Nevada have some Recovery money spent in Congressional District 00.  Ten of the states (Colorado, Idaho, Indiana, Kansas, Minnesota, Mississippi, Missouri, North Carolina, Rhode Island, and Tennessee) have 0 jobs saved/created in that row at some non-zero cost. Two states (Montana and New Hampshire), as well as the District of Columbia, have more than 1500 jobs listed in District 00. Most of the remaining states have fewer than 10 jobs listed in District 00, but a few have more than 100.

It looks like "00" is a default value, and a few states had almost all number of entries done without a specific Congressional District specified. Most of the states had a few remainders that resulted in this default value.

Posted by: karrde at 13:02 | Comments (1) | Add Comment
Post contains 567 words, total size 4 kb.

My daily routine used to involve lots of reading and writing. (I was a graduate teaching assistant, and later an adjunct professor of mathematics.)

Now, my daily routine requires looking at and thinking about computer code. (Somehow, I've learned a great deal about software...in spite of the poorly-organized programming courses I took, and an almost-complete lack of theoretical work in computer science...)

Computer software is challenging. Software design has often been described as like being a civil engineer designing a bridge of certain dimensions, while knowing that your boss may call you tomorrow and require you to double the length of the bridge, and switch it from a truss bridge to a suspension bridge.

Even bosses who know computer software (like my current boss) occasionally make demands of that type.

The most pernicious piece of trouble I see on a regular basis is version control.
more...

Posted by: karrde at 20:48 | No Comments | Add Comment
Post contains 419 words, total size 3 kb.

I spent part of the weekend overseeing a computer-database system. The database was used as a point-of-sale terminal for a used-book-sale.

Said sale was part of a yearly event. 
more...

Posted by: karrde at 18:34 | No Comments | Add Comment
Post contains 444 words, total size 3 kb.

[UPDATE: clever analogy inserted after the "more" link.]
One of my hobbies--recently turned into a job--is writing computer code.

The coded instructions that tell the computer how to handle files, display images, play music, and browse the internet are typically very complex.

However, the complexity is typically not in the specific instructions at each step, but in the information that the computer must manage while it does the job. Somewhat like complex proteins in a living cell, computer code is composed of a finite number of basic building blocks. The blocks (amino acids or computer instructions) can be combined in many intricate ways to produce a wide variety of tools.

In computer science, one fundamental memory structure is a list of data. Whether the data is meant to represent sports scores, the names in the phone book, or a list of art images, the computer stores the data as a string of numbers.

One problem which can arise is the problem of sorting the data--alphabetizing the phone book, finding the highest-scoring batter in the American League, or arranging the art pictures by date.

How would these jobs be done?
more...

Posted by: karrde at 06:46 | No Comments | Add Comment
Post contains 705 words, total size 6 kb.