Fun with statistics

Yesterday, I got a little involved in the comment-thread of this article by Megan McArdle.

At one point, another commenter posted something about the number of workplace deaths in the United States. Something like "4628 workers were killed on the job in 2012." *

This number might be valid, or might not.

However, I immediately noticed something odd. I remember that a much-higher number of people die in vehicle-related accidents on a yearly basis.

So, basically, I asked this question: since ~35000 Americans die in automobile collisions per year, is it more dangerous to drive to a job, or to be on the job? **

* This statement is a comment about a number, not an argument. But it was offered as if it was an argument.
I think I'm seeing these unspoken assertions:
1. A large number of people died on-the-job in the United States
2. Workers need some form of protection from this
3. Unions are both necessary and sufficient for this protection

This chain of reasoning has several holes.

About point 1: The data provided is only for the year 2012. No data was provided about how this number related to Unions, or to State/Federal regulations about workplace safety. No data was provided about how this number has changed as the level of Unionization has changed over the past three or four decades. No data was provided about how this number has changed relative to man-hours-of-employment over the past few decades.

Point 2 is hard to quibble with, but proposals coming from Point 2 depend heavily on the data that went into Point 1.

About Point 3: this also depends on data. There are many ways to reduce workplace fatalities in the U.S., and the Unions and U.S. Dept. of Labor have been trying to do that for many decades. Perhaps some data about these efforts, and how these efforts interact with business efficiency, corporate practice, and the change of Union membership in the United States, should be brought into the discussion.

** My source on this is the Fatal Injury Statistics website, which is provided by the U.S. Dept. of Health Center for Disease Control and Prevention.
The numbers are between 35000 and 36000 per year for most years since 2009.
Oddly, the number dropped from ~45000 per year in 2007 to ~35000 per year in 2008.
The per-capita rate had shrunk in most years from the 1980s to 2007, but the static value had remained near ~45000/year during most of those year. But there was a big drop in 2008, and it has remained since then.


Vehicle death

Another blogger had to get rid of a vehicle recently.

In his case, he had a Jeep which developed engine trouble...of the kind that rapidly turned into too-expensive-to-be-worth-fixing.

He ran into one of the long-term-ownership problems in the automotive world. Almost all vehicles are now made to have near-zero maintenance trouble. For something like 8 to 10 years.

But most vehicles have design problems that will begin making ownership expensive. Sometimes these problems are foreseeable, sometimes not. Sometimes they arrive shortly after year number 8. Sometimes they wait until 12+ years.

Sometimes they are avoidable. My current vehicle, a Subaru, suffers from potentially-catastrophic failures to the timing belt. These tend to come somewhere between 100000-miles and 110000-miles. However, the belt can be replaced before this becomes a problem.

Other times, these problems are not avoidable. An earlier car I owned (early-90s Taurus) had transmission trouble. This model of car, though extremely popular, regularly suffered transmission failure at ~80000 miles.

Robb's Jeep had a different category of problem, but one with similar results. The engine failed, and in a way similar to most other Jeeps of that model-and-age combination. Thus, most junk yards have such vehicles, but few have replacement engines handy. (Robb noticed that the salvage yard had 10 such Jeeps, 8 of them with the same engine failure. And two engines that did not have such a failure, but required expensive rebuilds anyway.)

And in his case (as with my old Taurus), the cost of repair is more than the resale value of a good, working vehicle of that type.

One of the frustrations of buying used cars is that such things can creep up on the unwary buyer. Yet one of the frustrations of buying a car new is that after you keep it for a decade, you want it for another five years...and problems like this can also sneak up on you.

Like Robb said, it's a First World problem.


Going, going...gone.

Sold a vehicle this week.

The old Jeep had sat for months. One relative had told me they wanted it...then they backed out. Another expressed interest, then they also backed out.

In the meantime, the vehicle had become a little cranky. I didn't establish a habit of running it one day a week. So when I tried to start it, I seemed to find trouble 50% of the time.

Finally got it back into predictably-running condition, and put it on Craig's List. But taking it for test drives reminded me of the parts I enjoyed about the Jeep.

But I was also reminded that I don't like keeping a vehicle I'm not going to use. And I wanted to switch to a stick-shift.

So when the one potential buyer came to look at the Jeep, I was happy to get an offer from them.

I was also happy that they paid cash, rather than cutting a check. We signed the title, shook hands, and they drove away.

Now that the Jeep is gone, I miss her. But I don't miss the cost of filling it with fuel.

And I enjoy a car that gives all-wheel drive and direct control of the shifting.



I feel as if I spent the last week in a deep, dark hole. (Blog-post wise.)

At the office, I discovered that somehow a large amount of merde had been fed into the ventilateur. (Pardon my French.) The result has been long nights, long waits for code to compile, tests of software, comparisons of day-by-day builds, careful combing of log-files, and endless worries about whether our customer will accept the upcoming release.

The bad thing about this is that the customer expects hardware-plus-software to arrive at their factory in final form...near the end of September. And we're working on the Bug-Fix releases, which (theoretically) should have already had most problems ironed out.

Except not all the problems have been ironed out, as we're discovering. And the remaining problems are hard-to-diagnose. Which makes life really hard on everyone in the team.



Riding a motorcycle gave me a slightly different view the weather and local climate.

I have to pay much more attention to rain forecasts. Because a two-wheeled machine can have traction and stability problems on wet pavement. And because the motorcycle has no roof.

Temperatures also have a different meaning when I'm rolling at 45 miles per hour. (Or 65 mph...) A day that goes from 60° F to 80° F means that I need the windbreaker jacket for the morning ride, and the vented jacket for the evening ride. However, a day that goes from 70° F to 90° F doesn't change the ride as much.

On many nights, I can feel the change in humidity when I descend into a river valley. Sometimes it's warm and moist; sometimes it's cool and clammy.

Last night, the temperature dropped from upper-70s to lower-60s. Which means that today, I may drive my car instead of ride my motorcycle.

This summer has been cool, which has decreased my desire to ride as much as possible. But I also have a car that is better than the car I had last year. So maybe I enjoy driving a little more, also.


Computer Problems

I've done quite a bit of computer repair in my time. I'm not a professional computer repairman. (There's more money in writing programs than in repairing hardware.)

Most of that has been software repair. Uninstall bad programs, upgrade programs, pull backup copies of pictures/documents, fix OS installs, clean out registries, etc.

Other things have been hardware upgrades. New displays, new network cards, additional hard-disk drives, replacement/upgrade of optical drives, etc.

Very few times have I needed to replace broken hardware.

Once, the broken hardware was a light bulb. A miniature fluorescent bulb that provided a backlight to an LCD display on a laptop. That fix was challenging, mainly because the laptop in question was not easy to disassemble.

My most recent repair on my own machine was a replacement of a faulty Hard Disc Drive. That drive had begun making strange noises during read/write access. And I sometimes got errors when I executed commands like "svn status" on the subversion repository stored on that drive.

Some time ago, I helped my parents recover a Hard Disc that was failing slowly. It wasn't making funny  noises. It was simply not saving important data in one location. This failure convinced the Windows bootloader that the drive didn't contain a valid filesystem.*

This hardware failure wasn't drive-destroying...but there was no way to restore the copy of NTFS on that drive and keep all the old files. And my parents wanted to extract all the Documents and locally-stored email that had been on that drive.

So I (with some trepidation) began searching for "file system recovery" tools online.

I rapidly discovered that if I could produce an image of the data on that HDD, I could pull all sorts of stuff from it. Including the files that my parents wanted. And deleted files, fragments of deleted files, data stored in the swap file, data stored in the Recycle Bin, etc.**

Among these files/deleted-files/fragments-of-files were lots of emails.

The entire process took me about 12 hours. However, these hours were spread across evenings of several weeks.

This is one reason why, when I read stories like this, I am not surprised. There are several different ways for an HDD to fail. Unless the failure involves destruction of the disc platters inside the HDD metal box, then most of the data can usually be restored/recovered.

If the IT team at the IRS didn't maintain central backups of internal emails, they should still have been able to recover most of those emails from a faulty HDD.

* The deep technical details: one of the data-storage sectors on the disc had gone bad. It could be read from, but all write operations would fail.
This copy of Windows was installed on an NTFS partition. Usually, NTFS can recover from this kind of failure. When the problem is detected during a write, Windows/NTFS can re-send the data to the drive with instructions to save it elsewhere. In many cases, this can happen without Windows telling the user!
However, this particular bad sector involved one of three redundant copies of the Master File Table. Windows and NTFS kept on trying to fix the bad MFT without moving it. The write operation would fail. Windows would check the attempted fix, discover the that the fix didn't succeed, and halt the Startup process. Then the user would see a message about a Startup failure, with a recommendation to try again.
Trying again produced the same result.

** Another aside: at one time, I worked for a company that performed contract work for DARPA, the Advanced Research agency of the US Dept. of Defense.
Several employees at the company had to get Security Clearance, and the company had to follow US DoD procedures for handling Classified data.
The US DoD-approved process for clearing Classified data from an Hard Disc Drive involves a custom program that will attempt to overwrite every bit of every byte of every sector on the HDD. Then the progam repeats this process four more times.
This knowledge came in handy. It also clarified for me that a determined attacker (or IT support person who wishes to comply with legal requirements for data retention) can extract most old data from a Hard Disc. Unless the disc is (a) put through the above process for cleaning out old data, or (b) mechanically disassembled and the components physically destroyed.


Camping Trip

This past weekend was swallowed up by a camping trip.

The church that I am a part of had organized the camping trip.

Good times, even if I didn't do as much out-doors stuff as I originally hoped to.

I hung around at campfire with my friends, had a few pleasant conversations during afternoon and evening. Awoke to a cold morning with mist hovering over the lake. Saw some sandhill cranes flying and simming on the lake. Watched some camp-site volleyball, rode a bicycle around. Did some cooking over a campfire.

At the culmination of the trip, we had an outdoor church service and a baptism ceremony in the lake.

The weekend was good, and restful.

I'm happy.