snopes.com  

Go Back   snopes.com > Urban Legends > Computers

Reply
 
Thread Tools Display Modes
  #1  
Old 01 January 2008, 05:40 PM
snopes's Avatar
snopes snopes is offline
 
Join Date: 18 February 2000
Location: California
Posts: 109,659
Computer Software catastrophe myths

Comment: I live in Australia, and work in computers. There are a few computer
horror stories we have heard (at least 10 years ago) which are used to
drive home the point of bad programming. I think they are Urban Myths,
but a friend thinks they are real.

1. Error 21. A machine used to give radiation treatments to cancer
patients started displaying 'Err21' during operation, but the process
seemed to complete normally. The technicians ignored the error until
people started dying from radiation poisoning.

2. Inverted flight. An updated software in a aeroplane (generally
believed to be a fighter plane) was being tested by the US Air Force.
When the plane flew across the equator (from the Northern Hemisphere to
the Southern Hemisphere), the autopilot software inverted the plane. When
the pilot righted the plane, the software informed him that he was flying
upside down. When the pilot crossed back over the equator, everything
returned to normal.

3. Flight path. In New Zealand an airline company offered a scenic
tourist flight. The path was entered incorrectly into the autopilot. The
pilot reported that the incorrect flight was actually better than their
intended path, but that in one place the altitude should be lowered as it
would create a stunning flight between two mountains. This was done and
the new flight path was a success. After some time a person noticed that
the flight path used by the tourist flight was different from the original
intended flight path and so updated the path without changing the altitude
settings. On the next flight, the plan flew into the side of a mountain
and killed everyone on board. As a side issue to this story, I remember a
tourist flight in New Zealand crashing into a mountain which happened
around 20 years ago, this is the flight generally linked to this story.

One of the reasons I think these stories are Urban Myths is that my friend
is very short on details which could be used to verify the stories, and
when I have talked to other computer people who have heard of them, they
also don't know any details. On the other hand, my friend is married to a
person who works in a hospital and she has heard the first story (error
21) from her work as well.
Reply With Quote
  #2  
Old 01 January 2008, 08:50 PM
Dropbear's Avatar
Dropbear Dropbear is offline
 
Join Date: 03 June 2005
Location: Hobart, Tasmania, Australia
Posts: 7,237
Default

The third one does sound very much like a mildly garbled version of the Mt Erebus disaster - where a change of coordinates was a factor in causing a tourist plane to crash directly into a mountain. This transcript of a science program from Australia has a nice summary of the causes.

Dropbear
Reply With Quote
  #3  
Old 02 January 2008, 06:59 AM
Penny Penny is offline
 
Join Date: 19 July 2004
Location: Sydney, Australia
Posts: 682
Default

The first sounds like the Therac-25 problems.
Reply With Quote
  #4  
Old 02 January 2008, 08:42 AM
Troberg Troberg is offline
 
 
Join Date: 04 November 2005
Location: Borlänge, Sweden
Posts: 11,580
Default

I've heard #2 as well, supposedly it was a F-14 that experienced it. While a bug could give a result like that when a number unexpectedly turns negative, it's still highly unlikely that such a problem would sneak past quality control.

Another southern hemisphere bug story:

http://gravitywave.blogspot.com/2007...falklands.html

"Why the delay? For years, physics textbooks have cited this battle as an example of the Coriolis force in action. The Coriolis force is a "fictitious force" much like the centrifugal force one feels on a merry-go-round. The Coriolis force has the effect of making projectiles deflect to the right in the Northern hemisphere, and to the left in the Southern hemisphere. The force isn't noticeable for small distances and velocities, but for long-range naval gunnery, it becomes important. The primitive mechanical firing computers of WWI could calculate and correct for the Coriolis force. However, the story goes that the British forgot that in the Southern hemisphere, the opposite correction needs to be made."
Reply With Quote
  #5  
Old 02 January 2008, 09:21 AM
hevach
 
Posts: n/a
Default

#2 reminds me of one that got into some of the professional PC magazines in the late 90's that the Navy had just deployed a new navigation software on a number of ships. They were thuroughly tested in shakedown cruises in the North Atlantic and around Hawaii. The first time an active duty ship crossed the equator using the software, a divide by zero error crippled the vessel.

I never saw confirmation of it, and I doubted at the time. The editorials both blamed it on the Navy switching to Windows operating systems (one said 95, which Microsoft marketed as a home or single user system, and the other said NT4, which is possible), but the description is clearly a software failure, not an OS failure.
Reply With Quote
  #6  
Old 02 January 2008, 10:16 AM
ganzfeld's Avatar
ganzfeld ganzfeld is offline
 
Join Date: 05 September 2005
Location: Kyoto, Japan
Posts: 23,785
Default

Quote:
Originally Posted by hevach View Post
#2 reminds me of one that got into some of the professional PC magazines in the late 90's that the Navy had just deployed a new navigation software on a number of ships. They were thuroughly tested in shakedown cruises in the North Atlantic and around Hawaii. The first time an active duty ship crossed the equator using the software, a divide by zero error crippled the vessel.
I'm wondering if you mean the incident on the Yorktown, which AFAIK wasn't an equatorial problem. (But I also seem to recall another older naval divide-by-zero incident mentioned in programming texts.)

http://en.wikipedia.org/wiki/USS_Yorktown_(CG-48)
Quote:
In September 21, 1997 while on maneuvers off the coast of Cape Charles, Virginia, a crew member entered a zero into a database field causing a divide by zero error in the ship's Remote Data Base Manager which brought down all the machines on the network, causing the ship's propulsion system to fail.
About the OP story, though, several months ago there was supposedly a similar problem with F-22s crossing the international date line. I can't confirm the validity of the story but it was mentioned on some blogs, eg:
http://www.defensetech.org/archives/2007_02.html

At that time, the OP story was told in blog comments about F-16's in a simulation, with links such as these:
http://catless.ncl.ac.uk/Risks/3.44.html
http://www.csl.sri.com/users/neumann...trative.html#8
I haven't seen any confirmation of this story, either.
Reply With Quote
  #7  
Old 02 January 2008, 10:31 AM
ganzfeld's Avatar
ganzfeld ganzfeld is offline
 
Join Date: 05 September 2005
Location: Kyoto, Japan
Posts: 23,785
Default

Quote:
Originally Posted by Troberg View Post
Another southern hemisphere bug story:

http://gravitywave.blogspot.com/2007...falklands.html

"Why the delay? For years, physics textbooks have cited this battle as an example of the Coriolis force in action. The Coriolis force is a "fictitious force" much like the centrifugal force one feels on a merry-go-round. The Coriolis force has the effect of making projectiles deflect to the right in the Northern hemisphere, and to the left in the Southern hemisphere. The force isn't noticeable for small distances and velocities, but for long-range naval gunnery, it becomes important. The primitive mechanical firing computers of WWI could calculate and correct for the Coriolis force. However, the story goes that the British forgot that in the Southern hemisphere, the opposite correction needs to be made."
This story is told as a true story by scientists every now and then but is it true?

One blogger looked into it a bit:
http://www.mememachinego.com/2004/09...tracts_my.html
Quote:
There are myriad references to these battles on the web.

Not one of them mentions the Coriolis force, or shells mysteriously missing their targets. Neither do Bennett’s books. All of them agree that in both battles, both sides were doing damage to each other as soon as they were in range. In fact, the only references to the Brits’ failing to correct for the Coriolis force I can find are in physics lessons by Americans.
Reply With Quote
  #8  
Old 02 January 2008, 11:22 AM
hevach
 
Posts: n/a
Default

Quote:
Originally Posted by ganzfeld View Post
I'm wondering if you mean the incident on the Yorktown, which AFAIK wasn't an equatorial problem. (But I also seem to recall another older naval divide-by-zero incident mentioned in programming texts.)

http://en.wikipedia.org/wiki/USS_Yorktown_(CG-48)
That sounds like it might fit. The dates are definitely in the area. The editorials were referring to a battleship, but that could easily be a poor understanding of war ship terminology.

Anyway, sounds like somebody doesn't sanitize inputs. It could be worse, I guess. Somebody could change their name to ';DROP DATABASE;--.
Reply With Quote
  #9  
Old 02 January 2008, 12:08 PM
Troberg Troberg is offline
 
 
Join Date: 04 November 2005
Location: Borlänge, Sweden
Posts: 11,580
Default

Quote:
This story is told as a true story by scientists every now and then but is it true?

One blogger looked into it a bit:
A comment further down supposedly links to a site that claims that the coriolis effect is significant in artillery, but it's a dead link.

It should be easy enough to calculate, though, if one knows how long a projectile stays airborne and the latitude where it's fired. Just calculate how far the earth surface moves due to the rotation during that time, and you should get a pretty decent estimate.

Any artillery specialists here?

If it's false, it would at least explain why the Invincible and Inflexible beat Gneisenau and Scharnhorst that badly, as they did not outgun them enough (even though they did have a significant advantage) to get such a victory with such a screw up.
Reply With Quote
  #10  
Old 03 January 2008, 01:39 AM
ganzfeld's Avatar
ganzfeld ganzfeld is offline
 
Join Date: 05 September 2005
Location: Kyoto, Japan
Posts: 23,785
Whalephant

Quote:
Originally Posted by Troberg View Post
Just calculate how far the earth surface moves due to the rotation during that time, and you should get a pretty decent estimate.
You should? I'm not a physics guy but that doesn't remotely correct to me. (When shooting from the poles, where the effect would be greatest, I think your estimate might work if you calculate the motion of the spot over which the projectile is. At the equator, where the effect is least, your estimate must be completely useless. Perhaps with a bit of subtraction one can come up with a rough estimate: the difference of rotation between firing spot and target. Maybe that's what you meant.)

The provenance of the story is so questionable that I don't see a lot of reason to investigate the possibility -- except for the fun of it and to possibly drive one more stake through its heart. Maybe one of the board physicists can help us out a bit.
Reply With Quote
  #11  
Old 03 January 2008, 02:01 AM
Beachlife!'s Avatar
Beachlife! Beachlife! is online now
 
Join Date: 22 June 2001
Location: Lansing, MI
Posts: 28,667
Skull

Quote:
Originally Posted by ganzfeld View Post
You should? I'm not a physics guy but that doesn't remotely correct to me. (When shooting from the poles, where the effect would be greatest, I think your estimate might work if you calculate the motion of the spot over which the projectile is. At the equator, where the effect is least, your estimate must be completely useless. Perhaps with a bit of subtraction one can come up with a rough estimate: the difference of rotation between firing spot and target. Maybe that's what you meant.)

The provenance of the story is so questionable that I don't see a lot of reason to investigate the possibility -- except for the fun of it and to possibly drive one more stake through its heart. Maybe one of the board physicists can help us out a bit.
But doesn't the atmophere and everything in it rotate as well? Otherwise, one would be able to go around the world by flying straight up and waiting for the earth to do the rest.
Reply With Quote
  #12  
Old 03 January 2008, 02:22 AM
UEL's Avatar
UEL UEL is offline
 
Join Date: 01 August 2004
Location: Fredericton, Canada
Posts: 9,381
Default

Quote:
Originally Posted by Troberg View Post
Any artillery specialists here?
I'm an artillery specialist, albeit, a specialist in surveillance systems. I have spent many a year doing the field artillery job.

Coriolis effect is taken into account when computing artillery fire. However, given the nature of the calculation of artillery fire, it is but one of 17 variables through which we have to calculate. The truth is that Coriolis effect is a very minor calculation, that is often dropped when doing manual work. It is always kept in when doing work via computer.

Example: we have a howitzer that can fire 24 kilometres. At the maximum range, the maximum deflection of azimuth for the Coriolis effect is 3 to 5 mils, which is about .17 to .28 of a degree. Over the course of the maximum range, the error introduced by this works itself out to be less than 120 metres.

120 metres may seem like quite a bit, however, when given the fact that meteorology (wind speed, direction, pressure, temperature, humidity) has an effect on the order of a couple of hundred metres (and your meteorology is never entirely current), temperature of the propellant has an effect, temperature of the barrel, altitude of target vs gun, accuracy of target location, accuracy of gun location, accuracy of gun orientation, dispersion of guns, barrel wear, weight of the projectile, it is not uncommon to have rounds 120 metres off for any number of reasons. We call it the "error budget". We try to minimise it as much as possible, but eliminating it is all but impossible. Met is the biggest reason the bullets do not hit.

Added to that, we rarely ever fire maximum range. This 120 metres is worst case scenario for Coriolis, most often it is much, much less.

Ships have even more errors to contend with when firing (velocity, direction, pitch and roll of the ship).
Reply With Quote
  #13  
Old 03 January 2008, 02:32 AM
ganzfeld's Avatar
ganzfeld ganzfeld is offline
 
Join Date: 05 September 2005
Location: Kyoto, Japan
Posts: 23,785
Hello Kitty

Quote:
Originally Posted by Beachlife! View Post
But doesn't the atmophere and everything in it rotate as well?
Yes, of course it does. The point of my post was partly that: That you have to compute the relative motion of an object that is rotating with a particular spot on the Earth. As for the resistance from the moving atmosphere and wind (which are also subject to the Coriolis effect but at much slower speeds and therefore to a much smaller degree), of course you'd have to compute that as well. The faster the object, the less it would matter but I have no idea how much that would be relative to the Coriolis effect for an artillery shell. (This is the problem with physicists -- they're never around when you need them. ETA - Ah, but we have artillery support now so we're good. Thanks UEL!)
Reply With Quote
  #14  
Old 03 January 2008, 06:16 AM
Troberg Troberg is offline
 
 
Join Date: 04 November 2005
Location: Borlänge, Sweden
Posts: 11,580
Default

Quote:
You should? I'm not a physics guy but that doesn't remotely correct to me. (When shooting from the poles, where the effect would be greatest, I think your estimate might work if you calculate the motion of the spot over which the projectile is. At the equator, where the effect is least, your estimate must be completely useless. Perhaps with a bit of subtraction one can come up with a rough estimate: the difference of rotation between firing spot and target. Maybe that's what you meant.)
You're right, and I was a bit brief in my description, merely listing the variables for the calculation.

Quote:
The provenance of the story is so questionable that I don't see a lot of reason to investigate the possibility -- except for the fun of it and to possibly drive one more stake through its heart.
I was only in it for the fun physics.

Quote:
I'm an artillery specialist, albeit, a specialist in surveillance systems. I have spent many a year doing the field artillery job.
Thanks, that's the kind of information I wanted.

Quote:
Ships have even more errors to contend with when firing (velocity, direction, pitch and roll of the ship).
That's one thing I've thought a lot about. When firing at the ranges that the big naval guns are capable of, even a minor error in angle will cause a huge deviation. How the hell do they compensate for firing off a moving platform that's also rolling and heaving in every direction? The same probably goes for firing from a moving tank.

Quote:
This is the problem with physicists -- they're never around when you need them.
And they tend to make their calculations using such devices as "easily movable pulley", "frictionless wagon", "ball falling in vacuum" and so on. That's why they usually don't have the 17 variables that UEL has.
Reply With Quote
  #15  
Old 03 January 2008, 12:20 PM
Hans Off's Avatar
Hans Off Hans Off is offline
 
Join Date: 14 May 2004
Location: West Sussex, UK
Posts: 4,660
Default

Quote:
Originally Posted by Troberg View Post

That's one thing I've thought a lot about. When firing at the ranges that the big naval guns are capable of, even a minor error in angle will cause a huge deviation. How the hell do they compensate for firing off a moving platform that's also rolling and heaving in every direction?
Timing and practice I would have thought!

Even so, as UEL said, at 25 kilometers the deviation can be up to 120 meters, for a WWII battleship such as the King George V can engage at ranges far above that, at a target such as the Bismark that is only 251meters by 36 meters, the effect is very significant.

Quote:
The same probably goes for firing from a moving tank.
Modern tanks have self levelling guns to negate any rock and roll, Older tanks tended to halt before firing.
Reply With Quote
  #16  
Old 03 January 2008, 12:55 PM
Troberg Troberg is offline
 
 
Join Date: 04 November 2005
Location: Borlänge, Sweden
Posts: 11,580
Default

Quote:
Modern tanks have self levelling guns to negate any rock and roll, Older tanks tended to halt before firing.
I know that (I've played Operation Flashpoint CRCTI a lot...), but I was more curious about how it's done technically. It's not a laser gun, to compensate correctly, the trajectory must be part of the calculation in some way, which means the distance is needed. It's quite an impressive problem.

Many modern tanks have only vertical compensation, not horisontal. I don't know why this is, but there must be some reason.

Quote:
Timing and practice I would have thought!
I think firing all the big guns at the same time also helps, as it eliminates the problem of recoil from one gun affecting another.

Quote:
Even so, as UEL said, at 25 kilometers the deviation can be up to 120 meters, for a WWII battleship such as the King George V can engage at ranges far above that, at a target such as the Bismark that is only 251meters by 36 meters, the effect is very significant.
One also have to consider the basic maximum precision in the system. Even if the gun was in a fixed position, firing a sequence of shells under identical conditions, they would not hit the same spot (with some degree of probability, of course). I wonder how large this effect is?

Last edited by Troberg; 03 January 2008 at 01:01 PM.
Reply With Quote
  #17  
Old 03 January 2008, 02:15 PM
Arriah's Avatar
Arriah Arriah is offline
 
Join Date: 15 August 2005
Location: Everett, WA
Posts: 3,663
Goldfish

Quote:
Originally Posted by Troberg View Post
That's one thing I've thought a lot about. When firing at the ranges that the big naval guns are capable of, even a minor error in angle will cause a huge deviation. How the hell do they compensate for firing off a moving platform that's also rolling and heaving in every direction? The same probably goes for firing from a moving tank.
Also, they don't shoot single shot wepons from the wobbly ship. They use one (or more) of these or its relatives. When you can shoot several thousand shots in a minute, I think it gets a bit easier to hit the target with at least some of them.
Reply With Quote
  #18  
Old 03 January 2008, 03:01 PM
UEL's Avatar
UEL UEL is offline
 
Join Date: 01 August 2004
Location: Fredericton, Canada
Posts: 9,381
Quote:
Originally Posted by Troberg View Post
I know that (I've played Operation Flashpoint CRCTI a lot...), but I was more curious about how it's done technically. It's not a laser gun, to compensate correctly, the trajectory must be part of the calculation in some way, which means the distance is needed. It's quite an impressive problem.

Many modern tanks have only vertical compensation, not horisontal. I don't know why this is, but there must be some reason.
Maximum range on a tank is about 5 kms. It is fired direct. Based upon the velocity of the tank round, the range and the manner of firing, I would estimate that the drift from the Coriolis effect would be in the order of a metre or two. It is much smaller than

Quote:
Originally Posted by Arriah View Post
Also, they don't shoot single shot wepons from the wobbly ship. They use one (or more) of these or its relatives. When you can shoot several thousand shots in a minute, I think it gets a bit easier to hit the target with at least some of them.
Interestingly enough, when firing onto land, they adjust their fire by firing only one gun, then when it's on target, they engage with all. At sea, that's another story.
Reply With Quote
  #19  
Old 03 January 2008, 05:47 PM
Hans Off's Avatar
Hans Off Hans Off is offline
 
Join Date: 14 May 2004
Location: West Sussex, UK
Posts: 4,660
Default

Quote:
Originally Posted by Troberg View Post
I know that (I've played Operation Flashpoint CRCTI a lot...), but I was more curious about how it's done technically. It's not a laser gun, to compensate correctly, the trajectory must be part of the calculation in some way, which means the distance is needed. It's quite an impressive problem.

Many modern tanks have only vertical compensation, not horisontal. I don't know why this is, but there must be some reason.
IIRC the verical compensation is achieved by the use of an inclinometer feeding back into the elevation control for the gun, so what ever the desired elevation of the barrel is set to (say, 18 degrees) the inclinometer will feedback whenever there is a deviation, thereby keeping the elevation set correctly.

I was told this by a bloke that worked for a geotechnical firm who said that the inclinometers that they used for borehole surveys were manufactured by the same firm that provided them for the Challenger II tank.

See... http://www.icefieldtools.com/docs/mi3_howitworks_en.pdf

It is the tube thing on the right!

The reason you don't have horizontal compensation is that having the turret swinging around on its own is more trouble than it is worth, there would be a danger of the barrel tracking into a wall or rocks or trees or whatever, or even the ground if the tank is near or on a large incline.

As an aside, I have ben toying with the idea of converting a 1/6 scale Challenger II to operate with a working inclinometer, but I don't think my electronic skills are up to it!
Reply With Quote
  #20  
Old 04 January 2008, 06:28 AM
Troberg Troberg is offline
 
 
Join Date: 04 November 2005
Location: Borlänge, Sweden
Posts: 11,580
Default

Quote:
Maximum range on a tank is about 5 kms. It is fired direct.
I've seen numbers twice that high (usually about the Soviet 125 mm smoothbore guns), but that may be part marketing hype, part a misunderstanding about which targets are fired upon and under which circumstances. It's quite a difference to fire on the move at another than on the move and to fire stationary at a house or bunker.

I think that today, longer distance engagements are usually made with missiles. The Soviets/Russians have some pretty impressive gun launched missiles, and both sides have various variants of missiles launched from external tubes. Being able to steer the missile onto the target makes long range attacks much more likely to hit.

Quote:
Based upon the velocity of the tank round, the range and the manner of firing, I would estimate that the drift from the Coriolis effect would be in the order of a metre or two.
I didn't consider coriolis in this case, I just drifted off the subject.
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is On

Forum Jump


All times are GMT. The time now is 08:12 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2019, vBulletin Solutions, Inc.