DailyDirt: Lies My Computer Told Me…
from the urls-we-dig-up dept
We trust automated solutions to perform all kinds of critical tasks, but how often do we verify that we’re actually getting the right results? We survived the Y2K bug, but there are plenty of other examples of software and hardware flaws that could be much more (deadly) serious. Here are just a few disturbing computer glitches that you might have missed.
- Recently, it’s been discovered that some Xerox machines don’t make exact copies, and in an effort to “clean up” images, they can introduce egregious errors that can swap numbers. The scary thing about this is that you might never notice that the photocopier changed your numbers around for you…. #youonlyhadonejob! [url]
- Intel sold millions of defective Pentium chips that didn’t quite perform calculations correctly, and a math professor (Thomas Nicely) discovered the problem while he was working on a computational number theory project. Intel ultimately recalled the processors, but it publicly stated that the flaw would only affect a typical PC user with a spreadsheet less than 1 time every 27,000 years. [url]
- In the mid-1980s, the Therac-25 medical accelerator had a software problem that killed at least 5 people and others were seriously injured from radiation overexposure. A race condition in the Therac-25’s operating system that allowed the device to turn on an electron beam even when its metal X-ray target was out of position. [url]
If you’d like to read more awesome and interesting stuff, check out this unrelated (but not entirely random!) Techdirt post via StumbleUpon.
Filed Under: bug, computer, defective, error, flaw, glitch, mistake, pentium bug, photocopier, race condition, software, therac-25, thomas nicely, y2k
Companies: intel, xerox
Comments on “DailyDirt: Lies My Computer Told Me…”
Intel Pentium Bug
The infamous Intel Pentium math bug is an intersting problem: There was a reward, never claimed, for someone who could find a practical situation where the bug actually made a difference, besides among those using the coprocessor to do computational number theory.
That is, since the problem showed up in the fifth or sixth significant figure, there were no practical calculations in which the bug mattered.
Now, let’s see if my beautiful, sacred theory survives contact with profane reality. (Grin)
Re: Intel Pentium Bug
Yeah, no one at home needed many significant digits with accuracy back then, but things might be a bit different today. Lots of software is available for home use, or one might write it themselves.
Amateur astronomers need accurate ephemerides
Electronic hobbyists need accurate circuit simulation
Who knows, you might want to calculate some standard deviation
……
Therac-25
Not really a “software” problem. The T-25 was controlled by a relay ladder on a state machine. The instructions for operating it (as I recall) said that the technician should select the appropriate aperture to deliver the prescribed dose and then turn on the electron beam (another button on the controller). The problem arose because the targets were surrounded by lead and moved very slowly. There was no interlock to check that the target was in place before the beam was energized, nor was there an external indication that it was so if the technician was impatient, the patient was zapped.
The case became a standard safety failure example in the design of programmable logic control ladders.
Re: Therac-25
The linked article goes into quite a bit of detail about the software race condition that caused the problem, which would lead to the display showing a different set of parameters than were being executed by the beam. Maybe you’re thinking of am earlier model?
OK, so Iraq war started because somebody copied the initial WMD report on a Xerox machine, I see…
Intel sold millions of defective Pentium chips...
2 + 2 = 4*
*Except on a Pentium
Serious Bug - just wait
In January of 2038 a VERY serious bug in the time keeping software in MANY computers and applications will cause all sorts of problems. 32bit roll over on the time counter.
2147483647 – Mon, 18 Jan 2038 22:14:07 -0500
The next second is -Fri, 13 Dec 1901 15:45:52 -0500
Much more serious than the Y2K problem.
BTW: both DEW and some Gov. agencies were down for several days, but they kept it quiet.
Re: Serious Bug - just wait
This is true, however we have lots of lead time and mitigation is being worked on as we speak.
I wouldn’t bring up Y2K in relation to this, for two reasons. First, it’s a different kind of problem and second, the unbelievable overhyping of Y2K was crying wolf. If we want the Unix rollover to be taken seriously, the best thing we could do is to distance it as far as possible from the BS that was Y2K.
We’ve been told that although people make mistakes, and lie, computers don’t. And who told us that? PEOPLE!
Mike is not man enough to admit it. Instead, he is deliberately lying and pretending like he has no idea why it’s happening.
How dishonest can one man be? Mike breaks the mold, it seems.
Mike breaks the mold, it seems.
How dishonest can one man be?