Posted By Paul Kafasis on August 24th, 2006

Recently, a user (call him John Smith) emailed us indicating that his Mac Mini would no longer boot following an update to OS X 10.4.7. He determined that this must be due to Audio Hijack, as his other Macs updated fine (and presumably don’t have Audio Hijack installed). Our reply is likely fairly predicatable:

“Hi John,
I’m sorry to hear about the issue you’re having. However, I can assure you it’s not related to any of our products. Nothing that Audio Hijack does occurs until you’ve launched it, let alone at boot time. I’m afraid you’ll need to look elsewhere to find the issue here. Good luck with it.”

To be sure, there’s just no way for Audio Hijack to interfere with the booting of a machine. Nothing at all occurs until Audio Hijack is actually launched, and even then it’s just an open application the same as TextEdit or any others. Once you hijack audio, it’s affecting the source and nothing else. Put simply, there was no doubt the problem lay elsewhere, not with Audio Hijack.

I mention this because it’s something we see too often. Specifically, I mean that a problem occurs and as a user has just installed our software, the problem is associated with us. To be sure, problems do occur with our software, like any software. But in most cases like these, the problem is unrelated to us, and the user has committed a logical fallacy. Specifically, this fallacy is known as a Post Hoc fallacy, from the latin Post Hoc, Ergo Propter Hoc, which means of course “Beware the Ides Of March”. No, no, it means “After this, therefore because of this”.

In logical terms, it works out as:

* When A occurs, B occurs.
* Therefore, A causes B.

Perhaps you’ve seen this chart before (source: Venganza.org):

The idea is the same – the rise in global average temperature is due to the decline in pirates (a false conclusion)*. Obviously this isn’t the case (that is obvious, right?), but the same sort of mistake is often made regarding support issues. In John Smith’s case, isolating differences between machines is a good way to go. However, there are certainly more variables than just “Is Audio Hijack Pro installed?”.

I don’t expect this sort of thing to ever go away entirely. It’s natural to look for causes and effects, and often times, it is helpful to know what occurred prior to a problem. However, the automatic assumption that a problem happened after an event and was therefore caused by the event, is definitely worth re-examining. If nothing else, it may lead to faster troubleshooting in the future.

* Technically, this is an example of Correlation implies causation or cum hoc ergo propter hoc (With this, therefore because of this), but the fallacy is pretty much the same. Also, the chart amuses me.