I was going to start a thread on this, so I'll hop on this one. After years of deciding which side of the line I fall on, I've eventually come to the conclusion that it is safer to use torque wrenches these days. Sure, in the past it probably wasn't all that critical. But now, things are engineered to a higher and higher precision, and so the tolerances are becoming lower. I found an interesting article
here. Of particular interest to me in this article is this part:
"At Interbike [2008], we had a transducer set-up to a 0.25in breaker bar and hex bit socket, and we asked for volunteers to tighten the bolts at the stem/steerer to the recommended torque value [4.5Nm – a common spec for those types of parts]," said marketing product manager Glenn Kalnins. "Of the nine volunteers, only one person was within spec (+/- four percent) of the 4.5Nm value [and] some people pulled five and six times the value that was set on our clicker.
"Imagine your stem faceplate needs to be tightened to 5Nm to hold your brand new carbon bar in place and you or your local mechanic tightens it to 25Nm or 30Nm. This variability is totally unacceptable in terms of safety and liability in the bicycle industry."
Given that quoted safety factors in some high-end parts can be as low as 1.1 (meaning that exceeding the spec by just 10 percent can result in failure), it doesn't take much of a ham-fisted mechanic to do some real damage. Conversely, the narrow window also means it's easy to leave a bolt too loose, resulting in unwanted movement or slip.
So, I shall be ordering a torque wrench just as soon as I can persuade my wife that I need one! Trouble is, though, and this is often not known, one torque wrench probably isn't enough: one is needed for low torques, and another is needed for high torques.