I think that article is misleading in that while it's commonly acknowledged that an engine dyno is more accurate at measuring engine output, the article seems to be trying to make the case by claiming engine dynos measure torque directly while chassis dynos calculate their data. To be honest ALL dynos "calculate" their power data. When comparing engine dynos with chassis dynos here's the truth: Engine dynos main benefit is that they eliminate most of the variables inherent in engine power testing but they are still bound by the same problems of manufacturer software calculations and operator error. Regardless of what the article states, chassis dynos can be very accurate and very repeatable. Their strengths include both ease of operation and availability to the public. The main difference between engine and chassis dynos is that there are many variables that can affect the power output data on a chassis dyno that have nothing to do with the engine. These can be anything from vehicle specific items like gear oil weight and tire pressure to non-vehicular items like strap tension etc. In the end, it really depends on what you want to do with the dyno. If you want to test a particular tune, a chassis dyno is likely your only choice but even if you regard the chassis dyno as inaccurate for obtaining "absolute" power output data, it's just fine for "relative" data, and that's all you care about anyway, right? You might not know the exact power output but you know your tune picked up (or lost) 5hp, and that's a for-sure number. Alright, sorry for rambling on. Just wanted to try to clarify some points about dynos.
engine dynos are fine,, lets you know what the engine will actually put out, but the chassis dyno will let you know what a drive line will do attached to it.
The main take-away should be that ALL dynos utilize software calculations to form their data, and that the main difference between an engine dyno and a chassis dyno is that the engine dyno simply has far fewer variables that affect the data. One other key point is that although pretty much everyone zeros in on horsepower numbers, torque is actually a more accurate and reliable number to reference. After all, dynos measure torque, not horsepower. There is no way to actually measure horsepower. It's merely a calculation which is a measure of work performed over a specified timeframe.
If those common sense answers do not satisfy, go to hagerty.com/media/archived/horsepower and scroll to the stealth316 link. This will probably give most of us a headache. Like Katshot said, so much work over so much time and distance will equal something that we equate as "horsepower". The debate should and will continue along with the Ford vs. Chevy vs. Mopar with maybe an AMC SC/Rambler or an AMX 390 thrown in for good measure.