The application note is based on the physics of electrons which still work the same today as they did when the article was written.
I am in agreement that proper equipment grounding and bonding are extremely important, but in practice distant connection points that may have been exactly at the same potential on day one of the equipment's life will experience some corrosion or oxidation and connection resistances will or could change over time due to this. By connecting both ends of the shield you are increasing the probability of a ground loop. If you provide a path for current to flow, it will, if anything changes in the potential from point A to point B. If you do not provide a current path then no current will be able to flow in the shield thus preventing any induction of noise on the cables signal conductors. These currents and induced noise are tiny but consider if this cable was being used to connect a Penning type vacuum sensor to its remote amplifier (common in the Titanium vacuum casting industry) the signal produced by this sensor is in the micro amp range and easily skewed by small amounts of noise or think of a 4-20 mA current loop out put from a laser distance measuring system who's 16 mA difference is 1,000,000.0 microns (a meter), that's 16 nano-amps per micron so to measure a common machining tolerance of 0.1 mm (100 microns) a signal change of 1.6 micro amps needs to be discernable, not happening if noise is present in the cabling. In digital signaling where voltage levels determine state and have pretty good noise margins (0.7 volts for TTL and higher for CMOS) the small amount of signal noise is unlikely to have any effect, but in the small signal analog world it can be a deal breaker. Read the app note; just as Einstein's explanation of gravity from 1915 is still the same today as it was then, so to is this article.
I am in agreement that proper equipment grounding and bonding are extremely important, but in practice distant connection points that may have been exactly at the same potential on day one of the equipment's life will experience some corrosion or oxidation and connection resistances will or could change over time due to this. By connecting both ends of the shield you are increasing the probability of a ground loop. If you provide a path for current to flow, it will, if anything changes in the potential from point A to point B. If you do not provide a current path then no current will be able to flow in the shield thus preventing any induction of noise on the cables signal conductors. These currents and induced noise are tiny but consider if this cable was being used to connect a Penning type vacuum sensor to its remote amplifier (common in the Titanium vacuum casting industry) the signal produced by this sensor is in the micro amp range and easily skewed by small amounts of noise or think of a 4-20 mA current loop out put from a laser distance measuring system who's 16 mA difference is 1,000,000.0 microns (a meter), that's 16 nano-amps per micron so to measure a common machining tolerance of 0.1 mm (100 microns) a signal change of 1.6 micro amps needs to be discernable, not happening if noise is present in the cabling. In digital signaling where voltage levels determine state and have pretty good noise margins (0.7 volts for TTL and higher for CMOS) the small amount of signal noise is unlikely to have any effect, but in the small signal analog world it can be a deal breaker. Read the app note; just as Einstein's explanation of gravity from 1915 is still the same today as it was then, so to is this article.
