Don't agree with ya Joe
A short in a DCC signal looks a lot like 180 degree out of phase signals connected together. That's true. But I don't think the statement 'shorts in AC and DC are different animals' is true, at least not until you get up into megacycle frequencies where wiring capcitance and inductance do funny, non-intuitive things - and DCC is a low frequency signal.
I suspect - and this is a guess - that when running a DC loco, the power pack output may simply be too low for a short to draw enough current to be readily detected as a short circuit (and not just an open frame motor that can pull 1.5 amps at full throttle).
How does the circuitry know there's a short circuit? By the amount of current going through a frog. With DCC there's about 12V (and sometimes 16V) on the rails. A short circuit will mean the booster will put out max current (until the booster detects the short and shuts down - which takes between 20 milliseconds to 100 milliseconds depending on the brand and model of booster and how its configured.
Lets look at DC. Suppose the power pack is outputting 2 volts to the loco - dead slow speed. The power pack often has a large (physically because this resistor needs to disapate a lot of heat sometimes) potentiometer in series with the track wires. If the max voltage output of the power pack is 10 volts (more likely it's 12 or 16 but 10 makes for nice arithmetic), reducing the 10 volts down to 2 volts and 100 milliamps requires at resistance of R = volts / current = 8 volts / .1 = 80 ohms. That resistance is connected in series with the motors of the locos.
Now, presto the loco hits the backward powered frog and we got us a short circuit. The result is the current becomes I = E / R = 10v / 80 = 1/8 = .125 amps. That's not much! If the frog juicer thought that 1/8 amp indicated a short circuit it would be constantly flipping frog polarity and that wouldn't be very workable.
Now it's true that most DC power packs these days don't use a dropping resistor in series with the track to control voltage. But the transistor circuitry that regulates the output voltage/current behaves in a similar manner (well sort of). So if the throttle is set to notch 1 out of 10 and the outputs of the power pack are shorted you may not see a whole lot of current flowing.
DCC voltage is always applied to the rails at full-tilt-boogie and it's up to the locos to regulate the power to their own motors. When there's a short, current jumps through the roof often exceeding (momentarily at least) the 5 amps that boosters are commonly rated for. When the juicer (or auto-reverser) sees such current levels it can assume there's a short circuit (or something real nasty is happening) and its time to change the frog polarity.
So are juicers a good idea? Well aside from the cost the answer is 'perhaps'. They sure are convienient, no more trying to adjust micro switches on hand throws, no more worrying about Tortoise contacts getting burned out by over 1 amp going through 'em or needing to wire an external DPDT relay to a Tortoise to handle the sometimes high track currents when a short (or long loco consist) occurs.
But, when the loco wheels bridge the closure and frog rail gap there's a little arc (zap!). The size of this arc depends on voltage - you'll have trouble seeing it when room lights are on. You might be able to see it if the lights are off. These can, over time, pit loco wheels and turnout rails. It's best if the polarity shift happens with extreme rapidity. Under 10ms response really reduces the 'zap' time (at least in theory). which reduces pitting. Tim, if you're reading this perhaps you can comment on speed of reversal and wheel/track pitting? btw DCC autoreversers have this issue, too. There's a discusison of it over on the Tony's Train Exchange web site (at least the used to be) - sorry I don't have the URL of this page (but Tony's can be found through the logo buttons on the MRH website's 'Sponsoring Advertisers' area - use the tab in the row above the banner photos).
It turns out that designing an auto-reverser (or frog juicer) that operates with extreme speed, doesn't switch polarity for hallucinations, does switch polarity when appropriate and is robust enough to withstand the spikes and transients occuring on DCC powered track, is a non-trivial exercise.
I hope this helps (and that I haven't wandered out in the weeds of speculation to far with my answer)
Charlie