The sacrificial anode does indeed have to be in the same solution as the metal it is protecting. And the solution must not be overly hindered by current decreasing effects like distance and poor conductivity of the solution. Another consideration is anode area versus cathode area. A small sacrificial anode ( some Mg ) does little to protect a big cathode, your car.
 The best Detroit has come up with is the electrogalvanize coating on the steel all over. The zinc applied this way is a truly effective sacrificial anode because it is in intimate, pervasive contact with the steel it's protecting.

Mcguire,

I know the anode has to be in the same solution with the protected metal.  What I meant is that the conductive glue represents the wire connecting anode and metal object in a circuit.  This, as opposed to using non-conductive glue and relying on the salt water on the car to carry all the currents for all reactions between Mg and iron.  It would be nice to use ordinary non-conductive glue but I think I need it.

I know a small piece of Mg will do little to protect the whole car, what I'm hoping for is a more localized effect.  If any protection was happening the Mg should corrode very fast because of its small size, instead most pieces have lasted many years (maybe 10!)  I can't explain this.

About the electrical gizmo, it has a measurable voltage wrt the body of the car where it is glued (conductive) to the car's undercoating or paint.  Where it got glued to bare metal it seems to short out.  I can't explain why this needs to be insulated from the body and the Mg doesn't.  They seem to be working differently from each other.

One final question, as I said my car is undercoated, can I apply a zinc spray paint over the undercoating and rusted areas both?  Will one area work and not the other?  Maybe the thin zinc layer would mostly corrode away in one season but I wouldn't mind reapplying it in key areas each year.  I realize it's not a simple "circuit", and there will be areas electrically isolated from others by missing zinc or discontinuous salt water layer.  I'd like to better understand the basics of what's happening.  Thanks.

The Mg sacrificial electrode must remain in electrical contact with the body. That can be difficult with sacrificial electrodes because, when they work, they become covered with corrosion products just like a battery terminal. The contact point itself should be isolated from being corroded. Drill and bolt, then protect the junction woth conductive glue or some similar scheme.
 The gizmo is doing exactly the same thing as the sacrificial anode is trying to, but the contact is more secure. The fact that the Mg anodes are still there says they aren't functionning sacrificially. Either they aren't in the solution or they are electrically isolated.
 The gizmo produces it's own voltage. it doesn't rely on being dissolved by the solution to provide electrons. That's why the puzzling difference.
 And, as you cite, zinc-rich paint over undercoat doesn't function as a sacrificial anode as galvanize coatings do.
 Good luck with those chicago winters. I know I got too old for them. Now I worry about the effects of warm salt water on my cars.

Just remember with cars there are many different parts that are all insulated from each other be rubber seals and grommets, grease and plastic components. You just can'tr get a good electrical connection for them all. You also suffer from all the small difficult areas that are difficult to reach, which are perfect for crevice corrosion to occur.

The whole concept of cahodically protecting the bodywork of a car with sacrificial or applied current systems IS NUTTY and fraudulent!!!  
If you wanted to use CP, you would have to put your car in a swimming pool or some other electrolyte of enough volume and conductivity that the protection can work, because the anode effect you are seeking is limited to what can be conducted in a thin film of electrolyte - which is VERY LITTLE INDEED.  This is why the only way to do this is to have the sacrificial metal everywhere, as is accomplished with zn galvanized sheet.  I know this isn't what you want to hear, but you are trying to do something that is not feasible as your experience clearly shows!

I was hoping to splice into the lines for the electrical gizmo and run wires to key rust prone areas like behind the wheel wells.  I got enough conductive rubber so far to make 4-6 additional protection spots.  The conductive rubber and glue are quite expensive so I don't want to hook the electrodes up wrong.  It sounds good in theory and I got the kit pretty cheap so I figured it was worth a try.

Do I remember the mftrs of autos standardizing on "Negative Earth" to reduce the risk of corrosion ?  Buying an auto which is galvanized halfway up the bodywork seems to be the best solution - otherwise, clean the vehicle once a week, especially under the wings.

I'm more of a "clean it once a year" type of person.  I'm always looking for a way to do less work.

I have finally stopped laughing and have returned to the chair from which I fell off.  Seriously, you cannot stop rust, because there is already such a state of oxidation by the time your new car has been shipped to the dealer, that you would need to do a "frame up" on it.  Manufacturers, in their quest for $$$, really don't do the careful measures required prior to galvanizing, drawing, trimming, welding, post-welding, pre-finishing, etc etc etc in order to kill rust.  You might have marginal luck with a caustic wash-down and then use zinc-rich paint.  I have never thought of magnesium as a very good sacrificial metal, zinc is much better.  Good luck!

Magnesium is a very good sacrificial metal, one of the best you can use that won't ignite on contact with water.  The reason I'm experimenting with magnesium is that I have a big chunk of it but only have zinc spray paint.  It looks like the best places to put either are where it gets plenty wet, like around the wheelwells.  A couple pieces of magnesium were placed near, but not touching, the conductive rubber electric anode and they corroded away quite fast.  I believe the proximity to the electrical anode accelerated its corrosion when the car got wet.  This would indicate that the current from the electrical anodes is having an effect.

To work, the sacrifical metal (whether aluminum, zinc, or magnesium) needs to be well distributed over the surface to be protected and electrically connected to it.  Otherwise the water film would be too highly resistant for sufficient protective current to reach all of the meatl parts to be protected.
This is what plating does in the electrogalvanized and hot dip galvanized materials.

Barry,
Thanks for that tip.  I guess I'll concentrate on applying a good layer of zinc paint to all vulnerable areas of the underbody.