In order to qualify directly whether a coil is "better" than the OEM coil, you need for it to satisfy two conditions. It needs to store more energy at full charge and it needs to achieve full charge as fast or faster than the OEM coils in order to respond in the time allowed. Now, the current achieved over infinite charge time is just I=V/R. And the energy stored by that current is E=(1/2)L I^2 = (1/2) L (V/R) ^2 = (1/2) V ^2 * (L / R^2)... that is, the energy rises with L but falls off with R^2. Now, to ensure that the charge time be similar to the OEM coils, we have to define T = L/R. That is the characteristic time for a RL circuit to rise to 70% of it's max possible current. So, keeping R/L roughly the same as stock, I can explore what happens of their either BOTH bigger than stock or BOTH smaller. Rewrite the energy to keep T = L/R fixed and we have, E = (1/2) V^2 T/R. That is, keeping T fixed the energy still depends on 1/R. So, if you want to store MORE energy and keep the same charging time, you should DECREASE the resistance AND the inductance while keeping the ratio of R/L approximately fixed.
SO, the stock coils have R=0.8 Ohm and L=3.9mH which puts T=5ms. If you want to replace them with BETTER coils then you'll want to look for coils which have BOTH R and L smaller than these values while maintaining T = L/R ~ 5 ms. The intrepid coils, unfortunately, do not do this. They come in at R=0.8Ohm and L=2.9mH when wired in series. That makes the value of T slightly shorter (which is ok) but also lowers the stored energy (which is less awesome). Hense the doubling of the igniters which should lower the system resistance and bring T back into line as well as raising the stored energy.
And remember, resistance and inductance BOTH add normally in series and reciprocally when in parallel. So, if you're looking at a set of four coils you have to figure out whether to put them in series or parallel in order to get them to duplicate the values of the two OEM coils.
As for the question of the spark going the wrong way over the plug. Unless you're dealing with an inherently directional material, such as a diode or a pnp junction, all of electricity, including currents, fields, potentials, plasma currents, etc, is invariant under interchange of sign. There is NO reason to expect that it should be harder to ionize the spark gap from the ground strap to the electrode than from the electrode to the ground strap.
To see that this MUST be the case, inspect the diagram for the OEM coils. Each coil only has ONE secondary coil and it connects to both spark plugs (one on each end). When it discharges the current flows down through on plug and up through the other to complete the path. The plugs themselves couldn't care less which direction the charge is flowing. In fact, I suspect that the discharge current is actually A/C and not D/C given that the oscillating field necessary to induce the breakdown voltage would have to be EXTREMELY FAST in order to build up to the necessary strength. That means that in the time it takes to discharge the spark the current over a single plug is likely switching direction over and over, making polarity an inapplicable concept.
If you say that you had issues with wiring one of the coils backwards then I SUSPECT that it was a wiring problem. Notice that the Intrepid coils on have three available terminals. For simplicity lets call them T1, T2 and SP. Normally you hook power to T1, ground to T2, and the spark plug to SP. When the secondary fires T2 is disconnected so the current must be across T1 and SP. That means that T2->SP must be an impossible path or else we'd have had a short over T1/T2 during the charge cycle. Now, if you wire the coil backwards then it charges fine (in reverse) but the discharge goes wrong. That's because you're switching T1 to open at the point of discharge. But SP is attempting to pull current from T1 which is not connected to anything and thus kills the discharge event.
For my proposed variation to the wiring, the above problem is solved by hooking T1 from two different coils together. The first coil in the series charges backwards and the second coil charges normally. Then, at discharge, the T2 of coil two is not an open channel since it goes to the igniter ground which is open at this point. And the T2 from coil 1 is also not in the current loop since there is no SP -> T2 connection in the coils. The first coil tries to push from SP to T1 and the second coil pulls from T1 to SP. It's like a snake eating it's own tail and it makes a local current loop, feeding down over the first plug, through the head, up the second plug, through coil 2, across the connected T1 ports, and down coil 1 into the first plug again...
Now this is all speculation and I haven't wired it up yet. When I do, I suspect that I won't notice any difference. The proof, if there is any, will be that it works just as well as the other way and the benefit will be unnoticeable. But it will, in principle, reduce the current demand of the ignition circuit just a little and that'll please my sensibility.
