What Die Type Should Be Used?
This question is, of course, system-dependent. For The One Ring (TOR) the Feat Die (the special d12) was the obvious choice, and the precedent for the mechanic already exists in a published adventure. A d20 might be appropriate in d20 systems, though I chose to instead opt for 2d6 with 13th Age (13A). As it turns out, I have some further ideas for modifying the ally check which become more interesting with 2d6 compared with a d20, which I'll discuss later. But what about a system with non-standard dice, like Edge of the Empire (EotE)?
The similarities between TOR's Feat Die and the Force, Challenge and Proficiency dice of EotE made using the Force Die or the red/yellow d12s pretty tempting. Theoretically the blank face and the Triumph/Despair could stand in for the Eye of Sauron/Gandalf Rune, with Success/Failure and Advantage/Threat standing in for good or bad results (faces with both being neutral). On the Force Die the applications of light/dark side pips should be pretty easy to guess. The ultimate problem with such a system is that the results of the roll aren't easily or intuitively as modifiable as a numerical system. This presents a couple of big problems, the first being that you can't "custom fit" the probabilities to account for the difficulty of the odds your allies are facing, and second being that bonuses or penalties (due to PC intervention, terrain, narrative changes, etc.) are tough to apply. It's easy to just say "use a regular d12, or 2d6 just like the other examples!" but ideally you'd use a die or dice that are already used in the game, since not all players can be expected to have a set of standard RPG dice (ok, ok, so most probably have a couple of d6s from board games or whatever). Thus the obvious choice becomes the d% dice.
You'd basically treat the Ally Roll like you would a roll on the critical injury table (again, you can pre-make a custom table or just wing it). Off the top of my head, I'd probably break the results down into 1-5 very bad (Eye of Sauron), 6-40 bad, 41-70 neutral, 71-95 good, and 96-100 very good (Gandalf Rune). The odds are stacked against the allies with the bad range having a 35% chance of being rolled, neutral a 30% chance, and good 25%. The percentile roll provides a lot of granularity for adjusting based on the difficulty/conditions of the battle, and of course you need to leave PCs the opportunity to be the real heroes by making it an uphill fight.
Oh, and while we're at it I'll note that in the 2d6 system I employed for 13th Age the equivalents to an Eye or Gandalf would be double 1's or double 6's, respectively. While this makes these results less likely to trigger than in the other examples, it also has the advantage of simplicity. GMs can get around the reduced probabilities by interpreting those results with a heavier hand, but the probabilities are also a little more malleable when the roll is modified (see the "Game Specifics" section).
Allowing for Player Interaction
There can potentially be problems associated with running different sections of the battlefield with differing degrees of abstraction. On the one hand, the PCs and their immediate foes will be operating on the scale of individual actions, individual Hit Points/Endurance/Wounds/etc., and individual initiative slots. The allies all get lumped into a single roll which represents how both sides are faring in relation to each other, and the results are more often than not going to be much more narrative with much less bookkeeping. While you can add some bookkeeping (see the unit strength idea that I presented in the original 13th Age post, which was the second link at the top of this article), my personal preference is to reduce that level of tracking and instead focus on the broad needs of the plot. The PCs shouldn't care about too much detail, because in-game they have their own problems, and so they'll be perceiving their allies in broad strokes (besides, this keeps the game moving more quickly, with less time spent on the GM "fighting themselves").
But what happens when a PC wants to interfere? This is best handled by a modification of the Ally Roll. Here are 4 distinct examples.
- A PC kills the foe he was fighting, and wants to effectively "join" the ally group. Simple enough, just pick out a single opponent or two (or a mook/minion group, if the game supports that), pull them out of the ally roll and into a new personal combat with the PC, using PC scale actions, HP, etc. If it makes sense that the enemy would already be a little wounded, go with whatever feels right. To represent the loss of some enemy forces, you could grant a bonus to the Ally Roll (magnitude dependent on the strength of the foes that the PC "drew out").
- A PC blankets the enemy forces with an area effect (grenade, fireball spell, etc.). Depending on the size of the enemy group and their durability, this might result in bonuses to the Ally Roll, or perhaps even an automatic positive result. This is most likely the case if your positive result reads that an enemy/enemies are wounded or slain, and it makes sense that the area attack would have wounded/killed some enemies. If the PC spends his entire turn and what is probably a limited resource (grenade or daily-use spell), yeah the effect on the Ally Roll should probably be pretty huge.
- A PC wants to use a skill to bolster the allied forces. An obvious choice would be Leadership (EotE), Inspire (TOR), or a Cha check with applicable background (13A). Again, if the PC is using their action to do this then the results should be pretty big. For some systems that call is automatic (for example, in EotE using a skill is almost always an action), but for others it'll be up to GM discretion (13A and TOR). GMs might even want to give players a choice of doing things quickly and getting a simple bonus, or spending their action/turn and getting a more significant bonus (a bigger numberical bonus, or perhaps even an automatic "good stuff" result depending on the nature of the action).
- A PC wants to command a specific ally to act upon something outside of their own conflict. For example, commanding an allied archer to shoot the enemy Wizard, or ordering a rebel soldier to fire at the Moff at the computer terminal. A successful check should allow that ally to be "pulled out" of the ally group for a turn or more, letting them act as an individual at the same scale as the PCs (this is basically example 1 in reverse). The price you'll pay for this is that you'll probably add a penalty to the Ally Roll, since without that individual's aid the other allies will have a tougher time of it. Perhaps with a really good roll (Great/Extraordinary success in TOR, Advantage/Triumph in EotE, or a Natural 16+ in 13A) this penalty can be reduced or eliminated.
Ok, so with the general principle and some examples laid out, let's provide a summary of some baseline mechanics that a GM can draw upon. Keep in mind that this stuff is still mostly theoretical, as I've minimally playtested it with 13th Age and The One Ring. Not only that, but because there is so much variance in specific conditions this is designed to be modifiable and somewhat improvisational by the GM.
The One Ring - You'll be using the Feat Die for the Ally Roll, usually with a spread of Eye (really bad), 1-4 (bad), 5-7 (neutral), 8-10 (good), and Gandalf (really good). Bonuses/Penalties to the Ally Roll should generally be between 1 and 4, with +/-2 and +/-4 being good starting points thanks to the Complications Table (pg. 48 LMB, GM Screen). Keep in mind degrees of success, with a simple success on any PC's skill check to interfere providing a modest bonus, and Great or Extraordinary successes have better results.
13th Age - Use 2d6 for the Ally roll, with 6 or less being a bad roll, 7-9 as neutral, and 10 or more being a good result. Optionally, double 1's can be very bad, and double 6's very good. While simple numerical modifiers certainly work for bonuses or penalties (I'd use the range allowable by Backgrounds: 1-5), I really like the idea of successful skill checks adding a die to the pool to provide a functional "re-roll." In other words, roll 3d6 and keep the highest 2 values. Severe penalties could go the opposite way (roll 3d6 and keep the lowest 2 values). This boosts the low probabilities of the double 1's/6's for very bad/good results, and in my opinion is a more satisfying impact for players than a straight modifier (a modifier can't save a horrific roll, but a re-roll very well could). Optionally, a 16+ on the PC's roll can add the extra d6 in addition to a numerical modifier. The base dice pool of 2d6 works really well here because re-rolling the entire Ally Roll comes across as a bit too powerful to me.
Edge of the Empire - Using a d% for the Ally Roll this system potentially offers the most granularity, but realistically things should probably be kept to intervals of 5%. As I stated before, the baseline breakdown might be as follows: 1-5 (very bad), 6-40 (bad), 41-70 (neutral), 71-95 (good), and 96-100 (very good). A bonus of 10% per success on a PC's skill roll seems about right, though with checks that have a goal other than simply boosting the Ally Roll, allowing Advantage to be spent to also boost the Ally Roll by 5% per Advantage seems reasonable. Additionally, keep in mind that with EotE the minion rules already provide an easy way to adjudicate battles between small groups of allies and enemies, and so GMs should consider carefully when using these Ally Rolls is really appropriate. Probably only with large squads that would result in unwieldy dice pools if treated as minions, or if the GM wants the combat to run especially quickly and doesn't want to deal with actual dice pools and minion tracking.
Miscellaneous Note on Cumulative Bonuses
Sometimes it makes sense for a bonus to an Ally Roll to last more than a single round. For example, if you give the roll a penalty for one of the ally combatant being wounded or killed, then that penalty should probably stick around until the playing field is even with an enemy being wounded or killed. If the Ally Roll gets a bonus because the enemy leader was taken down, that drop in enemy morale should probably last for the entire battle.
Some player actions will result in advantages of variable duration. If a PC rolled to inspire his allies, how long does that bonus to the Ally Roll last? It could be just 1 round, it could be longer for a higher degree of success. Maybe it simply lasts until a low roll comes up (bad or very bad) and the allies are demoralized again. Go with what makes sense. Some "tactical" rolls where a PC tries to direct allies to a better fighting position should obviously last as long as the allies hold that position. You might even alter the outcome of a roll (on the fly, even if you're using a pre-made table) so that a bad result, which would normally wound or kill an ally, instead drives the group off of this position, nullifying the bonus.
Ultimately, one last thing to keep in mind is to go with what makes sense in the moment. This house rule isn't designed to be an in-depth new mechanic tacked onto the game, and doesn't need to strike a perfect balance. Rather, it's designed to provide some structure for handling groups of allied combatants narratively. It's for when "making it up as you go along" feels too arbitrary. It's there as a baseline, to get your creative juices flowing, and to basically provide a status update on how those other guys are doing, especially during rounds where the PCs aren't direction involved with that side of the battle. If you use it in a game, let me know how it goes.