So it occurred to me as I was thinking about the G11 project one day that the 'salvo' approach to improved hit probability may be somewhat flawed. Obviously, it was sound enough for H&K to develop a weapon system to maturity, so I must be missing something, and I'm going to ignore the "just learn to shoot" argument for a moment, because that's a separate ramble.
For those not familiar, here's the abstract: use caseless ammunition to remove the extraction/ejection portion of the firing cycle and speed up the firing rate to 2000+ RPM, fast enough that the gun can fire three rounds before the shooter feels the recoil; in effect, the gun throws three rounds at the same place and kicks - the shooter would perceive it as a semi-auto. The problem is that all three rounds are going the same place, and they're not far apart; there's a very slight time lag between round 1 and round 3, but at 3000+ feet per second, not much at all (~.03 of a second x 3000 feet/second = ~90 feet between rounds, with a total salvo impact time of .09 of a second). The accuracy of the barrel determines the dispersion; a little is good, because if hit probability is to increase we want to have a larger 'beaten area,' but an inaccurate barrel is going to cause problems when attempting to shoot precisely instead of 'taking an average.' Say the barrel shoots 2 MOA; that's a 2-inch diameter beaten zone. Not much of an improvement on a single round. With .09 seconds between the first and last impacts, there wouldn't be much of an advantage against moving targets. I'm reminded of a 'spread of torpedoes,' a tactic whereby a submarine fires three torpedoes dispersed at a target ship; if the ship continues abreast to the submarine, two hits are guaranteed, and the dispersed torpedoes diminishes the effectiveness of evasive maneuvers. On a land battlefield, not much is going to move fast enough for .09 of a second to matter, so it would essentially be three hits simultaneously. Compare to an M16A2's three-round burst: you have a quarter-second between the first and last round, and those three are going to disperse more in close quarters, giving a larger, though less regular, beaten zone. The irregularity is the tricky bit; automatic small arms take some skill to use to their full efficiency.
So we have three rounds essentially going in the same hole, give or take a few centimeters at range. There would be an improvement in terminal effects - close hits break through body armor, which was the stated objective of the AN-94 Abakan's 2-round burst at 2000 RPM - but if that were the case, why not just use a bigger, traditional round and save the complication? Furthermore, while there is an option of 'suppressive' traditional automatic fire at 480 RPM vs the 3-round 'salvo' for improved terminal effects, the stated goal is improved hit probability.
Project SALVO, the US Army's 1960s vintage project with similar goals, produced a modicum of experimental cartridges, with duplex slugs and flechette rounds the main avenues explored. And this here round seems to present the most logical approach to improved hit probability from a non-human engineering (sights, ergonomics) perspective.
The cool feature of that round (6.35x53mm) is that the duplex load has a symmetrical first round and an asymmetrical base to the second round - it's slanted slightly, producing a first hit that strikes the point of aim and a second hit that lands somewhere nearby. In automatic fire, the rounds would be randomly oriented, so as to produce a pattern around the impact point over time. Since it's a traditional cartridge, a single load, with optimized ballistics, can be used for precision work. With a larger cartridge to push the heavier duplex load, the simplex target loads could be powerful enough for single-shot interdiction at extended ranges. Say a cartridge size a bit larger than 6.8 SPC - 115gr at 2650 fps for that cartridge; bump it up a bit, with two duplex rounds of 5.56 size (50-60 grains, so 100-120 total), at 5.56 velocities to ensure terminal fragmentation. I'm no ballistician, but the first round could be optimized for stability while the second is given a calibrated dispersion.
If that were scaled up to triplex or more, there would be a very significant improvement in hit probability at battlefield ranges; say, give it 3-8 MOA worth of wobble (in the games biz, there'd be a note scribbled in here: playtest this!). Package the lead in a weak jacket and throw it fast enough to fragment, a la M193 ball ammo, and it'd do well, at least in close quarters. Since your rate of fire is effectively doubled, the cyclic rate of the firearm can be backed off so as to make very controllable; since it doesn't need to run at warp speeds, something like the constant-recoil system used by the Ultimax 100 SAW or the AA-12 automatic shotgun would result in very dense saturation for a rifleman. Such firepower could be devastating in close quarters, where the shooter can turn his rifle into a submachine gun with merely a magazine swap. Build the system in a bullpup format and you have a winner.
The problem comes in when one needs to account for every projectile fired; this would definitely be a military expedient, not a civilian law enforcement one. It would only be legit for a free fire zone, unless the users policed simplex and multiplex rounds carefully. Also, this would improve the lowest common denominator at the disbenefit of the skilled operator; for a disciplined and experienced operator of automatic weapons, the extra dispersion could be an unnecessary distraction. I suppose the cyclic rate could be boosted and the system used with simplex rounds only in this instance.
I'm tired, so there's probably a hole in the reasoning. But effectively it's a 'shotgun' approach scaled to 'assault rifle' size.