I've tried to attack the problem of "fixing"
40k from a number of different angles over the past few years, and I've come to the conclusion that the status quo contains a large number of degenerate relationships that prevent any real fixes, most of which have become entrenched within the lore and within
GW's business model.
The basic problem is size creep; models get bigger, guns get bigger, "basic" models become irrelevant in the face of larger and fancier guns, which leaves us with a lot of false granularity (the difference between a "Space Marine statline" and a "veteran Space Marine statline" is +1A/+1Ld, which under the current rules is pretty much academic), trickle-down size creep (2W Space Marines shouldn't be necessary, but they are made so by size creep everywhere else), and because
GW would rather release a new model to "fix" the issues than fix older models we end up with list bloat where there are options that were relevant in older editions but aren't relevant now because size creep and rules changes have rendered them irrelevant.
I've started wondering recently if it would be possible to give the 9th Age treatment to 7e; instead of tearing up the rules and starting over go back to basics, clean up older rules, and build some of the
USR bloat into the core rules rather than requiring a giant appendix of "special rules" to handle it; the problem there is I don't have a benchmark for what "works" and "doesn't work" about 8e to build off of, so I'd better enumerate that to begin with.
And on that front I've spent so long waffling and trying to consider various properties of the game that I don't really have good answers, so at the end of the day I'm going to start needing to throw some ideas down on paper and hoping for the best. Arguments could be made either way for a lot of the following, but here's the position I'm working from:
--->Alternating Activations; In my mind this is a recipe for abuse if you don't have strong controls on the approximate cost of a "squad"; it doesn't solve the alpha-strike problem if you let superheavies exist, and unless you're willing to cut down on the length of the turn the number of priority passes that'd have to take place to alternate activations in five phases each turn (move/psychic/shoot/charge/fight) would make games take a lot longer than they do already.
--->Dice: I'm sticking with
d6s simply because I don't regard the amount of granularity
GW is trying to introduce by putting more and more models into the game as desirable. When the stack of Imperial infantry has to go Conscript->Guardsman->Guard Veteran->Stormtrooper/Battle Sister->Celestian->Space Marine Scout->Space Marine->Veteran Space Marine->Primaris Space Marine->Space Marine Terminator->Gravis Armour->Custodian->Warden->Custodian Terminator you might care about trying to make the stats more granular, but I regard trying to write a ladder of 13+ steps of "this model must be very slightly better than this other model" as a much bigger problem than the fact that
d6s don't simulate the ladder of 13+ steps very well.
--->Vehicle Facings: The point of vehicle facings is that I can put down a model that you can't kill because of how I've positioned it, but you can kill if you can get around it. It isn't a huge game-slowdown problem if you've got a few vehicles with basically rectangular shapes or with obvious bases; it becomes a huge game-slowdown problem if you have a lot of irregularly-shaped vehicles. For me the idea that "movement->shooting" aren't perfectly sequential (i.e. my Flyer doesn't "move 20", then shoot", it shoots something somewhere in the middle of its movement) is enough to justify vehicles shooting "outside of arc" (the line of sight requirement represents the fact that the "turn order" isn't something obeyed in the chaos of the battlefield; a vehicle that has to poke part of its nose outside of cover to take a shot is actually driving out a short distance, shooting, and reversing back into cover (anyone who's played World of Tanks will be familiar with the concept), but the enemy has an opportunity to make a shot while they're doing that, even if the vehicle has "enough movement" to make it back out of line of sight in "one turn"

. I can't remember who but someone suggested to me in another thread that 180-degree "front"/"back" arcs might be easier to measure and create a similar effect, I'm leaning towards doing something with that suggestion at the moment. I also find that the vehicle degradation tables mostly translate to a lot of looking things up and don't matter all that much because vehicles tend to either die all the way or have stratagems that make them fight at top profile, so I'm thinking of a universal subsystem-damage table instead.
--->8e numbers v. 7e numbers v. Other numbers: 8e and 3e-7e both run on three-rolls-to-kill, and seem to require high fire rates to do anything as a result. I don't mind fixed to-hit rolls out of 8e simply because the comparison table for
WS in 7e mattered so little; it's relevant in 30k because all the Space Marines have very similar
WS so nudging it a point has repercussions more often than not, but shift the game to include fewer Space Marines and more Guard/Xenos and giving
WS as a fixed number is faster/easier than checking every fight to make sure your
WS is still higher than the other guy's. I prefer the to-wound table from 7e because it just comes out and tells you "stop firing lasguns at tanks, you can't hurt them" instead of the "go ahead and roll two hundred dice against my airplane at 1/216 to do a wound, it'll be fun" of 8e. Armour saves are problematic for similar reasons; in 7e it was nigh-impossible to price 2+ armour/AP2 unfairly because it was 6x as effective or 1/6th as effective depending on matchup, so a "fair price" just made 2+ armour/AP2 wildly overpriced half the time and wildly underpriced the other half of the time. 8e has a similar problem where the value of
AP/armour swings wildly depending on your matchup. Three-rolls-to-kill has produced some fundamentally degenerate things; you shouldn't need to fire 3-4 shots per pulse rifle for a pulse rifle to be relevant, but in an environment when you might be shooting at un-Markerlighted Primaris Marines in cover and need 36 shots to drop one 17pt model you sort of have to. On the other hand two-rolls-to-kill in Lord of the Rings feels frustrating because you spend so much of the game fishing for 5+ on a small number of attacks and you need the "trapped" conditional effect that lets you swing twice with everyone to ever kill much. My current thinking is sticking to 8e to-hit and save-mod mechanics, but rearranging durability to try and produce something more like a 7e meta where one shot killed more things and you didn't fire as many shots.
--->Blast weapons: One of the things I've been grumbling about for a very long time is uniform blast damage; when
GW set out to make a blast weapon that's also a semi-functional anti-tank gun in earlier editions they had to give it AP2 against all targets hit, which made it way too strong at killing heavy infantry. The thing that really took Terminators off the battlefield in earlier editions was the proliferation of AP2 large blasts; you Deep Strike, everyone's all clustered together, a Riptide ion cannon blast lands in your face and they all need to make their 5++ or die. Blast templates themselves are fun but if you're trying to be precise/careful with them they're a huge source of game slowdown; I'd rather see something closer to how Bolt Action does "blasts" where they do a random number of hits but only ever do one hit to armoured targets in an effort to make anti-armour and anti-infantry weapons actually different instead of keeping the "let's fire
2d6 d3-damage shots at any target heavier than a Guardsman!" meta going. I really like the conversion beamer on the Contemptor Dreadnaught in 8e; it shoots one very powerful shot, then if that kills a model it does a large random number of hits at a much lower S/
AP/D, thereby simulating non-uniform blast damage (your explosion may kill the Terminator caught in the middle, but pelting the rest of the unit with shrapnel isn't going to do a whole lot) and making it function differently from a gatling gun making a large number of uniform-damage shots, so I'm thinking that's a good model to extend to other "blast" weapons.
--->Army lists: I'd like to move to a 30k-esque model of one "primary detachment"/one "allied detachment" with "Rites of War" that modify which units live in which slots and grant detachment-wide bonuses/penalties to everyone to simulate strange and unusual armies; more structured army-building lets me fiddle more precisely with things for balance purposes without straight-up saying "no, you can't do that." I've complained endlessly about skew in other contexts where taking only units vulnerable to one type of weapon renders armies overly-strong against all-comers lists that don't have enough of the hard-counter; with
40k's loose detachment structure there isn't a lot you can do about it, but the way 30k does it I can say things like "sure, take Russes in Troops, but the ones in Troops have more restricted equipment and the enemy has an extra secondary objective and your non-Russ choices are restricted in this other way." They'd probably end up looking more like Theme Forces in Warmachine (2-6 per faction depending on size of faction) rather than the 15 generic and 2 Legion-specific per Legion in 30k and would be a lot of work to write, but it's a way I can reconcile my desire to avoid skew armies with other peoples' desire to play skew armies without saying "Nope, only T4- 1W models in Troops, you must all take infantry to take objectives." At least I think I can balance them that way; I don't see Armoured Breakthrough (Predators in Troops) or Fury of the Ancients (Dreadnaughts in Troops) winning everything in 30k.
--->Command Points: I like the idea, but I dislike the implementation; there are a lot of things that feel like they ought to be a rule on a unit that
GW implements as Stratagems (flakk missiles, for instance). One of the things I actually like quite a lot about v2 Age of Sigmar is "command points" generated per turn used as a constraint on special abilities; if I tie
CP/turn to what characters you've got left on the table it makes sniping characters more interesting; I'd also like to try and cut down on
GW's insistence that there be stratagems attached to specific units/weapons since it feels like that produces a lot of unnecessary stratagems you don't need that you have to dig through to get to the important ones.
CP/turn also feels like it'd make the 'one use of each stratagem per turn' bit less necessary if you can't blow all your
CP for the game doing something massive and dumb on one turn.
--->Psykers: I have a deep and abiding hatred for one cast attempt per power per turn; it feels like it encourages writers to make individual powers too good since they can always say "oh, don't worry, you only get one cast/turn," and it discourages taking multiple psykers since you get diminishing use out of them for each one you take, which screws over armies that have a lot of psykers. My current thinking on psykers is to kill the "psychic phase," make any "psychic power" that isn't an attack function like a stratagem, and define the ones that are attacks as a weapon with the "psychic" tag, then making them all harder to use depending on nearby potential "denial" sources and riskier to use depending on how powerful they are.
--->Types: One of the major sources of bloat in 7e since by then there were eight non-vehicle types with their own page of rules and twelve types of vehicles with their own page of rules, plus eight weapon types that had to define how many shots you got and whether you could move and fire them based on what type of unit they're attached to. One of the great things 8e did is make all those types "keywords" that didn't do anything on their own but could be referenced by other things, but then they promptly shot that idea in the foot by taking rules that would be more easily done by attaching them to a type (Explodes, Supersonic) and writing out the full paragraph on every datasheet with minor variations, requiring you to dig through a paragraph of identical text to figure out if one or two variables are slightly different. I'm currently thinking that anything that's a paragraph of identical rule should be written as "this is what this type does" and stuck into the core rules to avoid needing to scan through paragraphs of text to make sure there are no slight variations in every single unit's version of the rule with the same name.
I've done a lot of theorizing/philosophizing on most of these points over the years and always run into hurdles that send me back to the drawing board when I try and write up a test version; going to try and actually forge ahead and post something this time around. Expect more later today, expect every model to have a use but don't expect every profile to survive intact, and don't expect to see every army book in the near future, I know most of the Imperials well enough to do a lot of their stuff off the top of my head but I can't do Orks/Tyranids/Necrons without a lot more reading and consultation with people who actually know them.