Systematic Lobster

[PhoneLobster]

OK. So 4th ed pisses me off. Already.

3rd ed. Well the only reason I haven't already stopped playing it already is I wanted to give the tome material a work out.

Everything else is also dumb.

So I am wandering back into the dark wilderness of pure home brew.

So I'm making an attempt at converting my existing "its broken all to hell and I'm pretending that's just cool flavour" home brew system into something remotely balanced that I can use when I'm not running my "cool flavour" home brew by the seat of its unbalanced pants.

But I'm making some assumptions, maybe some big ones and I want to hear some opinions and analysis from anyone who could be arsed.

Big Core Assumption
I got this core mechanic for attacks that I like. Perhaps too much.

It goes a bit like this.

1d20+some kinda attack bonus= your attack result

10+some kinda defence bonus= your targets defence.

If your attack result is equal to or greater than the defence then it is "successful".

If it is greater by a certain margin it is a critical hit, and removes your target from combat.

If it does not reach that margin but is a success by some smaller margin that margin becomes a penalty to the target's defence in future.

Simple enough. I think its a mechanic that can work. (though any discussion of its implications might be of interest).

The problem probably comes in some of the particulars of the EXACT implementation I'm currently considering.

Them thar particulars, part 1, Numbers and gaps and extended stabbing duration
OK. So in my crazy system it used to be that success by 10 cops you the big critical hit and variance between poorly corralled bonuses could range anywhere up to something stupid like +11 or rarely +13, but most of the time should be more like a +3 probably in favour of the defender.

And that made for a pretty harmful fairly swift results to combats. And that was OK because it was in keeping with the "cool flavour" of high mortality rates and higher come back from the dead stronger and crazier than ever rates.

But see I'm looking to move away from the spare lives mentality of that setting for this project, not that it won't have coming back from the dead junk, just that I don't want it to happen to at least one PC EVERY fight this time round.

So I thinks to myself. OK. So the new system will have a success margin of 20 to make a critical hit.

Also it will have an iron fisted reign of various capped bonus stupidity.

So the new system can see some pretty large MAXIMUM margins between characters, between levels and other junk adding in you could have like JILLIONS of points favouring either attack or defence.

I mean seriously If you are fighting someone at the wrong end of your attribute specialty, who is the lowest level considered appropriate for you to face, and they are stark naked without any skills or enhancements, and you are fully equipped using your skills to your best benefit with every viable enhancement under the sun then you will have a margin of like +20 in your favour on attacks and will kill them real fast.

BUT. The assumed "standard" encounter says hey, you are the same level as your target and everyone should find it RELATIVELY easy to max out all their other bonuses at the hard caps set for each bonus type.

Leaving you with a situation where if everyone maxes out everything the defensive side gets a 10 point advantage (because the defensive side gets 10 more points of cap from two additional named bonus types).

So that the idea then is that an average hit in that situation sees a small margin of success and something like a 15+ on the roll will take a large chunk out of future defence.

With more than one attack expected to be coming each targets way each turn (see later details for some limits on how that works) within perhaps 3 to 4 turns in I expect to see a critical and someone drops out.

But. Thats the assumption if everyone just maxes everything out.

I'm assuming also a potential ten point variance either way as well based on the variation I expect and hope to introduce in some of the bonuses with situational junk and various minor trade offs.

So favouring the defence side that would see a situation where only a roll of a 20 would be a success on the attack at all. Having a ruling that a 20 is always a success and always deals at least 2 points of damage penalty means its not a total farce and that if you can totally get the best reasonable drop on your otherwise equal opponent they can hurt you, but only just, and it may take quite a few turns to even be in danger of being knocked out.

With the ten points swinging in favour of the offensive side if you manage to pull the best reasonable drop on your otherwise equal opponent you have a 1 in 20 chance of knocking them out first turn and will likely drop them by turn 2.

Then you can throw in additional variation based on adding a "level" bonus directly to everything and having the potential of fighting higher or lower level opponents and you can wade naked through a sea of negative leveled peasants killing them only with your little finger, or alternately fear the little finger of the mighty level 10+ GM who is compensating for something NPC.

Now I feel supporting that sort of level stupidity may be unwise. But I went ahead and did it with the proviso that really you should keep a close eye on that and its probably best to keep encounters within a range of 5 levels of the PCs leaning toward equal or just a few levels lower.

Them thar particulars, part 2, Other particulars
There are several other mechanics that play directly into this core mechanic that I am concerned about for one reason or another.

1) The attributes
I'm having two main combat attributes. One is "Fast", the other is "Hard".

The idea is Fast represents being agile and accurate of body or mind.

And Hard represents being tough and powerful, of body or mind.

So Muscle Face the barbarian and Omnipotence the Wizard are both Hard characters while Acrobatica the Assassin and the Sly Fox Demon Voodoo Witch are Fast characters.

Anyway. Any attack can be Fast or Hard. The attacker chooses when they make their attack. Fast or Hard attributes of the respective parties are then added directly to the attack and defence of the respective parties.

Fast is both the attack and defence attribute for Fast attacks and Hard is both the attack and defence attribute for Hard attacks.

You set your Fast and Hard attributes to between 0 and 5 at character creation (five points to distribute between two attributes with 0 and 5 being the minimum and maximum possible values for each) and there they stay forever.

I have two grand hopes about the Fast and Hard attributes.

I hope that they are basically fairly balanced. And they should be as far as I can puzzle out, or vaguely remember puzzling out ages ago, without committing myself to the harder math that I have blocked all memory of after my career change into horticulture.

I also hope that they will be all the raw numeric no specific special power attached differentiation that players need to represent the combat attributes of their characters and that people feel the difference between a 5 Fast, 0 Hard and a 2 Fast, 3 Hard character means something to them on some sort of vaguely descriptive level.

2) One hit rule
With attacks rapidly weakening targets against future attacks being able to attack the same target multiple times really really hurts.

So getting extra attacks on the same target is one thing the system isn't going to be giving out too easily.

But that isn't enough, you got 4 dudes fighting four dudes even with one attack each if all four on each side target the same guy each turn you are looking at losing a character turn 1 perhaps a little too often.

So I got this here 1 hit rule. And it ties in somewhat with my zany simultaneous resolution thing, which I hope not to go into right now, but maybe later.

So anyway each turn only ONE attack can deal either a critical hit or a damage penalty to a targets defence. If multiple successful attacks are made against a single target only the MOST successful has an effect.

To add to this mechanic (in fact to make it legitimate at all) and to make sense in a wider simultaneous resolution er, thing, attacks that deal a damage penalty to a targets defence do not apply that penalty to the targets defence until the END of the turn, so the full turns worth of simultaneous attacks competing to be the best attack of the turn do not benefit from the damage penalty of the prior contender for winning the single hit rule.

This means, I hope, that there are good reasons to gang up on a target, especially if your sucky side kicks are missing or landing mediocre hits. But that there are also very good reasons to spread the loving around, especially if your competent companions are dishing it out with the big round soup ladle.

3) Sneaking up on some fool
OK, so you sneak up on some fool and steal his lungs and sell them on the black market before he even sees you, with an axe.

I want this to be a pretty big fat bonus. Back in the old crazy system it was like a +5, its one thing I really haven't thought hard on for converting to the new scale yet, I'm toying with what amounts to a +10, but I dunno.

The End
So anyway, I can only hope someone with half a brain neglected it long enough to read that but not too long to understand it and think about it. Any opinions, suggestions? Where do these often sweeping assumptions fall down? What have I gotten massively wrong?

And of course what have I simple completely failed to explain that needs clarification?

[Manxome]

So, some observations regarding your Big Core Assumption.

Imagine, for a moment, that the defense penalty you inflict on a successful hit were a non-stacking penalty. That is, your penalty is not the sum of the margins of success of every attack that has hit you thus far, but is the single biggest margin by which anyone has beaten you.

I notice that this is mathematically equivalent to not having defensive penalties, dealing damage equal to the margin of success, and having HP equal to the critical threshold. Your target's defense penalty means you win by X more than you would have, but you need to win by at least X in order to weaken him, and the amount by which you weaken him is your margin minus X, so the penalty gets you closer to eventually landing that critical, but doesn't actually accelerate how quickly you're hurting the target.

I point this out not because I think that's how your system works, or how it should work, but because lots of people are familiar with HP systems, so it's a useful reference point for comparison.

With that in mind, if we wanted to look at your cumulative penalty system as if it were a HP system, you still have HP equal to the critical threshold, but now, any time an attacker manages to roll above your original defense, it's like your total damage from all previous hits is doubled and then you take damage from the new hit on top of that. Plus, some hits that wouldn't have hit you at all before now do hit you and inflict some damage (but less than what you've taken up to that point). Even so, if you've been hit by 2 or 3 attacks, someone could attack you, make a roll that wouldn't even hit you if you were at full health, and still deal as much damage as one of the earlier hits.

In other words, the expected amount of damage a character sustains per hit grows exponentially as the battle continues. On average, it more than doubles every hit.

You can run with that if you want, of course. But IMHO, that's a truly scary degree of snowballing, and I'm philosophically opposed to that for two reasons:

(1) Most players tend to have the most fun when they feel like they were in serious danger of losing but still managed to win. Basically, overcoming long odds is fun. It makes you feel cool and powerful.

A system in which you spend most of the fight feeling like you're really healthy and then suddenly you're dead in the blink of an eye has a really difficult time creating this feeling; on one side, the player feels like he was never in serious danger, and on the other, he just got clobbered, and there's very little space in between. This makes the game hard to balance and means that players rarely survive a "close call." Players may very easily be complaining that the game is way too easy right up until the moment they die.

(2) It makes for better drama if a fight is close up until the last moment. The literature is full of scenes where the hero is beaten within an inch of his life before pulling out some cool trick that saves the day, or the villain unleashes his most powerful curse with his dying breath, or whatever. It usually doesn't make sense, but it still gets used over and over in books, movies, and games of all descriptions because it really is dramatically satisfying. But it's really hard to accomplish when it takes 10 attacks to deplete half your defensive resources and then only 1 additional attack to finish off the other half.


Of course, those need to be traded off against the fact that getting weaker as you get the snot beaten out of you really is generally realistic, and that's important to some people.

It's your system, so it's your call, but that's my two cents.



One other thing you should note is that if you have mid-combat healing in this system, healing 1 point of defensive penalty is way more powerful on a lightly damaged target than a heavily damaged one. You'd pretty much need to have any healing ability recover a percentage of defensive penalty, and even then it would likely be amazingly difficult to balance.

If you don't want to have combat healing, that doesn't matter. But if you do...that will be rather problematic.



That just leaves one problem with your math:

1d20+some kinda attack bonus= your attack result

10+some kinda defence bonus= your targets defence

Leaving you with a situation where if everyone maxes out everything the defensive side gets a 10 point advantage (because the defensive side gets 10 more points of cap from two additional named bonus types).

So that the idea then is that an average hit in that situation sees a small margin of success and something like a 15+ on the roll will take a large chunk out of future defence.

If I understood those formulas, then if the defense bonus is 10 points higher than the attack bonus, the attacker should only hit on a roll of 20. I suspect you accidentally wrote that "10 point defensive advantage" into two different places.

[PhoneLobster]

Imagine, for a moment, that the defense penalty you inflict on a successful hit were a non-stacking penalty. That is, your penalty is not the sum of the margins of success of every attack that has hit you thus far, but is the single biggest margin by which anyone has beaten you.

I've considered that for the past version of basically this system, and as an alternative to the one hit rule, and decided against it for two reasons.

1) eh, just didn't like it.
2) It actually makes the math to figure out where the damage is going to even more complex.

Because if your prior penalty does apply to defence when you determine the new penalty it still manages to contribute, if in a somewhat disjoint manner.

So for example under the current system if your attack roll succeeds by 3 on the first roll and you get the same result on the attack roll every round there after.
Damage Penalty
0
-3
-9
-21
Critical

If damage penalty doesn't stack but DOES apply to defence when determining damage penalty it goes...
0
-3
-6
-15
-33
Critical

Basically the same place but the second one does it a turn later with a little bit of added complexity and more of an unpredictable pattern.

I'm not sure where you are seeing the non stacking penalty as equivalent to HP

Unless you are suggesting that the damage penalty ONLY apply in determining margin of success for the purposes of a critical hit and not for determining new penalty.

But even then I fail to see entirely the comparison since as far as I can tell that scenario has the rather problematic potential to have a situation in certain odd ranges of numbers where it is impossible to ever land a critical or further improve margin of success. Among other implications.

Now as to the other concerns with the snowballing.

As to doubling (or worse) the damage penalty I don't especially care if it becomes big numbers, at least as long as its not big enough to be an administrative problem (which it pretty much can't be since before it gets too large a critical hit is assured).

I pretty much just care about how many hits over all a character can take over how many turns and that they fall down on que somewhere between turn 1 and maybe turn 8 depending on what sort of defensive or offensive advantage was in play. If the defense penalty grew scarier at a rapid pace in between it doesn't really worry me.

A linear alternative might be nice but its basically only doable using some sort of fixed damage penalty per successful non critical hit in place of margin of success. And that is genuinely a serious option, especially if its determined that the math of having SOME idea of when people will fall over is just too hard under the current system.

Also maybe I over emphasised my desire to not have critical hits be as easy to achieve. One way or another the system still revolves around them, a "critical hit' is the ONLY way to remove anyone from combat after all. Perhaps I should point out that the only certainty of a critical hit is that it removes the victim from the current combat, any other properties are basically fluff determined by the attacker, copping a critical is not fatal to the PC.

As to your note about the 10 points thing I think I screwed something up in my post, or alternately in the hazy morrass of trying to convert my base mechanic to a wild and exciting new scale. I'll figure it out and edit at some point.

[PhoneLobster]

A linear alternative might be nice but its basically only doable using some sort of fixed damage penalty per successful non critical hit in place of margin of success. And that is genuinely a serious option, especially if its determined that the math of having SOME idea of when people will fall over is just too hard under the current system.

Maybe a flat damage penalty per hit is a really good idea this time around, I remember tossing it last time for some reason I can't remember but in retrospect a solid -5 per hit this time round would look really nice, it certainly would see a much more predictable drop out rate for characters.

Mind you it still wouldn't be truly linear as the rate at which attacks dealt hits would accellerate due to prior hits, but hey...

Its now under heavy and immediate consideration.

[PhoneLobster]

Anyway, now for the report back on the glaring number discrepency.

There is a glaring number discrepency.

I've checked like five times and yes, I actually stupidly introduced a range about ten points larger than it can handle.

Its not just a typo on the post, at some point when I threw an extra couple of named bonuses at defence late the other evening thinking "lets make defence big and beefy" I just sorta forgot about the whole base 10 point on the defensive side.

However this error gives me the opportunity to do something I've been considering for a while.

Instead of reducing the range of potential bonuses I might just drop the arbitrary +10 given to the defense stat in place of the dice roll that the offensive result gets.

I've been thinking for a while about messing with a range where a net +0 bonus on both sides saw a 100% success on a d20 roll to give a bit more stretch to the bonuses that can fit in the RNG.

And since the slogan on this one is "Ah Screw It" I think I'll run with the forgotten 10 points from nowhere and have it work like this.

1d20 + Attack Bonus = Attack Result

ZERO + Defence Bonus = Defence

And tadda. It's back on track.

In a typical everyone max out the bonuses scenario defence maintains a 10 point lead. With best likely variance leaning to offence all things are even and there is a 1 in 20 critical first shot, and with max "likely" variance leaning to defence gives that whole 1 in 20 success.

Dropping the base defence bonus from 10 to zero makes it do everything it was intended to.

I just need to decide whether to convert over to fixed increments in damage penalty. Which is a smaller change than it might appear when you think about it.

Now someone give it another kick or I start talking about my order of resolution mechanics.

[JonSetanta]

The alternative combat rolls looks similar to "BESD" or "Anime d20", a now-unsupported variant.

However, as fresh as it may seem to scrap all that came before, be aware that there is a possibility to make all the same old RPG design mistakes all over again, from ignoring the past.

[Manxome]

If damage penalty doesn't stack but DOES apply to defence when determining damage penalty it goes...
0
-3
-6
-15
-33
Critical

No, it doesn't.

Say that my attack equals your defense if I roll a 10. I roll a 13, beating you by 3. You now have a -3 defense penalty.

Therefore, on the next attack, I'll equal you on a roll of 7. Say I roll a 13 again. I've beaten you by 6 (13 - 7), so that replaces your previous -3 penalty with a new -6 penalty.

I now hit on a 4. I roll a 13, beating you by 9. You now have a -9 penalty and the next attack hits on a 1.

So if I roll nothing but 13s, it goes:
0
-3
-6
-9
-12
-15
-18
-21
Critical

And damage is linear, not exponential.

If I ever roll a 10 or less, regardless of your current defensive penalty, I'll beat you by less than your current defensive penalty, which means I'm not actually hurting you if it doesn't stack. Which is why it's mathematically equivalent to having hit points instead of defensive penalties.

Again: this is just an example for illustrative purposes. Using a system that works just like HP except it's harder to keep track of is not something I would ever recommend.

As to doubling (or worse) the damage penalty I don't especially care if it becomes big numbers, at least as long as its not big enough to be an administrative problem (which it pretty much can't be since before it gets too large a critical hit is assured).

I pretty much just care about how many hits over all a character can take over how many turns and that they fall down on que somewhere between turn 1 and maybe turn 8 depending on what sort of defensive or offensive advantage was in play. If the defense penalty grew scarier at a rapid pace in between it doesn't really worry me.

Whether you intend it or not, if you have a numerical value that represents how damaged someone is, most players' instincts will be to take it as a linear measure of how close they are to dying.

You should also note that with an exponential damage scale, the utility of a defensive or offensive bonus is not linear. If you calculate how many attacks it takes to kill someone on average, I believe you'll find that it increases exponentially with their defense score (until you get to "only hit on a 20," at which point your rule for minimum damage on a 20 kicks in and they're working on an effective hit point system until their relative bonus drops below 20).

I wrote a quick computer program to crunch some numbers and estimate the number of attacks before a critical in this system.

Looks like the expected number of hits to drop someone climbs up to 3 when the defender starts at a +2 bonus relative to the attacker. Each additional point of defense adds less than one round of average suvivability up to around +10, where it takes about five and a half attacks you drop you. Then you creep above the one round/bonus threshold and it takes around 10 attacks average to drop you at a +14 bonus (notice: the tactical difference between +10 and +14 is over twice the difference between +2 and +10, despite having half the numeric difference).

Then things start to reach scary-fast levels. By +17, it takes around 18 attacks to get you. At +18, it takes 27 attacks. At +19, it takes over 37 (you initially only deal damage on a 20), and at +20 it takes around 46 attacks to take you down. After that I think it should grow around 10 attacks per defense point, given that the only way you make progress is if you roll a 20 and inflict -2.

(These stats based on 100 random combats for each defense score.)

You'd certainly need to keep bonuses within a very constrained range...

Even without your rule limiting damage sourcer per round, one higher-level character can totally mop the floor with a team of lower-level characters...

[PhoneLobster]

No, it doesn't.

Say that my attack equals your defense if I roll a 10. I roll a 13, beating you by 3. You now have a -3 defense penalty.

Therefore, on the next attack, I'll equal you on a roll of 7. Say I roll a 13 again. I've beaten you by 6 (13 - 7), so that replaces your previous -3 penalty with a new -6 penalty.

I now hit on a 4. I roll a 13, beating you by 9. You now have a -9 penalty and the next attack hits on a 1.

So if I roll nothing but 13s, it goes:
0
-3
-6
-9
-12
-15
-18
-21
Critical

And damage is linear, not exponential.

You know I should really only try to discuss this stuff when I'm not in between cross state deliveries and other related stuff.

But 3 things.

I screwed up my example damage progression again (nuts, I'm having a bad run of it).

You set the critical 1 turn too late under that suggested progression you can't have a damage penalty of -21, that was a critical.

Ok now it looks linear. Just like setting a fixed stacking damage penalty rather than a non margin based penalty looks linear.

But I suspect it still isn't linear.

Yes its linear when considered as damage per hit but it should not be linear when you consider average damage per attack.

A quick spread sheet with some dirty rounding looks like its a bit wonky really, but maybe thats the dirty rounding and the 5 hours driving an unairconditioned delivery van that I just got back from.

I'll do a version with better rounding later on and see how it turns out. My vestigial math instincts tell me that the numbers on the replacement method are not nearly as straightforward as you think.

[Manxome]

You set the critical 1 turn too late under that suggested progression you can't have a damage penalty of -21, that was a critical.

In both of your progressions, you wrote "critical" one line below the first penalty exceeding -20, so I assumed that was your convention. Heck, your first example actually has a -21 on the line before "critical," exactly like mine (the second goes to -33).

Ok now it looks linear. Just like setting a fixed stacking damage penalty rather than a non margin based penalty looks linear.

But I suspect it still isn't linear.

Yes its linear when considered as damage per hit but it should not be linear when you consider average damage per attack.

No, it's really exactly mathematically equivalent to having 20 HP and no damage penalties in all cases. I swear. You'll drop on exactly the same hit for any possible series of attack rolls (ignoring or converting special-case rules governing a natural 20).

In the initial case, say the defense bonus exceeds the attack bonus by X. You need to roll an X to tie, which means you only actually apply a penalty if you roll above X, and if you roll Y, you inflict a penalty of Y - X.

Now assume the target has some defensive penalty P. His relative defensive bonus is now (X - P), so your roll wins by Y - (X - P) = Y - X + P. This replaces the previous defensive penalty (if it's larger). In that case, the new defensive penalty is your margin of success, Y - X + P.

Notice that the total resultant penalty in the first case is Y - X. The total resultant penalty in the general case is (Y - X) + P. So you're taking the damage you would have done with no defensive penalty at all and adding the prior defensive penalty to get the new total.

Total Penalty = Prior Penalty + Damage you would do if there was no penalty

That's a simple accumulation formula.

You don't add anything if the new penalty is equal to or smaller than the old one. When does that occur? It occurs if and only if your attack roll wasn't good enough to deal damage without the target's defensive penalty. So even though you technically "hit," the target is no worse off than before you attacked, so that's mathematically equivalent to a miss.

No matter what the actual numbers or die rolls involved are, you can algebraically prove that the net result is exactly the same as a hit point system with max HP = 20 and damage = margin of success and no defensive penalty stuff at all. Every time.


Now, you don't really need to get caught up on that equivalence if you're not planning on using that as your system (clearly I'm great at following my own advice, eh?).

But just...think about some RPG you've played that uses hit points, and imagine what it would be like if every time you took damage of any kind, you first took damage equal to the difference between your max HP and current HP, and then took normal damage on top of that. Your system is kind of like that, except it also makes it so that if you're currently hurt, you can be further hurt by attacks that wouldn't even have hit you when you were at full health.

Just like setting a fixed stacking damage penalty rather than a non margin based penalty looks linear.

Actually...it's not. Because the amount of damage per hit stays constant in that model, but the average damage per attack still goes up dramatically, because you become easier to hit each time you take damage.

It's not as dramatic as penalizing the defender by the margin--off the cuff, I think the damage grows quadratically rather than exponentially--but it's still definitely not linear, and increasing initial defensive bonuses still gives faster-than-linear improvements in survivability.

I'm not sure exactly how much you're concerned with game balance, but I would suggest it would be a good idea to figure out what you want a character's actual progression from full health to death to look like, and then construct your game mechanics based on that.

I generally like any bonuses the PCs can spend resources to acquire to give diminishing or, at most, linear returns, because that tends (under various assumptions) to make optimal builds stable (you could trade a +1 here for a +1 there or vice versa and still be close to optimal) rather than pushing players towards breakpoints or arbitrary maximums. That, in turn, usually means that designing a good character is relatively easy, designing an optimal character is fairly hard, and tweaking your game mechanics produces small changes in the optimum builds, rather than wild swings.

But that's just my personal design philosophy.

[PhoneLobster]

Heck, your first example actually has a -21 on the line before "critical," exactly like mine (the second goes to -33).

Yeah but, partly due to screwing up one of the progressions the accrued penalty was not actually the same value as the last margin of success.

No, it's really exactly mathematically equivalent

When I ran a sloppy example I'd swear it looked like average damage per attack was growing and the rate at which it was growing was getting smaller.

But I'll admit, it was quick and sloppy and by no means am I sure it was a valid model of what you are describing.

Your system is kind of like that, except it also makes it so that if you're currently hurt, you can be further hurt by attacks that wouldn't even have hit you when you were at full health.

I'm aware it definitely does that and it is kinda the point.

Actually...it's not.

That's why I said it LOOKS like it is. I am fairly confident it isn't.

However it does make it a little easier to predict the behaviour of the whole thing over the long run and on the fly. And potentially also easier for balancing some special attacks against standard attacks.

In addition I suspect that a fixed penalty in the range of 4 to 5 for all successes will hurt high defence characters a little harder and low defence a little easier. Which is a moderatly attractive prospect compared to the current situation.

but I would suggest it would be a good idea to figure out what you want a character's actual progression from full health to death to look like, and then construct your game mechanics based on that.

Well that is basically what I'm trying to do right here. Because I'm uncertain as to whether my core mechanic really does exaclty what I want it to, especially at its extremes.

I generally like any bonuses the PCs can spend resources to acquire to give diminishing or, at most, linear returns, because that tends (under various assumptions) to make optimal builds stable (you could trade a +1 here for a +1 there or vice versa and still be close to optimal) rather than pushing players towards breakpoints or arbitrary maximums.

This is a fairly good point. But I really do hope to reduce the severity of the issue by rather brutishly pushing players toward arbitrary maximums.

Seriously, the slogan on this one really is "Ah, screw it" its written at the top of my document in 24 sized bold where the title should be.

Arbitrary maximum caps to bonuses are an idea long championed by RC and generally disliked by me. But I want to put together a moderately stable system with minimal effort and it seems like a really handy way to cheat my way into a shortcut to getting there. So distasteful as they are its looking like they are going to be a rather notable foundation of this system.

But despite that it DOES mean that one particular trade off, the trade off between Hard and Fast attributes will need a quick look over to see if my stubbornly clung to base mechanic has actually introduced some unfairness between generalists and specialists that I missed. So yeah. Nice point.