Engine balancing issues in ARM?

[SpaceGremlin]

Remember, a KR-2L attached to a 2500 ton asteroid can only accelerate it at 1 m/s^2. These things are meant to move huge masses, and yes that makes them overpowered. But can we come up with something more creative than a raw stat nerf? Great gameplay mechanics often rely on less tangible aspects than pure stats (see: every rpg ever).

According to the Squadcast on March 14th, these are meant to be end game rewards. Having access to them all the time with no drawbacks is bad, but there must be a way to reward a careful player with a little OP fun without letting it dominate the end game.

These parts were simply implemented too early. Their tech tree placement is wonky, and they are overpowered in the only area that currently matters. Stat adjustments will always be an option, but I'm sure a little brainstorming will yield some better ideas. Career mode improvements are coming, this is the time to be thinking about ideas to support that. If we simply lock these parts into a narrow region on a graph, we limit the potential variety of future parts. Pure statistical balance of parts is likely going to be obsolete. I'd rather embrace what the devs are trying to do here and look for ways to get these to mesh well with future updates than hold these parts to an old standard.

Also, multiplayer. More efficient parts reduce part count, which will probably reduce lag in multiplayer. Just a thought.

[maccollo]

Remember, a KR-2L attached to a 2500 ton asteroid can only accelerate it at 1 m/s^2. These things are meant to move huge masses, and yes that makes them overpowered

As result of simply having high thrust, but as has been stated over and over and over and over and over again, the total thrust is not the problem.

[Rakaydos]

Remember, a KR-2L attached to a 2500 ton asteroid can only accelerate it at 1 m/s^2. These things are meant to move huge masses, and yes that makes them overpowered. But can we come up with something more creative than a raw stat nerf? Great gameplay mechanics often rely on less tangible aspects than pure stats (see: every rpg ever).

Make the engines much heavier, and slightly stronger. They can still move 2500 ton asteroids just as well (or better) because the engine weight is low enough in comparison not to matter, but they'll be harder to maneuver unladen, making intercepts best done with a different engine.

[SpaceGremlin]

Right, right. I forgot that for a payload that big the mass of the engine starts to become irrelevant. Does anyone care to comment on any of my other points though?

[vger]

Make the engines much heavier, and slightly stronger. They can still move 2500 ton asteroids just as well (or better) because the engine weight is low enough in comparison not to matter, but they'll be harder to maneuver unladen, making intercepts best done with a different engine.

Even just nerfing their atmospheric performance would be SOMETHING.

If they're only here so we can move asteroids, then they shouldn't be getting used for liftoff at all. Cutting fuel efficiency would be another thing, making them useful for near-Kerbin maneuvers (where all the asteroids are), instead of accidentally flying into interstellar space.

Edited April 5, 2014 by vger

[KASASpace]

Remember, a KR-2L attached to a 2500 ton asteroid can only accelerate it at 1 m/s^2.

Assuming the right ship mass, added on to the asteroids mass.

I think that the Isp of the KR-2L should be increased to 390 in vac, the engine made shorter (looks wrong to me as is), and the thrust slightly reduced (It has way too much thrust for an in space engine, so you can modify that yourself with the slider)

In MP it would help. Not to mention bigger parts get you less part count for a given ship mass.

[KASASpace]

I admit, the KS-25 x4 is OP, I made an SSTO with one.

[Rakaydos]

Right, right. I forgot that for a payload that big the mass of the engine starts to become irrelevant. Does anyone care to comment on any of my other points though?

I would argue that balancing for sandbox and putting into carear, is better than ballancng for carear and tossing it into sandbox.

Balancing for sandbox is about variety and niches. A turbojet/Ion SSTO to Laythe and back is an amazing piece of engineering, but it's not going to be landing on the Mun, as it just doesnt have the right engines. A Moho lander is going to look different from a Eve Return vehical, which is different from a Kerbol Escape probe and so on. This same logic transfers to career mode- you unlock new engines because they let you do new things, go new places and do SCIENCE! there. Budgets become an optional "hard mode" limiting, but also focusing your creativity.

"Balancing for career" the way people have been talking about it here, leads to engines that are useless, clogging up the parts screen when there are better engines. When was the last time you used a Poodle engine, or an Inline Ant engine? Their sole reason for use (other than looking nice, like a space shuttle replica that uses Rocomax 55s as shuttle engines) is to handicap you until you unlock the REAL engines furthur down the tech tree- the ones like the s3 cluster that can, by itself, single stage to laythe... or mun, or Duna, all with the same design. And while you can throw arbitrary cost limitations on them to limit their abuse in carear mode, those limitations vanish in sandbox.

In my opinion, engines with different specialties and sizes and focuses is a better choice than a "technological progression" that has you paging throgh obsolite parts to get to the "real" engines, and that is why once Stupid_Chris's rebalance mod gets posted, I'm going to get it, and promote it every chance I get.

[KASASpace]

"Balancing for career" the way people have been talking about it here, leads to engines that are useless, clogging up the parts screen when there are better engines. When was the last time you used a Poodle engine, or an Inline Ant engine? Their sole reason for use (other than looking nice, like a space shuttle replica that uses Rocomax 55s as shuttle engines) is to handicap you until you unlock the REAL engines furthur down the tech tree- the ones like the s3 cluster that can, by itself, single stage to laythe... or mun, or Duna, all with the same design. And while you can throw arbitrary cost limitations on them to limit their abuse in carear mode, those limitations vanish in sandbox.

Some, like retro gamers, could argue that they (less advanced engines) still have their place. I like using those parts all the time, and typically use a Poodle as an upper stage for a HUGE rocket, a tiny upperstage, even so.

Besides, a SSTO to Laythe? I wouldn't even WANT to do that. I want a decent challenge, where when I fail, it's funny, not frustrating. And when I succeed, it is immortalized in Kerbal History for that save.

If I even have to FORCE myself to use those parts, I will.

[stupid_chris]

To those interested, I've put on the table the changes talked about here in practice http://forum.kerbalspaceprogram.com/threads/75272

[Z-Man]

But being a lifter engine isnt the problem. Being a high TWR lifter engine AND a high ISP space engine sounds awesome for moving high mass asteroids around!

Yes. You can read the intention from the atmospheric I_sp; the KR-2L has a lower one than the other new engines, and you can attach to it from below. It is meant to be an awesome space engine for larger vessels. Increasing its mass would be the right move to reinforce that.

If you really want to explore those infinity solutions, the engine would have a TWR of very near infinity, and an ISP of very near 0. This means it would burn all it's fuel instantly but give an infinite amount of thrust.

That would be an engine balanced according to your graph. But in practice, it would be useless. You would not be able to build a rocket clearing even 1 km in its first stage with it. I am using this extreme example to demonstrate that your analysis method is wrong there, but it actually is wrong everywhere. That you have to cull outliers also shows that.

Amaroq's metric is much better. Specifically, single stage payload fraction to LKO is an excellent measure for lifter engines. I personally think that larger engines should be at least as good at that particular job than smaller engines so players are not forced to decide between part count (use large engine) and efficiency (use cluster of smaller engines).

[Rakaydos]

That would be an engine balanced according to your graph. But in practice, it would be useless. You would not be able to build a rocket clearing even 1 km in its first stage with it. I am using this extreme example to demonstrate that your analysis method is wrong there, but it actually is wrong everywhere. That you have to cull outliers also shows that.

Sure, because the Sepratron is a useless rocket engine. And the Launch Escape Tower is too.

They are extreme cases that are almost useless for conventional asents... but they get used for OTHER things. And besides, there's a video somewhere of someone using stacks of LES engines Orion style. Stage! Stage! Stage! Stage! ...all the way to space.

[MKI]

The NASA Pack is a Pack of parts. The same way KW is a parts pack or B9. They shouldn't be compared with Stock parts at all.

The parts are balanced for what they are. They are "End Game" reward parts that reward those who get to the end of the Tech Tree in Career. Their high efficiency, and effectiveness (and low part count!) do make them worth getting. This whole thread is about how great they are, so im sure plays who get through career will appreciate getting these parts and utilizing them.

The argument about Sandbox vs Career is IRRELEVANT. Most if not all players here are veterans. So we know what it means to make the "Right Stuff" (pun) when it comes to making good rockets and ships. Its an accomplishment, a trophy. Now these new parts make all of that seem pointless. But These parts are here for 2 reasons.

1. To be an easy way to get to space, even for large payloads.

2. To be a reward for those who get through the tech tree in career.

Sandbox has been until recently the way to play the game, until now. The Dev's are focusing a lot of effort into career mode. These parts are one of the many things implemented FOR career mode. Yes they exist in Sandbox, but they don't exist FOR sandbox. These parts are just a different side of the same coin as the Science parts. The new parts help you fly to the stars, where as the Lab and Goo containers are the reason you can't.

In time Sandbox will not be a challenge, it shouldn't be. Building rockets shouldn't be hard when you have all the parts available. This is probably the goal Squad is going for. Career needs to have its own benefits to make it more the core KSP experience. These parts are part of that. (get it?)

Yes these parts blow all the other parts out of the water. Yes these parts make doing certain things very easy(like making an SSTO) And yes these parts could totally be nerfed. But, These parts are also part of a pack, that just so happens to come with the game. If you REALLY did not agree with these new parts, don't use them. Or mod them so they are nerfed (a mod for a "mod" lol)

I feel the argument for nerfing these parts isn't on the most stable of grounds, unlike actual unbalnced parts (48-7S) which are stock but have no point in actually being that effective. The parts are designed to be the biggest(they are) the baddest(they are from NASA!) and the best(low part counts, PWNAGE stats)

How does nerfing them make anything better?

How does changing something that is strong to something "less" strong get anywhere? Again they are at the end of the Tech Tree and probably be very expensive to use. It means designing a single HUGE launch utilizing these parts are possible. Nerfing these parts is redundant to the idea they are to be used for things not possible before.

Squad fixed 2 massive game play issues this update that kept players from getting bigger and more awesome things into space. They changed the Joints to allow bigger ships. Gave bigger, massive, and highly effective parts to utilize the new joints.

All to allow players the ability to do more with less(less cpu that is) and players don't want it?

Seems sort of spoiled to me.....

[Tiberion]

and players don't want it?

Seems sort of spoiled to me.....

Which players? no one is trying to take the parts away. Making them balanced consistently with the rest of the game doesn't "ruin" anything.

I think its a huge shame that people have to get defensive and launch into hyperbole rather than just trying out the changes proposed (its an easy edit, and there's even a ModuleManager file now) to see how it would play when balanced properly.

They certainly feel more like newer, better engines made by the kerbals, rather than Sci-fi engines running on fairy dust or whatever powers the current setup.

[CaptRobau]

The NASA Pack is a Pack of parts. The same way KW is a parts pack or B9. They shouldn't be compared with Stock parts at all.

The NASA pack can't stand on its own. KW or B9 can do so to a far greater degree, as they have a larger range of engine types and sizes. Stock and the NASA stuff need to be used together, which means that they can be compared.

[Z-Man]

Sure, because the Sepratron is a useless rocket engine. And the Launch Escape Tower is too.

They are extreme cases that are almost useless for conventional asents... but they get used for OTHER things. And besides, there's a video somewhere of someone using stacks of LES engines Orion style. Stage! Stage! Stage! Stage! ...all the way to space.

Do they have an I_sp of 1 like the hypothetical example I stated? No? Well, then.

But still, where would they be in stupid_chris' graph? Keep in mind that for SRBs, you need to take the maximum TWR to be somewhat consistent. I tell you where they are. Deeply in the region he considers OP. Are they really OP? What's their payload fraction to LKO? With staging allowed in this case.

[Pthigrivi]

So, generally, I feel like "balancing for sandbox" is a red herring. Sandbox is a testbed for developing the real game, which is career. Imagine playing a version of starcraft where every unit was unlocked from the start and nothing cost any resources, and then complaining that battlecruisers are OP. Sandbox is fun, but ultimately not more important than gameplay in career mode.

That said, even in career, even with cost considerations, I think it still makes sense for parts to be somewhat balanced. I think in general later tech tree engines should be slightly better to incentivise progression, I also think once a person has maxed out their tech the reward should not be "Oh I guess I just always use this engine now." So sure, the new engines are better than the old ones. Im not bothered by that in principle, especially considering their cost. But I could see an argument for a middle solution, nerfing them somewhat, so that they were still useful for moving asteroids and lifting huge payloads but still left room for late game usefulness of the 2.5m engines. I imagine there will be a lot of tweaking like this in the future, especially after .24 when we see just how much impact cost has on the game (which will be huge).

[SpaceGremlin]

The current engine balance was an artifact of sandbox being the only game mode. With career being the primary focus, that opens up the option for the devs to implement direct upgrades that have no tradeoffs statwise (other tradeoffs being difficulty to acquire, price, etc). Maybe someday a balance pass will affect every engine and throw out the old balance curve. Would that ruin the game? No. Does nerfing the SLS parts to be like every engine ruin the game? No. These are subjective design decisions that will have to be tackled by the devs to fit their vision of the stock game. For everything else there's a mod.

[ferram4]

I think there's another problem with the SLS engines, but one that hasn't been addressed and that I think causes people to dismiss the "overpowered" argument and the numbers out of hand: their thrust per unit cross-sectional area doesn't scale up like it should; this makes the engine able to lift less fuel / payload, and makes it "feel" weaker. As a comparison between the heavy lift engines:

LVT-45 (not 30, since that would require trying to balance thrust vectoring)

Thrust = 200 kN

Diameter = 1.25 m

Area = 1.2271 m²

Thrust per unit area = 175.2 kN / m²

Mainsail

Thrust = 1500 kN

Diameter = 2.5 m

Area = 4.9087 m²

Thrust per unit area = 305.6 kN / m²

KS-25x4

Thrust = 3200 kN

Diameter = 3.75 m

Area = 11.0447 m²

Thrust per unit area = 289.7 kN / m²

Really, the thrust per unit area for the KS-25x4 should probably be at least on par with the Mainsail, if not a little bit higher.

Now, for illustration, I'm going to throw in the KW Griffon XX, since that's where I (personal opinion) think a heavy lift 3.75m engine should be, and the Titan, which still feels a bit better than the KS-25x4 for launching rockets that need boosters to get going:

KW Griffon XX

Thrust = 4900 kN

Diameter = 3.75 m

Area = 11.0447 m²

Thrust per unit area = 443.7 kN / m²

KW Titan T-1

Thrust = 3600 kN

Diameter = 3.75 m

Area = 11.0447 m²

Thrust per unit area = 325.9 kN / m²

You can launch quite heavy rockets on the Griffon XX, and it feels like an incredibly powerful engine because it can lift a tall, heavy rocket above it, despite the fact that its Isp is poor. The KS-25x4 feels much more lackluster because, despite it's good Isp, it can't lift a tall rocket above it, requiring either heavy augmentation through boosters, which makes it feel like a less effective engine because it's not moving gigantic stages and payloads. In comparison, the Mainsail can lift quite a tall stack of fuel tanks, hence why it has always been fairly popular.

As a further comparison, let's throw the Skipper and Poodle against the KR-2L:

Skipper

Thrust = 650 kN

Diameter = 2.5 m

Area = 4.9087 m²

Thrust per unit area = 128.3 kN / m²

KR-2L

Thrust = 2500 kN

Diameter = 3.75 m

Area = 11.0447 m²

Thrust per unit area = 226.4 kN / m²

Which makes me think that the KR-2L thrust is probably okay, maybe needs a minor nerf, but the Isp and dry mass should probably go up to compensate.

This is just something I'm throwing out there, since it does affect the way an engine ends up feeling, which might affect how people perceive an engine's performance and how they perceive the numbers. It has been talked about, and I think it might be relevant to the topic at hand, since we not only want engines to be balanced, but to feel balanced and fun to use. Any thoughts on these as possible metrics?

[LethalDose]

As far as the graph on the first page goes, is there any justification for fitting an exponential function to fit ISP vs TWR? (e.g. do engines IRL conform to a similar function?).

I'm also really concerned that the inclusion of the nuke is really driving the entire equation being fit on that chart. If you know anything about regression or statistics, that whole chart and equation (which most of this point seems based on), throws up a lot of red flags.

[Tiberion]

Ferram: The new engines could be bumped up in thrust without issue, but their mass needs to go up quite a bit too. But there is a limit before you have to start using out-of-line Isp settings to get a reasonable engine + tank stage to fliy long enough to be useful. I've toed that line in NP to keep masses down (since its been game breaking for so long)

All three of the new engines would be better (and more useful) if they were balanced to fill a specific role (like any rocket engine, ever) rather than be hybrid. (again, something I've tried in NP without much success)

Lethal: see here: http://forum.kerbalspaceprogram.com/threads/74448-SLS-engines-Probe-Rockomax-buffed-ions-and-a-larger-solar-system?p=1057132&viewfull=1#post1057132

All the "unbalanced" engines on the chart are excluded, including LV-N

[purpletarget]

I'm also really concerned that the inclusion of the nuke is really driving the entire equation being fit on that chart. If you know anything about regression or statistics, that whole chart and equation (which most of this point seems based on), throws up a lot of red flags.

It probably should. The trendline is based only on the cherry picked engines which are previously assessed as "balanced".

If you run a polynomial trendlines with all the conventional engines (No LV-N), you get a much different picture.

Full trendlines for conventional engines.

Breaking them up by size class also ends up with a bunch of funny lines with no real pattern.

Multi-trendlines, broken down by sizes

Finally, IRL engines...don't really follow any trendlines. It's all just a mish-mashy-mess.

Smattering of IRL engines compared to KSP incl ARM

If we want to take a look at another factor for example, how about how the thrust of an engine relates to the mass...the trendlines aren't nearly as offensive, it's almost like they mostly fit. If anything the 25x4 starts to look too heavy.

Thrust to Mass

Here's another factor where the size-3 ARM engines really don't bother me, ....the 25x4 needs more fuel just to move itself than the smaller ones....yes, even the mainsail. And while the KR-2L needs less fuel than the mainsail in Vac due to a superior ISP, the ASL leaves the Mainsail as requiring slightly less fuel, which ties in with the KR's high atmo area.

ASL Fuel Mass

Vac ISP

These charts are not intended to necessarily prove anything particular about the engines, but merely to demonstrate that there's many ways to make "math" say whatever you want it too....so don't think that just because it's math, that it's beyond dispute.

[LethalDose]

All three of the new engines would be better (and more useful) if they were balanced to fill a specific role (like any rocket engine, ever) rather than be hybrid. (again, something I've tried in NP without much success)

I fully agree with this statement: The engines should be designed to fulfill particular roles.

Lethal: see here: http://forum.kerbalspaceprogram.com/threads/74448-SLS-engines-Probe-Rockomax-buffed-ions-and-a-larger-solar-system?p=1057132&viewfull=1#post1057132

All the "unbalanced" engines on the chart are excluded, including LV-N

I'm not sure what I'm supposed to be looking at here, it looks like posted chart there is identical to what's posted in this thread.

It probably should. The trendline is based only on the cherry picked engines which are previously assessed as "balanced".

If you run a polynomial trendlines with all the conventional engines (No LV-N), you get a much different picture.

These charts are not intended to necessarily prove anything particular about the engines, but merely to demonstrate that there's many ways to make "math" say whatever you want it too....so don't think that just because it's math, that it's beyond dispute.

These charts are way more useful, but in several places you've fit polynomial curves using just three points, which can always be perfectly fit with a second degree polynomial (i.e. a quadratic function).

Briefly, models used to explain data have to be justifiable, and "this is what Excel gave me when I gave it data" is nowhere near justifiable.

Due to previous experience, I'm not really interested in explaining details behind the issues here. Anyone who's interested should look into a concepts of "leverage", "residual analysis", and "influence". These are basic quality control steps that are required to demonstrate validity of results. Visual observation is enough to determine that what's presented here isn't valid. I'll simply leave it with a quote from George Box:

"All models are wrong, but some models are useful."

The models presented by Stupid_chris are not useful.

[stupid_chris]

Have some better plotted data then

Javascript is disabled. View full album

You're not the only one who had statistics classes.

[LethalDose]

You're not the only one who had statistics classes.

And anyone who has had statistics classes knows that systematic rationale is the most important part of model specification; none of which has been provided here.

I've already said I'm not discussing it further. Especially not with a confrontational mod.