Rethinking KSP's career mode

[Rombrecht]

Greetings fellow space nerds!

Since early alpha I've been an enthusiastic player of KSP. The game has thought me alot about the mechanics of spaceflight and motivated me to study the history of spaceflight.

When career mode was added, at first I was quite exited about the prospect of playing the game like a real space program manager. However, I find myself exclusively playing sandbox mode these days again because I keep getting frustrated with the way career mode functions in KSP.

Being the stickler for historical accuracy and realism that I am, I always found it rather odd that the KSP career begins with a manned mission. unmanned probes only come into the game after some research. While in reality of course, every destination in space has always first been visited by unmanned robotic probes, paving the way for manned exploration.

Also, the fact that one needs to gather "science" by measuring the properties of the different regions of space and celestial bodies in order to develop rocket parts seems a bit absurt. One does not learn about the mechanics of liquid rocket engine turbo pumps by studying the craters of the moon. In the real world, knowledge about rocketry is gained by spending money on testing parts on the ground, and by flying (and occasionally blowing up) equipment.

So, I've been finding myself RP-ing a career in sandbox mode. Would it not however be swell if we could imagine a more realistic career progression and mechanic then is currently implemented in the game? Mods like Better then starting manned partially tackle the issue, but technological progression is still based on "science" gained form studying space in stead of spending money on R&D.

And yes, I know this is a game about little green men riding insane rockets into space, and less about an accurate representation of historical spaceflight. But hey, speculating about the mechanics of a game never hurts. What are everyone's thoughts?

[tater]

Many of us have been saying such the same for a while now. If they can't deal with stuttering, or wheels and landing gear no longer working properly I think the chances of any improvements (read: scrapping and doing over) career are nil.

[Rombrecht]

57 minutes ago, tater said:

Many of us have been saying such the same for a while now. If they can't deal with stuttering, or wheels and landing gear no longer working properly I think the chances of any improvements (read: scrapping and doing over) career are nil.

Well now, no need to be so negative. I'm not expecting SQUAD to immediately change their entire game. However, what stops us from imagining alternative career mode architectures, perhaps for inspiring mod makers?

[tater]

Sorry, I just think that's the way it will be at this point, and we can only hope for mods to make a decent career mode. Every change to career has been to double-down on the bad paradigm of "contracts" and planetary science somehow furthering rocket parts. For reasons.

[Rombrecht]

34 minutes ago, tater said:

Sorry, I just think that's the way it will be at this point, and we can only hope for mods to make a decent career mode. Every change to career has been to double-down on the bad paradigm of "contracts" and planetary science somehow furthering rocket parts. For reasons.

Well, how would you like to see the career mode then? This is the place to discuss it I would hope so.

[regex]

2 hours ago, Rombrecht said:

When career mode was added, at first I was quite exited about the prospect of playing the game like a real space program manager.

I, too, would enjoy a "light management sim" but KSP is trying to accommodate too many play styles to go that route.

2 hours ago, Rombrecht said:

Being the stickler for historical accuracy and realism that I am, I always found it rather odd that the KSP career begins with a manned mission. unmanned probes only come into the game after some research. While in reality of course, every destination in space has always first been visited by unmanned robotic probes, paving the way for manned exploration.

Kerbals are the star of the show in KSP; sending probes first is not their way.  Furthermore, this is not a historically accurate simulator at all; the fact that we're using nebulous things like "LiquidFuel" and "Oxidizer" should clue you in.  That's a design choice and one that got the game so popular, IMO.  You would literally be better served by installing the Realism Overhaul suite of mods.  Launching spin-stabilized sounding rockets in early career is pretty satisfying.

[John FX]

For me the whole reason I like building rockets is because of the human exploration of space and my desire to recreate it. I don`t want to go the full orbiter route, I quite like the way you can just build a crazy rocket out of the wrong parts and see if it works.

What I would really like though is for there to be more meaningful game elements between starting the game and getting to orbit. If we are going to have jet engines and wings and unmanned probe cores then we should have a good reason to use them, for example building planes and probes...

Not just as something to do once you have finished the game.

For myself a good approximation of what I would like career to be is RP-0 based in a solar system just like ours where you are almost forced to send unmanned probes before manned because sending manned (with the associated life support, power generation etc) weighs too much until you get better engines.

It used to be an issue that installing RSS/RO/RP-0 would put you very near the memory limit but hopefully once all the mods are updated for 1.1.1 that will be a thing of the past.

I think KSP has an identity crisis at the moment, it started out wanting to be a cartoon style lego building fun rocket game with little green men but then we got more and more realism and better and better looking parts that matched more real life parts that it no longer is that game it started out as. Now we have little green men and women who launch in shuttles and have spaceplanes that have to be careful because the atmosphere will destroy you otherwise.

We are getting a Delta-v readout.

To me that will be the pivot point where we go from `slap some pieces together and see if it explodes` (the kerbal way) to `you are starting to have the tools in stock you need to build proper rockets` (closer to real life)

I`d like the stock career to be better but I don`t think Squad want to change it much. Luckily we have a wonderful community who have stepped up to the challenge and are well on the way to making a very decent career based in a Real Solar System.

[tater]

Yeah, in stock I'm not wed to starting manned, but career lacks any sense of a, well, "career."

Career needed a hash out of what the point was, then it needed to be designed with that goal in mind.

Assuming that the goal is to follow the arc of a space program, with the player as a sort of administrator, we need some new tools and mechanics.

1. Time needs to be a thing. Currently long time only passes due to time warp, and the only reason to time warp is to progress a mission in flight, or to warp to a transfer window. If building things took time (including rockets), then time would progress more naturally. The usual complaint is "needless" warping, but warping is a thing, and moving the calendar forward is not "needless," it's in fact needful. KCT as a mod does this, but stock likely needs a more simplistic approach (but not much, KCT is a great mod).

2. The science/tech paradigm is broken. This is harder, since buying tech is currently the only psychological reward system in career. Intermediate fixes are better tech trees (CTT and ETT, as examples of mods that do this).

3. To actually be a management game at all, kerbals should be able to do things on their own if the player wishes---execute maneuver nodes without the player needing KAC to do it themselves, for example. I'd add in even completing routine missions alone if assigned (station resupply, etc). With meaningful time this becomes more important as you usually have a few balls in the air at once.

4. Ideally there would be some sense of science giving you something for its own sake---doing some science could unlock an atmospheric trajectory prediction, for example. Some could allow better map zoom, and/or the ability to place markers on the map (to use as targets just like a landed ship) to facilitate better landings.

5. The ability for the player to set goals/programs explicitly. Ie: set the goal as landing kerbals on the Mun, and returning them to Kerbin.

6. This won't happen, and doesn't really deserve to be in the list, but if they can make AI kerbals, they can make an AI space program. I'd love to see a "Space Race" mode (which is implied in the game's rescue contracts and "Firsts" implicitly, anyway). Such a mode would be a foil for the player, and the time constraint of trying for "firsts" would create interesting decision branches for the player... wait till you get that part you want, of fly for the Mun without it in your sketchy craft to beat the Koviets.

Edited April 29, 2016 by tater

[Stoney3K]

I'm inclined to agree that KSP is going through an identity crisis which is perhaps due to its "puberty" phase where more advanced features are being tested and fleshed out, until it becomes a full-featured game.

KSP is a sandbox game which offers some railroading in the form of contracts to give the player a degree of challenge. But it's basically the same as buiding in SimCity or Transport Tycoon with unlimited resources enabled -- which can be fun, but it requires the player to role-play their own limitations if they want to have certain challenges.

If they aim for KSP to be more like the classic "tycoon" style games, it means adding more challenges, more constraints and competition. Currently, the biggest thing missing from the career game is, as you say, time management, as time is pretty much an unlimited resource. You can warp forever without any penalty except contracts that expire -- no life support running out, and no competitor on your tail who chases you to get to the Mun first. Even a simple "leaderboard" style which records a player's achievements and succesful missions, which can be published online, can provide a sense of achievement and competition in lieu of a crude multiplayer mechanism.

[tater]

@regex has put it well that contracts are "random side quests" (my apologies if anyone else said it that I missed). 

I edited my post above to add something like Strategia, where the player makes missions/goals, instead of them being either automatic, or random contracts.

The possible contextual contracts addition makes some of this more plausible within the current framework. I would like to see "internal" contracts that come from your program, that fit the goals you might set in something like strategia. So your goal is the Mun, but your science team then suggests possible landing spots (via survey, etc, "contracts"). Instead of rewards being funds/science upon completion, perhaps (assuming time became a thing) the funds come from a budget, and perhaps science points can be awarded in advance---to buy tech for the mission in question.

Say the whole funding paradigm became an annual budget, paid out to the player in chunks every month or so (a Minmus month is ~50 days, or it could be a Kerbin year/10, or whatever). This would make time be a thing, instantly, as if you are low on funds, you warp to the next month. Bam, 40-50 days pass. With real mission planning, your goal is, say a crewed Duna Program at this point. Taking that goal would award some science to spend on stuff for that mission. Contracts and missions (Duna is the program, I'll use "missions" for the internal to KSC contracts) are then generated for Duna. Linus suggests orbital science at Ike, and various science on Duna at specific places. Taking those gives you extra science to spend, and perhaps can borrow from total annual funds, NOW for that mission (reducing your later payments, obviously, as you took an advance). Time limits on missions and contracts would be slashed across the board. You'd not have 10 years to complete a rescue, it would be 10 days. In the case of Duna, completion would be set for some years that make sense. Perhaps in picking a Program you could set a starting date yourself (aiming for transfer window).

[Knaapie]

Funny, I was just thinking about time and budget games. Time warping and getting free funds seems to be the issue there.. I had a thought to work around it:

What if: time warp becomes: "warping into the future.." This will enable you to do your entire mission, yet, when you return to KSC, you return to real time. Only being able to warp to the next month and get your next funds/costs balance and maybe be able to build a new rocket.. You will see the progress of your running (completed) missions every month ^^..  And you'd need many more Kerbals ^^

Edited April 29, 2016 by Knaapie

[HebaruSan]

2 hours ago, tater said:

6. This won't happen, and doesn't really deserve to be in the list, but if they can make AI kerbals, they can make an AI space program. I'd love to see a "Space Race" mode (which is implied in the game's rescue contracts and "Firsts" implicitly, anyway). Such a mode would be a foil for the player, and the time constraint of trying for "firsts" would create interesting decision branches for the player... wait till you get that part you want, of fly for the Mun without it in your sketchy craft to beat the Koviets.

I think an AI foil would pretty well wreck KSP, frankly. The huge range of player skill means it would be way too easy for some players and way too hard for many more, and the ability to choose which missions you want would be reduced, as would replayability, somewhat, as you'd have to play to beat the AI's (likely hard coded) strategy in every save rather than setting a goal you haven't tried before. You'd lose control over how many active vessels there were, so performance could tank without you doing anything or having any way to fix it. And who knows how interactions with "enemy" craft would be handled. Should the AI be allowed to shoot down your ships while you're warping to morning for a launch?

Then imagine if the 1.1.x issues with legs and wheels were transmuted into equivalent problems with an AI player. "It just sits there," "It reported a Mun landing two seconds into the game," "It cheats on delta V," "I can win in one mission," and so on.

I find the ideas for meaningful passage of time, realistic-ish discovery and investigation of bodies, comm delays for probes, annual budgets, and so on much more viable.

[regex]

22 minutes ago, tater said:

@regex has put it well that contracts are "random side quests" (my apologies if anyone else said it that I missed). 

Someone else said it before me but, yeah, that's what they feel like.

7 minutes ago, Knaapie said:

Time warping and getting free funds seems to be the issue there.

Well, the idea behind monthly/yearly budgets is that you get a set amount of funds per budget period each time, they don't just increase infinitely for doing things like the game does now.  In that case you could base the available funds per period off of reputation or something.

3 minutes ago, HebaruSan said:

I think an AI foil would pretty well wreck KSP, frankly.

Totally agree here.

Edited April 29, 2016 by regex

[Knaapie]

20 minutes ago, regex said:

Well, the idea behind monthly/yearly budgets is that you get a set amount of funds each time, they don't just increase for doing things like the game does now.  In that case you could base the available funds per period off of reputation or something.

Well yes, that would result in infinite funds in combination with time warp. So you'd need costs as well and fund benefits from completing missions.. But that would result in tedious filler missions during every long mission (like Jool) as discussed earlier (I forgot in which topic)..  That's why you'd need to be able to fly in the future and get monthly funds for your active missions. Your Jool mission could give you 360.000. If it were to be completed in 3 years you'd earn 10.000 a month, once the mission is completed in the "future warp".

/edit: 

It suddenly makes time important: do you launch this month? or build a better rocket next month ?

Edited April 29, 2016 by Knaapie

[regex]

6 minutes ago, Knaapie said:

Well yes, that would result in infinite funds in combination with time warp.

How so?  Because the budget "pool" refills every period?  Sure, I guess you could call that "infinite" but the limitation is that the budget "pool" never gets above a certain point.  Therefore you are limited in what you can fund during a certain period.  You could certainly use orbital construction to build bigger but that would mean a lot of time-warping to build and launch all the component parts plus potentially lock you out of other missions while all your funding was tied up in the larger project.

This is not an issue, it's a management consideration.

6 minutes ago, Knaapie said:

So you'd need costs as well and fund benefits from completing missions.

Not at all opposed to upkeep costs.  Another management consideration.

6 minutes ago, Knaapie said:

But that would result in tedious filler missions during every long mission (like Jool) as discussed earlier (I forgot in which topic).

Why?  If you have a certain budget available every period and choose not to use it (aside from upkeep) that should not be a problem.

In the budget scenario, where some external value determines your maximum budget, perhaps based on missions you've previously accomplished, every mission is meaningful because it adds to the value determining your current budget.  Kind of like now, only you don't just add more and more funds, you increase your fund maximum.

[HebaruSan]

5 minutes ago, regex said:

Totally agree here.

To elaborate a bit more, much of the gameplay fun that KSP delivers early on is in learning how to do basic tasks. Getting to orbit for the first time is a satisfying achievement because when you started, you may not have even understood what that meant. Likewise a Mun landing and return. You try things, you watch them succeed or fail, and you build up a body of knowledge and rules of thumb that allow you to improve and eventually achieve cool goals.

But an AI doesn't have to learn. It can launch pre-determined optimal missions rapid-fire from the first minute as long as it has funds, and pilot them without errors. (And I expect that rather than spending time in the VAB/SPH, the AI would use all stock crafts; craft design is an entire portion of the game that the AI can't even participate in.) While the new player is exploding things on the pad and incinerating suborbital pods without parachutes, the AI has already burned for the Mun and is mere in-game hours away from presenting a "You've lost!" popup to the player. You can forget about time warping to Jool, because the AI will be biome hopping and contract farming on Mun and Minmus the entire time you're en route, building up an insurmountable lead.

Well, what if the AI doesn't use optimal strategies? Then regex and tater and others will wipe the floor with it every time and complain about how this AI wasn't worth adding because it's such a pushover. A difficulty slider might mitigate some of this, but some of it is just inherent in the way science, contracts, and funds are structured.

[Knaapie]

7 minutes ago, regex said:

Sure, I guess you could call that "infinite" but the limitation is that the budget "pool" never gets above a certain point.  

Right.. That's an (better)easier solution

Edited April 29, 2016 by Knaapie

[Stoney3K]

1 hour ago, HebaruSan said:

To elaborate a bit more, much of the gameplay fun that KSP delivers early on is in learning how to do basic tasks. Getting to orbit for the first time is a satisfying achievement because when you started, you may not have even understood what that meant. Likewise a Mun landing and return. You try things, you watch them succeed or fail, and you build up a body of knowledge and rules of thumb that allow you to improve and eventually achieve cool goals.

But an AI doesn't have to learn. It can launch pre-determined optimal missions rapid-fire from the first minute as long as it has funds, and pilot them without errors. (And I expect that rather than spending time in the VAB/SPH, the AI would use all stock crafts; craft design is an entire portion of the game that the AI can't even participate in.) While the new player is exploding things on the pad and incinerating suborbital pods without parachutes, the AI has already burned for the Mun and is mere in-game hours away from presenting a "You've lost!" popup to the player. You can forget about time warping to Jool, because the AI will be biome hopping and contract farming on Mun and Minmus the entire time you're en route, building up an insurmountable lead.

Well, what if the AI doesn't use optimal strategies? Then regex and tater and others will wipe the floor with it every time and complain about how this AI wasn't worth adding because it's such a pushover. A difficulty slider might mitigate some of this, but some of it is just inherent in the way science, contracts, and funds are structured.

You're forgetting that an AI may have a totally different goal from the player. If the goal is "Get to the Mun first" as a contract, then such a contract would only become available once the player is (feasibly) within reach of doing so. It doesn't make sense to offer such a contract when the player only has the first tech node unlocked.

An AI would only do just as good as a player would if it gets offered "Test the Kickback booster on a sub-orbital trajectory near the Sun" contracts just as the player would. It gets bogged down in grinding all possible test and tourist contracts that would be achievable in any given time, not setting any long-term goals unless it is offered a contract. A player can already think ahead and predict that it may get a "Fly by the Mun" contract in the near future, and budget for it accordingly.

You can see the same behavior on Transport Tycoon's AI. Those will keep hunting for every possible subsidy which offers them a higher reward than regular, trying to get ahead by buliding microscopic networks with subsidised passenger runs, which pay almost nothing per completed run. When you play against those as a player, you can easily beat them by running heavy networks of bulk freight, which have a higher net payoff, but take more time and money to set up, but keep generating reward once they are running.

An AI may not be the best option here but we need a competitive element to keep the game challenging and rewarding. You may feel a sense of accomplishment once you get your first payload into LKO, but once you have flown 100 payloads into LKO, the task becomes repetitive and boring, so you would rather have something that automates it and enables you to focus on different tasks. If I have the tech to build a lifter that can haul 100t of payload into LKO, I can reasonably expect my crew to build that lifter, attach any payload up to that 100t, and drop that payload into a predefined parking orbit without my intervention.

[HebaruSan]

3 minutes ago, Stoney3K said:

You're forgetting that an AI may have a totally different goal from the player. If the goal is "Get to the Mun first" as a contract, then such a contract would only become available once the player is (feasibly) within reach of doing so. It doesn't make sense to offer such a contract when the player only has the first tech node unlocked.

Indeed, I'm assuming that a "Space Race" as such depends on the participants having the same or similar goals. Otherwise it's not a race, it's two unrelated space programs operating in parallel. I'm also assuming the AI would not have to deal with contracts at all. Doing so would require it to evaluate the success conditions for how difficult they are at the current level of technology, then somehow map that to an appropriate craft and generate a new and unique mission plan. That just seems totally unworkable to me; I don't think SQUAD would ever attempt to program that, and if they did, it would have far worse problems than I suggested earlier (such as, "The AI never completes any contracts" ).

So yes, I'm imagining a redesigned game mode to allow an AI to be implemented at all. It would need to have a sequence of predefined milestones that you and the AI both "race" towards, and some ability of the AI to achieve them. And while that's more possible to do, I don't think it would be as fun as is often hoped, because to an extent such milestone events are mathematically solvable.

3 minutes ago, Stoney3K said:

An AI may not be the best option here but we need a competitive element to keep the game challenging and rewarding. You may feel a sense of accomplishment once you get your first payload into LKO, but once you have flown 100 payloads into LKO, the task becomes repetitive and boring

A large multiplayer game with procedural planets and fog of war might be cool. Each player would have a separately unlocked knowledge level of the planets and technology with purchasing and trading allowed, so one could try to corner the market on astronomical data, another could focus on parts, another on landings, etc. It could have multiple victory conditions like Civ, e.g., funds victory, reputation victory, science victory, exploration victory. But that would take a very long time to play.

[tater]

Strategic AI could have widely variant goals. In a KSP context, they could be rival countries, or rival companies... Your goal might be Duna, their goal is tourism to the Mun. In any race mode (it'd be a new mode), rep would be the relative score.

[regex]

I can't see a game AI "playing" KSP in any way, shape, or form like an actual human player, it would be more like events that happen at specific times.  I also think that could work to push the player if there were tangible reward/penalties for beating the deadlines but, as @HebaruSan points out, that will be very artificial and easily gamed, even if we allow "difficulty sliders" (god how I hate that term).

I don't think "Space Race" works with KSP.  It either turns into a railroaded conquest of artificial milestones or it is totally irrelevant.

[Rombrecht]

On the identity crisis of KSP

Fully do I agree that a game featuring little green men should be taken with a hefty pinch of salt. And when the Game was in Alpha and there was only orbit and the mun to go to, slapping together some boosters and flying by the seat of your pants was indeed a viable form of gameplay. I think KSP's identity crisis started to appear once other planets were added, requiring much more complex mission architecture such as transfer windows Delta-V calculations. While slapping boosters together and watching them blow up can still be fun, that kind of gameplay isn't very sustainable in terms of keeping the player interested for longer periods.

I'm not expecting a full-on NASA/ESA/Roscosmos experience with Kerbal Space Program. I have no problem with the fact that the Kerbal solar system is scaled down dramatically for example. Nor am I expecting to program every step of the mission in advance into a computer like they do in the real world. (though one could of course do this with K-OS). I'd even go as far to say that it would be an improvement to the game if the rocket parts had a much more cartoon-esque look to them then the realistic shape they tend to have now. That would even be very appealing. 

The Strength of KSP lies in the marrying of cartoon-esque slapstick with real spaceflightscience. The Look and feel of the game is cartoonish, but the gameplay at it's core is science-based. Looking for a more realistic mechanic for career mode progress does firmly lie within the bounds of KSP's identity. And if the career mode gets to serious with itself, heck, throw in a ludicrous contract like landing a heavy golden statue of Werner von Kerman on the far side of Mun. The way career mode is structured now just isn't very logical.

Let me put it this way: Why would KSP take orbital mechanics and Newtonian physics serious but couldn't find a way to have a more realistic career progression?

Edited April 29, 2016 by Rombrecht

[tater]

34 minutes ago, regex said:

I can't see a game AI "playing" KSP in any way, shape, or form like an actual human player, it would be more like events that happen at specific times.  I also think that could work to push the player if there were tangible reward/penalties for beating the deadlines but, as @HebaruSan points out, that will be very artificial and easily gamed, even if we allow "difficulty sliders" (god how I hate that term).

I don't think "Space Race" works with KSP.  It either turns into a railroaded conquest of artificial milestones or it is totally irrelevant.

Yeah, that's just another mode I think would be cool, hence me saying it might not even belong.

A novel way to do this would be for the AI to borrow from other players. You play your space race, I play mine, many others also play. Our crafts can be picked up (provided all parts are available to the AI's tech tree) and used by other games. It could also possibly sample player choices about upgrades and tech tree to make things more fluid. It would require a better tech tree, however.

In order for any sort of Space Race to ever work well, there would have to also be random failure modes. You'd have the ability to drive the chances lower and lower via use/testing.

[regex]

15 minutes ago, Rombrecht said:

Let me put it this way: Why would KSP take orbital mechanics and Newtonian physics serious but couldn't find a way to have a more realistic career progression?

Why would you expect overly-enthusiastic little green aliens to follow the same path into space that humans have?  If you want "realistic" there are tons of mods out there to make probes first, contracts make more sense, etc...  And while I totally agree that contracts should make more sense and career should have more management (a "lite management sim" would be awesome), I also want Kerbals to take their own path into space.

Human spaceflight is kind of boring right now.  We've put men on the Moon and we have a fantastic orbital station but we have progressed very little beyond that point; economics and general public disinterest are probably a large part of that but we're also very cautious with our lives.  Kerbals are, and should be, different.  No matter how much I enjoy the detail of realism and fully support mods like Realism Overhaul, I fully believe that Kerbals themselves should be treated as aliens with their own motives and path into space.

Edited April 29, 2016 by regex

[Rombrecht]

On the reason for a career mode

Let us go back to the basics: why does KSP have a career mode? The focus of the game is building spacecraft and getting out there to the planets. The management part of that process should not be a goal in itself, but a mechanic to limit the player's abilities early on in the game to built a rocket manned by 10 Kerbals that can fly to Jool and back with the first launch. The "career mode" is nothing but a mechanic to give the game some form of progression. Therefore I don't think that KSP should aspire to become a Tycoon-style simulator.

The science collecting mechanic technically does a good job at incrementally letting you build bigger and more powerful spacecraft. Though, as I pointed out at the beginning of this thread, the "science" model of game progression is quite unrealistic.

Do we need a complex game mechanic that involves a space race with some other space agency on Kerbin? I think not. Once again, the game's goal is to build your own spacecraft and get to explore the solar system with them. Successfully navigating and landing your spacecraft on another world is enough reward in itself. One does not need a competition with some rival AI space agency to spur you on to do great things in KSP. Sure, a rival space agency and space race would be a cute gimmick, but it should not be the focus of the game.