Random failures

[GoSlash27]

Just a silly notion that occurred to me:

 Suppose KSP had an option for a "Part reliability failure" slider. Maybe one of your engines fails to ignite. Maybe a reaction wheel seizes up.

 It would certainly add complexity and difficulty to the game.

 Not saying it should be enabled in the stock game, just thinkin' out loud.

Best,

-Slashy

[Superfluous J]

Having played with the "Dang It" mod, I do not want to ever have to deal with part failures, ever.

[tater]

Dang it! has failures happen far too often. I'd certainly use such a slider if it were available as long as testing/use improved the reliability over time.

[Atlas2342]

I definitely agree. Planning missions would be more interesting if you take into account what could happen during the flight..

[cantab]

As a general feature I'm not sure it's good, KSP is hard enough as it is.

Where I think it could work well is for part test contracts. The experimental parts would have somewhat randomised performance, with I think the possibilities being:

Nominal. It works like the standard part.
Underperform. The part works, but some aspect isn't as good as the standard part. For example an engine would make less thrust, a decoupler might not eject the part straight.
Overperform. The part works and some aspect is *better* than standard. For example higher thrust in an engine.
Failure. The part does nothing.
Catastrophic Failure. The part goes kaboom, possibly taking out other parts. Or in some cases perhaps the part gets jammed, for example a control surface gets stuck.

In all cases just running the test should complete the contract, but it would still add some interest and unpredictability. Do you plan to rely on the test part, which might go wrong, or haul it as deadweight which makes a more costly mission?

To compensate, make the dang part tests have more sensible and less nitpicky constraints.

[stibbons]

Do we really need to directly quote the what not to suggest list?

[pandaman]

Definitely not 'standard' stock IMO, but as I've said before, it may be a viable option for an 'advanced players difficulty' setting.

[nikokespprfan]

4 hours ago, cantab said:

As a general feature I'm not sure it's good, KSP is hard enough as it is.

Where I think it could work well is for part test contracts. The experimental parts would have somewhat randomised performance, with I think the possibilities being:

Nominal. It works like the standard part.
Underperform. The part works, but some aspect isn't as good as the standard part. For example an engine would make less thrust, a decoupler might not eject the part straight.
Overperform. The part works and some aspect is *better* than standard. For example higher thrust in an engine.
Failure. The part does nothing.
Catastrophic Failure. The part goes kaboom, possibly taking out other parts. Or in some cases perhaps the part gets jammed, for example a control surface gets stuck.

In all cases just running the test should complete the contract, but it would still add some interest and unpredictability. Do you plan to rely on the test part, which might go wrong, or haul it as deadweight which makes a more costly mission?

To compensate, make the dang part tests have more sensible and less nitpicky constraints.

For experimental parts, this might be a good addition (or maybe difficulty option). I always wondered why it states "experimental part" while you can be sure that it will work just fine.

Edited January 1, 2016 by nikokespprfan

[Waxing_Kibbous]

IMO you'd need levels of failure, easy would be something like solar panels, lights, batteries, antennae, science- things that are not critical, and most likely have multiple instances onboard anyway. Medium could have the above + reaction wheels, decouplers, things that one could still probably deal with if they had to. The advanced would have more serious things like engine failure. Just an idea, but I still wouldn't play it- sounds like rage quit fodder The key would be balancing the system in a fun way, randomness sliders would probably be the best way.

Edited January 1, 2016 by Waxing_Kibbous

[Matt77]

11 hours ago, cantab said:

As a general feature I'm not sure it's good, KSP is hard enough as it is.

Where I think it could work well is for part test contracts. The experimental parts would have somewhat randomised performance, with I think the possibilities being:

Nominal. It works like the standard part.
Underperform. The part works, but some aspect isn't as good as the standard part. For example an engine would make less thrust, a decoupler might not eject the part straight.
Overperform. The part works and some aspect is *better* than standard. For example higher thrust in an engine.
Failure. The part does nothing.
Catastrophic Failure. The part goes kaboom, possibly taking out other parts. Or in some cases perhaps the part gets jammed, for example a control surface gets stuck.

In all cases just running the test should complete the contract, but it would still add some interest and unpredictability. Do you plan to rely on the test part, which might go wrong, or haul it as deadweight which makes a more costly mission?

To compensate, make the dang part tests have more sensible and less nitpicky constraints.

 

6 hours ago, nikokespprfan said:

For experimental parts, this might be a good addition (or maybe difficulty option). I always wondered why it states "experimental part" while you can be sure that it will work just fine.

Good suggestion, in most cases I can't see random failures adding any fun - usually if something goes wrong, it will end the mission one way or another.  I like the idea of experimental parts not necessarily doing what they should - you can plan for that, you can't plan for a 150-part ship having a chance of failure on any part.  As nikokespprfan alluded to, it would also deal with the exploit where you accept a contract to purely to access a part before it's been unlocked.

Also the randomised performance should be decided before the mission begins, so you can't quickload and get a differently performing experiment if you don't like the outcome.

Maybe even, the performance could carry over to the production model when it's unlocked?  Not failures though, you'd have to assume they eventually got it working in the nominal state.  It would give another reason to test parts - you might get a bit more ISP, or slightly more power stored in a battery.  But you might lose out...

[tater]

Yeah, I like the experimental parts notion... change the tech tree such that you get the parts provisionally as experimental parts, then a certain amount of player-testing (use), and/or parts testing (contracts) will eventually unlock the part totally, at which point it's reliable. Perhaps unlocking X parts could be cheaper, and to totally unlock you need to pay the full price (so you take X parts for your mun mission to unlock some stuff past tier 1 at a great discount in science, but they might fail).

[TrooperCooper]

 

I'm considering to use the Test Flight mod... similar to Dang it! but with increasing reliability of parts the more you use them. Though I dont know how high the failure rate in general is... might still happen to often for my taste...

 

[Claw]

Unfortunately it's on the WNTS list. If you guys want to discuss DangIt or TestFlight, that's cool, but perhaps the thread will have to slide to another area then.

Cheers,
~Claw

[GoSlash27]

Oops!

 Sorry!

-Slashy

[AlamoVampire]

Le sigh!  Claw question for you, why cant we have a discussion on it? I really do not see why it cant be discussed. 

Edited January 1, 2016 by AlamoVampire

[tater]

To me this functionality would be most useful in a new space race mode of play, vs the current game.

[jf0]

I always thought this would be good, but definitely a configurable option. The way I imagined it is something like this: all parts, perhaps in the r&d menu or in the VAB would have a slider. When full, the part is very expensive, but very reliable, with minimum rare faliures that are not critical (example an engine could lose 50% of its thrust or shut down early, or a battery could lose some capacity/whatever)/ With the slider at minimum, the part is cheap BUT has a higher rate of faliure, and the faliures are more severe (structural faliure, explosion, gimbal stuck at full angle etc). This could give an interesting dimension to career: make money more critical and difficult to get, you need to balance your trade off between reliablility (manned missions, critical parts like engines on your booster need to be relaiable, but maybe things like the landing light not so much), and being able to afford your missions. It could also drastically change the way you build manned ships: you might need to consider redundancy, you'll need to plan your missions and every burn you are going to make and know exactly what is the point of adding a certain part, rather than just "I'll put this stuff incase I need it, I'll work it out when I get there...".

You could extend this furthur: as you use a part more often, it becomes more reliable, and if you recover a part you get a bonus to the reliablitity increase. You could also tie it into part availability: at the moment you can launch one rocket every five minutes with 20 mainsails and 10 nuke engines and 50 solid rockets if you want. In real life there is a limit to how fast they can be built. If you 'order' a more reliable part, its availability could decrease because it takes more care to build.

I really like the idea of a 'space race' career mode too.

there are so many opportunities to develop career mode and the gam in general, I think that now because the game has been out for a while, people are 'bored' of just launching rockets and really want new elements to the game, I know I do at least...

Edited January 2, 2016 by jf0

[MalfunctionM1Ke]

I am torn about failures.

Yes, they would add complexity to the game.
But, it would be really frustrating if your parts would fail just after you hauled everything over to Jool on your Jool-5-Mission wouldnt it?

Failures are a two edged-sword, and the Contrary-Side is much sharper than the other.

[Superfluous J]

13 hours ago, AlamoVampire said:

Le sigh!  Claw question for you, why cant we have a discussion on it? I really do not see why it cant be discussed. 

I believe the official rule isn't that it can't be DISCUSSED (which we're doing) but SUGGESTED (which GoSlash did in the OP). It can't be suggested because it's been suggested 4,213,291 times already and there is nothing new that can be brought to the devs' attentions. If we're just going to discuss it, it's far better to discuss the current mods that do it, or perhaps the creation of a new mod and what it should or should not have. "Discussing" if it should be in the stock game is pretty much "suggesting" it

[tater]

Failures without a context don't actually add much, which is why Squad isn't a fan of the idea. I would say that in a CONTEXT of an entirely new game mode, it's not a "suggestion" for addition to the stock game as we have to now  Lawyerly enough?

 

The context I have suggested is Space Race. Such a mode would pit the player vs an AI (or novel form of offline multiplayer?) opponent nation/company. Time would have to matter (a simple KCT, with a "cost" per ton in time to make any given craft?), then failures would be setbacks so that every space race game would not be the identical, "ideal" set of craft to get to the Mun first, etc. You can have a good design, but a bad day, and have to reconsider your plans.

[pandaman]

A thought...

IRL stuff breaks unexpectedly,  so the design solution is to build in backup/duplicate/redundant systems for the critical stuff. So far the only way to do this in KSP is duplicate the part...

2 probe cores instead of one. Twice as many fuel lines etc. Which increases mass and part count a lot.

Perhaps a work around could be included, such as - pay more funds, and in proportion to how much extra you invest, get a slightly heavier part that gets 'multiple lives' or a reduced chance of failure.

[tater]

2 hours ago, pandaman said:

A thought...

IRL stuff breaks unexpectedly,  so the design solution is to build in backup/duplicate/redundant systems for the critical stuff. So far the only way to do this in KSP is duplicate the part...

2 probe cores instead of one. Twice as many fuel lines etc. Which increases mass and part count a lot.

Perhaps a work around could be included, such as - pay more funds, and in proportion to how much extra you invest, get a slightly heavier part that gets 'multiple lives' or a reduced chance of failure.

That's actually a really interesting idea for a mechanic. Some probably significant mass increase. Take an orbital engine with redundancy (like Apollo), all but the engine bell has backup systems, engine mass up by 30% maybe, and gets a 2d chance if there is a failure. I really like that. Note that some systems can have things that engineers could repair (a use for skill). Another use for pilot skill might be to be able to deal with maneuver problems... say holding attitude with RCS/SAS when either (or both) of those systems have some failure. Say 1 RCS nozzle is not functioning, an RCS input using it would cause a spin, but the pilot can deal with it within some range based on skill.

Science instruments could be repaired by a scientist (cause they need some use for skill)..

[GoSlash27]

I like the concept of all of these things put together, but then it might get a little too "simmy".

 I just thought it would be neat to have the consideration in the design stage of accounting for failures with redundancy, definitely not something I'd want enabled in a standard game.

 But my apologies for even bringing it up. I had read the DNS list before and it slipped my mind that this was on it.

Best,

-Slashy

[tater]

Yeah, again, the primary utility, IMO, would be in some sort of space race. You could opt for improved reliability, or just go with what you have to try and beat the other side, but at some risk. What makes for fun gameplay are meaningful choices to make. Honestly, this is a primary problem I have with the mini-scale of the kerbol system. The Mun (given 2 moons of Kerbin) is more interesting when it is large enough that a 2-stage lander, or apollo-like (munar orbit rendezvous) vs direct approaches are choices that are not entirely aesthetic, but functional (direct works just fine in KSP, to the point it is wasteful to do otherwise).

[r_rolo1]

Well, it is discussible if we can't discuss the issue, especially when Harv itself chimed about the issue elsewhere .I do not find his argument very solid, though ( all in all, it resumes on his belief that people don't understand randomness, a proposition that ...well, does not show a lot of respect for the potential and actual buyers and players of the game, IMHO ) and it does not exclude part failure when it is expectable to happen, like when using experimental parts or when you're pushing the envelope ( we could have finally a use for the 100%-120% area of the throttle gauge  ). Oh well ...

Edited January 2, 2016 by r_rolo1
Grammar, my old nemesis ...