[WIP] [Dev thread] Dang it! A random failures mod

[Ippo]

Yeah, that would be very cool but I don't even know if it can be done at all. However this window is available in all scenes, so you can check stuff in flight too, which is handy for repairs. If I ever find a way to hack into the astronaut complex though I surely will

[Ciber]

How about "Hard" and "Soft" fails. You get notified when hard fails happen and they generally result in full loss of function. Soft fails however will not be reported and will be harder to tell. Think things like your tank springing a leak. You would only notice soft fails by right clicking a part. Tech should play a role. Experimental parts should have a much higher chance. Perhaps keep a count of the number of ships that have contained a given part, and lower chance of problems happening as this increases. Testing contracts should significantly reduce failure rate.

[Ippo]

How about "Hard" and "Soft" fails. You get notified when hard fails happen and they generally result in full loss of function. Soft fails however will not be reported and will be harder to tell. Think things like your tank springing a leak. You would only notice soft fails by right clicking a part.

This distinction is not really possible in my framework right now. Well, kind of: a failure module can easily define multiple failure modes, but they all share the same properties (MTBF, Life time, and Silent).

In reality, it's possible to achieve the same result by defining multiple failure modules for each part, although it's not very clean for some other fine points.

I'll see when later on I'll add more failures.

Tech should play a role.

Agreed, but I plan to do it only in the sense that the later a part is unlocked, the better it is. Parts won't get buffed by unlocking new nodes, because it wouldn't make much sense to me (imagine that SpaceX creates a new super-cool engine and their research retroactively improves the engines of the shuttle... not really plausible).

Experimental parts should have a much higher chance.

Agreed.

Perhaps keep a count of the number of ships that have contained a given part, and lower chance of problems happening as this increases. Testing contracts should significantly reduce failure rate.

Disagreed: I think I already explained while I don't think it's a good idea somewhere in this thread, however it boils down to the fact that spacecraft components are not mass produced, or, at the very least, not in large enough quantities to make a significant difference.

I will also go ahead and confirm that I won't be implementing a "quality" system that will let you pay in exchange for better stats: funds are anything but scarce, so you could just toss money at the problem until it goes away and I don't think it makes for good gameplay.

[RobotsAndSpaceships]

Explodin' engines and parts of the sort. It would give abort mechanisms an actual use. Loving this mod.

[Ippo]

I want to add that too in the end, but disabled by default for those that don't like it

[Ippo]

Success! For the first time I was finally able to upgrade a perk using the GUI

It still needs a lot of polish, but most likely it will be ready for 0.25, or shortly after (unless they pull the rug from under my feet and release a lot sooner than I expect them to).

[Athlonic]

Great !

Well done looking forward your next update.

[Ippo]

Hey everyone, I need some feedback about balancing

So, I need to set how much it will cost to train an old kerbal new tricks: I have taken this as a starting point:

TRAINING
{
    // Cost of training (science:funds)
    Unskilled = 10:10000
    Normal = 50:50000
    Skilled = 150:150000
}

Each line is the cost in science and funds that you will have to pay to train him from the previous level. You can't skip levels, but have to get through all of them serially: this means that to go from Untrained to Skilled, you will have to shell out

• 10 + 50 + 150 = 210 science points
  
• 10k + 50k + 150k = 210k funds
  

I'd like some feedback about the values. Please keep in mind that they are supposed to be *VERY* expensive, to make sure that a Skilled astronaut is a real investment. ~200 science points seems fair in my experience, with the absurd overflow of science that we have. I have no idea about funds though: I can't stand the contracts as they are right now and thus I never played a career, I'm not sure how much it is.

P.S: I am also happy to announce that I have officially completed* the perk system!

* May contain unknown amount of bugs.

Edited October 1, 2014 by Ippo

[Tripod27]

Hey is there any way to stop time warp as soon as a failure happens? I really hate traveling at any decent level of timewarp and by the time I notice the little antenna symbol in the top right saying that I have a notification (and that notification being that my tank is leaking) my tank has already been completely leaked for hours. You could just make it stop timewarp whenever ANY notification happens if you can't figure out how to narrow it down for just your notifications, since they usually only pop for important things you want to check anyway like completed contracts, and it doesn't pop often either

If you're up for it you could even talk to the guy who made the chatterer mod, and see if one of the chatter effects (warning alarm going off followed by kerbals frantically yelling) can play whenever one of the parts fail as well. Just things that will make it more noticeable that something bad has happened while also increasing the "experience"

edit: also for the balancing feedback, the funds section is alright since you said do want it to be expensive, I don't really agree with the science cost since it doesn't make sense, but I guess if you just scour every biome on every planet and also have other science mods installed that add even more science experiments to the pool then yeah you will have a ton of extra science

Edited October 21, 2014 by Tripod27

[Ippo]

Hey is there any way to stop time warp as soon as a failure happens? I really hate traveling at any decent level of timewarp and by the time I notice the little antenna symbol in the top right saying that I have a notification (and that notification being that my tank is leaking) my tank has already been completely leaked for hours. You could just make it stop timewarp whenever ANY notification happens if you can't figure out how to narrow it down for just your notifications, since they usually only pop for important things you want to check anyway like completed contracts, and it doesn't pop often either

That's very weird: timewarp stops when a failure occurs, it has for a long time (since alpha 3, precisely). Can you try to reproduce the issue? Has anybody else encountered this problem?

If you're up for it you could even talk to the guy who made the chatterer mod, and see if one of the chatter effects (warning alarm going off followed by kerbals frantically yelling) can play whenever one of the parts fail as well. Just things that will make it more noticeable that something bad has happened while also increasing the "experience"

We already talked about it some time ago, we decided to postpone until after he did the integration with remote tech. Considering that now we got to that point, maybe it's time I get in touch with Athlonic again

edit: also for the balancing feedback, the funds section is alright since you said do want it to be expensive, I don't really agree with the science cost since it doesn't make sense, but I guess if you just scour every biome on every planet and also have other science mods installed that add even more science experiments to the pool then yeah you will have a ton of extra science

IMHO, the science cost is even more important than the funds cost, since science points represent scientific advancement in the game and so you can "spend" them to obtain more knowledge for your kerbals.

Anyway, feel free to set the science cost to 0, it's just a matter of changing the Training.xml file (it's in DangIt/PluginData/DangIt/Training.xml), you can do it with any text editor

[MasseFlieger]

IMHO, the science cost is even more important than the funds cost, since science points represent scientific advancement in the game and so you can "spend" them to obtain more knowledge for your kerbals.

Anyway, feel free to set the science cost to 0, it's just a matter of changing the Training.xml file (it's in DangIt/PluginData/DangIt/Training.xml), you can do it with any text editor

Quick suggestion, may make some sense:

- how about the science required to train a kerbal to a certain level is a one time science cost? That is to say, once scientific achievements are reached, the cost of training the next kerbal for the same level should be drastically reduced since the scientific progress isn't forgotten...

Example would be training one Kerbal once to a Skilled or Professional level of Mechanic, and that costs 100 science the first time, and subsquent training for the same level can be a fraction of that cost since, say ((original science cost)*0.25)/number of trainings performed for same level

Example with Numbers:

1. training kerbal to skilled level of Mechanic = 100 science

2. second kerbal taking same traning = (100*0.25)/1 = 25

3. third kerbal taking same training = (100*0.25)/2 = 12.5

and so on.....

Either way, I love the mod, and its a must for me on my save.

Edited November 4, 2014 by MasseFlieger
typos

[Ippo]

Quick suggestion, may make some sense: [...]

Mmm, interesting, but how could it be justified? I have to say it, I am more of a realism player: I can't really see an obvious reason for this mechanic. I'm not saying no, I'm just wandering why cost should decrease over time

[MasseFlieger]

Mmm, interesting, but how could it be justified? I have to say it, I am more of a realism player: I can't really see an obvious reason for this mechanic. I'm not saying no, I'm just wandering why cost should decrease over time

Hi Ippo,

Not funds cost, but Science, I'm making the assumption that the science amount is related to the effort in scientific research that KerbalKind had to achieve in order to offer such a level of training. If that assumption is "correct" then the scientific progress is already there once it's achieve the first time (say when you pay for the training of the first Kerbal).

Hope that made sense.

[eRe4s3r]

I am probably going to make no new friends with this post I am sorry in advance.

Is there some way to have random failures in a more.. uhm..sorry to say... more random way? Because MTBF vs lifetime dependent failures are not random failures. The chance for hardware fail at end of MTBF is exactly as high as it is at start of MTBF, but the random CHANCE of failure exists in a linear fixed fashion constantly and always. So this mod is kinda not doing what it's title would suggest, it adds a hardware decay mechanic with chaotic failures over lifetime of a part. That is not random In fact the chance for a failure is 100%

I was looking around for a failures mod that basically adds some constant tension but randomly based on my play-time in real life, not based on the in-game time or the part lifetime. (Mainly because I played that Buzz Aldrin Space Program Manager, and I really like RANDOM failures that are not predictable ,p)

Ie, something like this

Random fail chance 5% base value

Checked every X minutes real-time (not in-game time, actual, real system time) spent focused on active vessel

When check occurs, 5% chance 1 random part fails, 2.5% chance 2 random parts fail, and 1.25% chance 3 random parts fail. (amount of failures with such linear degrading chance an setting, ie 1 to "all", and base value as well as linearity as a setting (ie 0 to 99% chance on first part, and a reduction gradient setting in percentage, ie from 0 to 200% (so it can be 5 0.5 0.05 or 5 10 20% chances) this would allow a completely modular way to have random failures. And everyone could set it as they like.

Mechanic skill reduces those percentages in fractions. ie.

100% skill on 1 kerbal in vessel would reduce fail chance to 2.5%, 1,25% and 0.75%

100% skill on 2 kerbal in vessel would only reduce fail chance to 1.25%, 0.75% and 0.375%

Though in all honesty I think this skill should be something we upgrade somehow at the base. R&D quality or something that defines the base value of failures for all parts and that we can improve by hiring more scientists and engineers or training scientists/engineers (not the kerbals in the actual vessels)

Option for fail chance to never fall below 1% for 1 part failure over X minutes spent on active vessel (failure chance ceiling) too ;p

I admit this is not really what the mod is currently doing. And as I said, I don't want to be a bother to anyone. If this is impossible or outlandish ignore it ,) But I read suggestions should go here, and I would love a "simple" random failures mod ;p

[Ippo]

...

Well, to each his own I guess

I will object to your statement that the chance of failure is 100%: it is only for time approaching infinity, which of course you can't do, and in fact a lot of LKO and Mun missions are so short that more often than not I come back without a problem. But I digress

While I don't really understand why you'd want such a model, I'm afraid that it just can't be implemented in my framework without rewriting pretty much all of it... sorry (Still, DangIt is free software so feel free to fork it, or to find someone who will for you )

[Ippo]

Hi Ippo,

Not funds cost, but Science, I'm making the assumption that the science amount is related to the effort in scientific research that KerbalKind had to achieve in order to offer such a level of training. If that assumption is "correct" then the scientific progress is already there once it's achieve the first time (say when you pay for the training of the first Kerbal).

Hope that made sense.

Sorry... not really convinced about it, to me it seems it only adds complexity with no real benefit for the gameplay :/

[eRe4s3r]

Well, to each his own I guess

I will object to your statement that the chance of failure is 100%: it is only for time approaching infinity, which of course you can't do, and in fact a lot of LKO and Mun missions are so short that more often than not I come back without a problem. But I digress

While I don't really understand why you'd want such a model, I'm afraid that it just can't be implemented in my framework without rewriting pretty much all of it... sorry (Still, DangIt is free software so feel free to fork it, or to find someone who will for you )

That's quite alright Sadly I have the coding skill of a brick, I am artist AND I am very lazy when it comes to anything modding ,p I am using Dang-It currently (Except I adapted all lifetimes and MTBF) so I can't really complain and it does more or less what I want (Except now chance for failure is very very low if my math is right) Anyhow, I edited the values (of all parts) like this

MTBF = 350400 (per year failure rate around 2.5%)

LifeTime = 131400 (15 years)

So essentially, per year in-game time there is a 2.5% chance of failures.

Maybe overkill? But does what I want Anyhow I am happy ^^

[Ippo]

Maybe overkill? But does what I want Anyhow I am happy ^^

That's all that matters

[Ippo]

Well, well, well, look what has cropped up! Failures in background are now not only a possibility, but a very possible possibility

We'll need to talk balancing though. I'm thinking about a "soft mode" where the chance of background vessels is greatly reduced (ideally, dynamically: dividing it by the number of flying vessels), and a "hardcore mode" where everything has the same chance, regardless of being loaded or not.

More news in the future when I find the time / will to keep coding this mod

[klgraham1013]

That could impact a great many mods I use. Looking forward to seeing what you do with it. Fingers crossed that your perks system is coming along, as well.

[pjf]

A quick suggestion if nobody else has made it...

Procedural Parts provides both regular tanks and balloon tanks. The latter are lighter, but much more fragile, and collapse if they lose pressure.

This would be a perfect interaction for DangIt. Balloon tanks can have a higher failure rate, and those which fail can leak at a higher rate. Balloon tanks that are empty of fuel and are leaking have a chance of deconstructing under powered flight or if landed, as the tanks lose structural integrity. (Although I haven't thought through all the deconstruction scenarios, so I'd consult someone with more rocketry experience for actual conditions.)

[NathanKell]

For a while, the scenarios included "if you looked at it funny."

Also, since balloon tanks involve common bulkheads, leak will usually equal explosion.

[lextacy]

thinking about getting this mod, is it stable?

[Ippo]

Procedural Parts

Actually the tanks in procedural parts are not really tanks according to KSP, they are a part that doesn't contain any resource and thus my MM patches don't catch them at all: if you wanna cheat my mod, use procedural tanks

Likewise, proper integration with procedural parts will require to write "procedural failures" that adapt their parameters on the tank volume...

thinking about getting this mod, is it stable?

Kinda: it's been a while since the last bug report, so I'd say that *this version* is stable in itself. On the other hand, it's knee deep in alpha, with missing features and very likely to cause save-breaks in the future (I actually am already planning one for Alpha 6, which is funny considering that I might have to quit and hand development to someone before Alpha 5 is even done).

For a while, the scenarios included "if you looked at it funny."

For a second I thought you were replying to lextacy, then realized the order of the posts

[Daishi]

...which is funny considering that I might have to quit...

Thats not very funny at all Be a real shame if you had to walk away from this, you've done (destructive) wonders with this mod and it just keeps getting better. Cling on! Cling on dammit