Random failures of experimental parts

[M3tal_Warrior]

It's adorable you think people read.

I know - I somehow require people to actually read stuff, despite seeing quite a lot of them acting like typewriting apes (e. g. bug reports for modded installs despite the sticky). OK, I tend to get nasty, sarcastic and ironic towards people too stupid to read, but that's just my way to give the needy a head-butt reminder.

Well, never gonna give up.

"Run test" can.

he meant you can't do an EVA in flight, due to atmospheric drag acting on your Kerbal blowing him away.

Edited August 4, 2014 by M3tal_Warrior

[fairytalefox]

he meant you can't do an EVA in flight, due to atmospheric drag acting on your Kerbal blowing him away.

Oh, it seems like it's me who lacks some essential skill :/ What a shame.

Well, those "in flight" tests just should not have an "inspection" condition, that's all. Or the devs should implement some sort of "inspection" available in atmospheric flight. I don't see any major problem here.

[vexx32]

With the consideration that the parts are tagged explicitly as experimental in the first place... this could well be the only place I would actually approve of having random failures implemented.

My reasoning is thusly:

1. The part is, clearly, supposed to be experimental, and thus should not necessarily function perfectly.

2. With current implementation, you can accept a contract, and abuse the granted experimental part to give you a leg-up in regards to achievable goals when you're still in the early stages of the tech tree. Making experimental parts unreliable would curb this abuse, making it difficult or even pointless to try to use the contracts in this manner (which, I feel, isn't intended anyway).

3. Could conceivably cause some awe-inspiring havoc.

4. Since the contract asks you to test a part in a specific situation, the results may vary depending on the situation instead of being completely random. For example, contracts to test a jet engine in vacuum might have a small chance of the flameout causing an explosion big enough to destroy the engine. Experimental launch clamps could have weird failures such as not detaching, explosive detaching, etc. Experimental nuclear engines could well be spectacular. The list is endless.

5. Such a mechanic could also force you to choose more carefully which Kerbals you send on a mission, if the mission involves testing an experimental part. It may even be prudent to use unmanned flights when possible, simply because the experimental parts would be too risky to let Kerbals near them.

The extent to which it is implemented is of course entirely up to Squad, and they might feel it better to not have random failures at all... but given the above points, I personally rather like the idea of having the experimental parts be somewhat unpredictable -- that is generally how a prototype behaves, after all. Sure, you know to a certain extent what it'll do, but you never truly know what it can do until you give it a test drive.

This suggestion, I quite like. Have some rep, OP.

I've also noticed people in the thread getting a little snippy. Please keep things civil, folks.

[fairytalefox]

the results may vary depending on the situation instead of being completely random.

This sounds like a good idea, too.

[Kerbart]

Can't be done in flight.

I know. Which means that all those tests you have to do at various altitudes now have to be on a manned craft that makes it into orbit. Instead of an unmanned "straight up, straight down, and likely to land on the Kerbal Space Port and get me 90% or more recovery Funds" contraption.

[vexx32]

Forcing a EVA Kerbal to present for a test may be a bad idea, I should think, if only because if the part is going to explode spectacularly, your poor Kerbal is either fried to a crisp, or knocked into a wildly different orbit (or just flung a few hundred metres overland, if you're landed on a planet).

[Stargate525]

I know. Which means that all those tests you have to do at various altitudes now have to be on a manned craft that makes it into orbit. Instead of an unmanned "straight up, straight down, and likely to land on the Kerbal Space Port and get me 90% or more recovery Funds" contraption.

But I don't explicitly see what's wrong with having a jet plane with a payload capacity to take a part to the specified altitude and speed, test it, and return it. For the big parts, figuring out how to fly them is a challenge in itself.

[fairytalefox]

Forcing a EVA Kerbal to present for a test may be a bad idea, I should think, if only because if the part is going to explode spectacularly, your poor Kerbal is either fried to a crisp, or knocked into a wildly different orbit (or just flung a few hundred metres overland, if you're landed on a planet).

This could be an idea for another suggestion. Besides experimental parts, we have contracts to test some already unlocked parts, too. It would be cruel to make these parts unstable.

I'm not sure if this thread is a good place to discuss this idea, though.

[makinyashikino]

this would make it exciting when people use contracts only for early access to later tier parts

I do believe that is an issue, as if you are asked to test the nuclear engine, or one of those massive engines really early on. GG game becomes a cakewalk.

[Drachma]

I personally like this idea, if your gonna fly to the mun, why would you use "experimental parts?" Unless told to, I haven't gotten a mission like that yet so not sure . and having this would give experimental more of a meaning and would make you think twice which I like and having every thee thing wouldn't make sense because of contracts and other things that could happen.

[aleis]

I'm against the idea. With the crazy requirements you some times get I try to make the best by using that par to do something.

For example I needed to haul the large srb to orbit and test it. So I strapped a lander to the top with a little extra fuel and let the test part do most of the interplanetary burn. But now it might fall to actually go and I'm now going to be losing a ton of funds because the expensive lander is just going to orbit Kerbin for fun

And squad said they want people to be able to use the parts on rockets so you can see how cool it will be when you actually get to unlock it

[SkyRender]

I think a less cruel version would just be to make it so that experimental parts have slightly tweaked parameters out of spec (ie. 10% heavier and 10% less effective than the non-experimental version). The added challenge of experimental parts should be adapting to their limits, not in testing how patient you are with Revert to Launch.

[Whirligig Girl]

If this is done, experimental parts should have a red part tab instead of blue

Yeah, I like this idea. a failure of an experimental part should also count for science, rep and money, just not as much. A flight test failure is just as useful as a flight test success.

[Stargate525]

And squad said they want people to be able to use the parts on rockets so you can see how cool it will be when you actually get to unlock it

[citation needed]

It's also a bizarre rationalization. That works precisely once. After that, it's just a 'get this part earlier, for free' button.

[John FX]

How about giving you access to ALL the parts early?

The caveat being that parts you use earlier than the node they are on (only parts you can see would be available so next node only) have stats worse than the fully developed part and they have a chance of failing.

For example, on tier 0 you would have access to tier 1 parts but they would run -10% on all stats (+10% for weight) so a fuel tank would weigh 10% more and hold 10% less than the standard part and have a 5% chance to leak (1-5 units/minute of either LF or OX). In a similar way a solar panel would be 10% heavier, output 10% less and have a 5% chance to drain 1-5 electricity/min from your batteries.

A decoupler would have a 5% chance of not decoupling (and would weigh 10% extra)

And so on. Note that `failure` is not a digital thing, it causes minor but problematic issues for the craft and so would very likely shorten a mission but not destroy it completely. Also you don`t get the shiny modern part free early, you get an undeveloped heavy unreliable part early, just like you should.

This could also be used for the current experimental parts implementation in contracts too.

This mechanic would be good if we have to buy a node then gain science to use the parts, the parts could be used right away but gaining enough science (and full node unlocking) would bring the part to full stats and a lower weight and reliability.

Then you could take a 1 in 20 chance on an untested part and even if you get a dud your mission might just get hard instead of failing.

Maybe you should have to run tests on ALL the parts to make them (more) reliable?

Maybe not.

EDIT : Just another thought.

With the current system I always take experimental contracts and always try very hard to not complete them so I keep the part until I unlock the node it is in. I think this should be discouraged in the mechanic. It would seem obvious that one way to do it is to make the part you get significantly worse than the part you will eventually get when you unlock the node, there may be other ways but they do not immediately jump out at me.

As has been said, it is basically a `get parts early and free` mechanism and the ways to disrupt that are either make it not free, make it not early or make the part not `good so the actual part is still desirable.

Edited November 14, 2014 by John FX

[Technical Ben]

Perhaps a much more understandable option would be variable test results from experimental parts. So they all perform (no random explosions), but science recovery amount and comments vary.