An Open critcism of the early model for the Science System.

[KSP orbiter]

Just throwing an idea out here, but would having different types of 'Science' help give a bit more definition to discoveries and the tech tree? For example you might earn different kinds of Science points for conducting different experiments. Things relating to chemistry ('red science'), or aerodynamics ('blue science'), or life sciences ('green science') and so on. Nodes on the tech tree would require differing quantities of each - a new command pod might require advances in electronics and life sciences, but you don't need any improvements in chemistry or aerodynamics.

This would also tie in to allowing the player to choose what experiments they want to run, in order to unlock the parts they want.

"Hmm, I want to build multi-stage rockets, but I don't have enough red and blue science to unlock decouplers... better run some experiments in those areas."

"Now I need more powerful engines to lift the multi-stage monstrosity that I've created! Or better fuels... which one to go for?"

[sprawl15]

For a science system to be really integrated in a way that makes sense, there's a significant amount of underlying gameplay changes that need to occur on top of just the valuation of transmission/return of data. To cover what's been said in the thread so far, I'd like to see a system where 'data based' science (like thermometer readings) could be covered 100% with transmission, 'materials based' science only being fully researchable with a return, a balanced diminishing return that allows some additional value from multiple readings, and a relevancy system where the data gathered assists with directly relevant unlocks. But here are some of the underlying, more system-oriented changes I think really need to happen to let the above blossom into a solid long-term science setup:

1) Robots first. This means that incentives need to be balanced such that manned spaceflight has to have riskier mission planning and greater rewards than unmanned spaceflight, as well as an incentive to send the robots before the manned missions. Right now, using an unmanned vehicle is generally more prohibitive than using a manned vehicle because of the electricity burned off by the robo-brain (especially in career mode where solar panels/RTKs aren't immediately available), and it offers no benefit whatsoever - only the drawbacks of no crew reports, EVA reports, or surface samples. On top of simple things like manned pods being an unlockable after unmanned controls, there needs to be an introduction of overhead costs for manned vehicles; rations, life support, etc. Probes should be the primary tool for initial exploration because of their low cost to maintain and their only resource being easily and directly renewable (electricity). This can obviously play into talk of things like hydroponics etc, but that's out of scope of the discussion.

2) Modularity within modules. Within a given module could be several different 'slots' for sub-modules. These could be as broad as things like Hitchhiker containers holding four slots that can be filled with seats or storage (for supplies, etc) or direct improvements like a science module being improvable with better hardware, providing a new fount of science. This would allow for individual modules to continue being useful within a specific envelope with only a relatively small amount of effort required to upgrade them and gather more data. This would be especially pertinent when money is triggered - you could have an extremely expensive space telescope that gathers/transmits huge amounts of scientific data but it eventually loses value due to diminishing returns. Later, instead of launching a new space telescope when you research better optics, you launch a small manned mission to improve the existing telescope, allowing it to gather much more science at a small fraction of the cost of launching a new telescope.

3) Modules as containers, stuff as objects. The two ideas above require not just modules that can store objects beyond a simple meter (100/100 snacks) but modules that can store discrete objects within a shared storage space (60/100 storage, with 40 snacks, 1 mystery goo sample, 2 mun rocks). This could work for abstract 'data' as well, letting you store different results on a hard drive while keeping the results discrete. This would serve a couple different goals: this would be a vehicle for the greater overhead on manned spaceflight (you would have to budget weight and physical space for snacks, etc), a larger amount of hard drive space can be baked into robo-brains while a larger amount of physical storage can be baked into manned pods (pushing probes' role as exploratory data gatherers), and it could open the door for 'consumable' items to be loaded up, like emergency extra batteries for the life support systems.

These three fundamental changes could allow for far deeper interactivity with science. For example:

Science lab modules that can convert samples to data for transmission. These could have a varied efficiency based on quality and sub-modules slotted. It would allow for a sizable portion of the 'return' value to be gained remotely, but would also require a significant amount of infrastructure to run (a space station, rather than hot glued to an OKTO2).

Research gathered from a specific celestial body can unlock sub-modules that can boost later research from the same body. You would be motivated to send a probe into orbit around Duna to gather data from orbit which would then be folded into 'extra research gained from Duna' sub-modules for the later manned mission.

Stuff like 'mystery goo module' could be changed to simply a 'what does a sample look like module'. Mystery goo can be loaded up in your cargo and you can run it through tests (exposing it to the environment, setting it on fire, looking at it, etc). If you haul up a dozen goo, you can run a dozen tests. This could also play into things like seeing what happens when you release Eve water into Jool's atmosphere, creating an incentive to have a large network of science gathering rather than having a single manned vessel touch down and leave with 100% of the science (and 100% of the incentive to be there).

Also, kind of unrelated, but a minor balance tweak that I'd like to see to make the rocketry side of research a bit more useful: when you switch up to bigger fuel tanks, there should be a weight savings due to economy of materials for the empty tank. Say, instead of a Jumbo-64 weighing 4 tons empty (which 64x T-100's would weigh), it could weigh 3, making it objectively better to use single bigger tanks over multiple smaller tanks but not so much that it would override staging/etc.

[Leatherneck]

*snip*

I like that idea, and would go a long way towards defining a purpose. Not only would it create dependency on doing "other" Science, at some point the player would eventually be forced to expand their research to their "less desirable" areas.

[Seret]

I think it would be better if you could only do an experiment once, and then you could save or transmit that experiment on your bay/goo container.

Nothing stopping you from playing it that way yourself, of course.

[boomerdog2000]

You are missing my point. There is no guidance. It does no better than sandbox, because are you put it, "it lets you figure what it does, and then introduces more advanced parts." The game does not tell you squat. All you get is a pile of parts, with a carrot of more parts, but no direction of how to get there. The game does not even TELL you how to GET science in the first place. It is all assumed.

The point of the tech-tree is not to tell you how to use the parts, merely to introduce them. The job of telling you how to use the different parts falls to tutorials which we have none of yet since the system has just been introduced.

You act as if they give you a pinewood derby car, and magically you can build a Bugatti at the end because they gave you pieces a little at a time. Just because I give you a few bits of a car at a time, does not mean you are going to build a Bugatti Veyron any better than if I gave you all the parts at once. Especially, if I give you parts for the brakes when the only thing you figured out "on your own" was how the damn windows rolled up.

More like they give you the pinewood derby kit (granted without the instructions so you have to google those for now), then once you manage to do some stuff with the pinewood derby car they introduce RC car parts, then go-cart parts, then car parts. They don't tell you how to use any of these yet, but the system for giving you the parts slowly does work (it does need balancing however).

The idea of multiple color-coded science types sounds cool and might fix the issue of having zero science specializing (like we have now) and having too much science specializing (like most peoples suggestions seem to lean towards). Combining this with the idea of choosing the science that goes into science bays and such could work nicely. You wouldn't add any more parts, just swap out the Mystery Goo or items in the Material Bay to give you red, blue or green science.

[KSP orbiter]

The idea of multiple color-coded science types sounds cool and might fix the issue of having zero science specializing (like we have now) and having too much science specializing (like most peoples suggestions seem to lean towards). Combining this with the idea of choosing the science that goes into science bays and such could work nicely. You wouldn't add any more parts, just swap out the Mystery Goo or items in the Material Bay to give you red, blue or green science.

I like the idea of giving the player more freedom to choose what sort of research they do, and being able to choose what experiments you put in each Science module for a given flight would be an excellent way to make them more reusable. The game needs to have a layer of abstraction over what is being done in order to make it a game, so I wouldn't push for too great a level of detail. The number of science types should be small enough that it's not overwhelming, but large enough that it either allows or forces the player to make meaningful choices about what they'll research.

I favour perhaps 6 or so fairly broad topic areas:

Electronics (Blue)

Aerodynamics (Yellow)

Materials (Orange)

Fuels or Chemistry (Red)

Life Sciences (Green)

Information Sciences (Purple)

It might be possible to have each part be unlocked seperately, but in any case the types and quantities of science needed to unlock something should bear some resemblance to what it is or does. New fuel tanks might require mostly orange science, with small quantities of red and yellow. The 'Hitchhiker Storage Container' would be mostly green and orange, with a little blue and purple.

That's just an example for purposes of discussion, of course. What do people think should be on the list of areas to research?

[LoSboccacc]

I like more the idea of splitting science.

It is too early for it as money is not in game but there should be mission based experiments and goals to fund your program

Then there should be part that do science which unlock more parts. Like, fly this prototype large engine to space and back and it unlocks the skipper. Fly the skipper to outside the kerbin soi and it unlocks larger thanks.

Things like that, whit the science missions guiding in what you do and what you get from doing that.

This would proide a guidance to new player and also will provide more tweakables for adjusting the game difficulty (the current tree only supports that as a cost multiplier.)

[Andrew Hansen]

I think it would be better if you could only do an experiment once, and then you could save or transmit that experiment on your bay/goo container.

Nothing stopping you from playing it that way yourself, of course.

Yup, and that's the way I do play. I don't like spamming science.

[UberFuber]

I like the idea of giving the player more freedom to choose what sort of research they do, and being able to choose what experiments you put in each Science module for a given flight would be an excellent way to make them more reusable. The game needs to have a layer of abstraction over what is being done in order to make it a game, so I wouldn't push for too great a level of detail. The number of science types should be small enough that it's not overwhelming, but large enough that it either allows or forces the player to make meaningful choices about what they'll research.

I favour perhaps 6 or so fairly broad topic areas:

Electronics (Blue)

Aerodynamics (Yellow)

Materials (Orange)

Fuels or Chemistry (Red)

Life Sciences (Green)

Information Sciences (Purple)

It might be possible to have each part be unlocked seperately, but in any case the types and quantities of science needed to unlock something should bear some resemblance to what it is or does. New fuel tanks might require mostly orange science, with small quantities of red and yellow. The 'Hitchhiker Storage Container' would be mostly green and orange, with a little blue and purple.

That's just an example for purposes of discussion, of course. What do people think should be on the list of areas to research?

I like this idea. A possible extension could be that you "earn" those science points through a randomized list of science missions (kinda of like the Kerbal scientist discovered something potentially interesting, and want you to plan a mission to carry it out). Once cash constraint is added, some of those missions could also include cash rewards. The mission idea could also make it an interesting challenge for player to optimize a launch mission to

For example.

Mission: Near Space Telemetry Test

Description: Engineers have a possible breakthrough in transmission protocol.

Activity: Transmit 50 Mits of data from Near Space without interruption (must have enough battery capacity and/or power generation to sustain transmission).

Rewards: 10 Electronics RP

Mission: Zero-G Material Experimentation

Description: Scientist have developed a possible material with unique property in microgravity environment.

Activity: Start experiment on SC-9001 Science Jr, maintain zero-G condition on spaceship for 30 seconds.

Reward: 10 Material RP.

[RadHazard]

One thing I'd like to see, and I think they are already considering, is the ability to research improvements to parts. I think it'd be interesting to decide things like whether to research more efficient versions of the engines you have now or developing larger engines to lift heavier payloads. It could work well with the multicolored science that's been mentioned, and it'd interact well with money when that gets added (Choose between more efficient rockets to save money or larger rockets to carry more valuable payloads).

I'd also like choosing experiments to be a more difficult process. Right now, if I can launch a Mystery Goo canister and a Science Jr. somewhere, I can slap on the rest of the scientific instruments without worrying about anything. I'd like it if there were more heavy experiments to carry around. It'd also make sense if scientific instruments had a heavy power requirement. Managing your power budget on early missions before you get solar panels is fun, as it adds a sense of planning. If the more advanced science instruments had a constant power draw that they needed to work, it would encourage either sending smaller probes with a few solar panels or much larger ships with the power generation capacity to run more at a time. Additionally, if some instruments need power to run, it could be changed that those instruments accumulate science over time rather than all at once, which would make establishing an orbit around a body more valuable than a flyby.