SCIENCE! Rethought

[Captain Sierra]

So, this is gonna be long. If you don't feel like reading, please refrain from commenting. This will be mildly ranty and possibly a bit disjointed and incoherent. I'm pretty much pulling all this crap off my head, where it has been stuck for about three days. If something needs clarifying, just ask nicely. These are based off a lot of what I hear, as well as my own personal opinions and suggestions. I do not claim to represent the interest of everyone, but given the similar opinions which I have heard, I do claim to hold an opinion similar to a significant fraction of the community.

---

So, science. What do I think? I think it's too monotonous. I think science is an exploitable clickfest which allows a skilled player to clear the tech tree in less than five missions (current record stands at 3). I'd like to see science become more meaningful. Does that mean more difficult? No! Science difficulty is decent, however, I think the system by which it is implemented allows way too much in the ways of exploitation. Now, I will touch on a concept here called complexity scaling. In a nutshell, it involves an approach to difficulty which makes higher difficulties use more complex mechanics (larger, more varied resource trees, more realistic and challenging tech trees (rearranged, ScienceDefs remain the same), and possibly more part varieties. The most relevant part of that concept (which I'll write a suggestion on later) is more science tools. But I digress.

How do we improve science? A handful of ways. For starters, let's get into biome sensitivity. Biome sensitivity is little more than an exploit in it's current implementation. How so? Because experiments can benefit when they shouldnt. This brings me briefly to another topic.

Experiment type. For now, I have come up with three. Point data, local area, and visual. Point data is pretty straight forward. Point data is the type which all sensors fit into. These types of experiments produce numerical data. Numerical data science cannot be increased by laboratory analysis. It can always be transmitted at 100%. Local area are your materials lab, which I have some thoughts on later, the surface sample, the atmospheric analysis package, which I also have some ideas for, and the goo canister. These can have their science yield increased by laboratory analysis of results, which is another topic. Lastly, is visual. These are your crew reports and EVA reports. As you can see, the majority of experiments fall into point data and local area, which are only effected by the immediate environment. Orbit is orbit, regardless of what you are above.

Now, this brings up some complications. Some instruments which should be stock, like the DMagic magnetometer, would return the same readings over the ocean as over the desert, but may return different readings over the poles and possibly the mountains. The solution I see is to set up instruments with a ScienceDef which corresponds to all biomes it will pull that reading from, so that a magnetometer would pull the same reading over kerbin highlands, plains, desert, oceans, etc. but pulls a different reading over the mountains, and yet another different reading over the tundra and poles. In this respect, certain instruments can gain limited biome sensitivity without being overly exploitable.

Local area experiments, which point data is realistically a subset thereof, lose biome sensitivity (for the most part, as explained above) at the upper atmosphere, which starts at either 12000m or 32000m, I do not remember exactly. Biome sensitivity is only retained in the lower atmosphere. Beyond that, 'biomes' become altitude layers encompassing the entire body. There is upper atmosphere, low orbit, and high orbit, between which I feel the dividing line should be somewhere around a few thousand kilometers (it may be good as is, I do not remember where it actually sits). Local area experiments provide considerably more science. The atmosphere nosecone clearly would not work beyond the upper atmosphere.

Visual experiments are tricky. I feel like EVA reports and crew reports are two versions of the same thing, furthering the clickfest exploitation of science. I feel though that they are both necessary as, in certain circumstances, the results yielded would be different. One such scenario would be when landed on a planet. A report inside the capsule would be extremely limited compared to that of one taken outside the craft. I do not pretend to have all the answers and would like some alternative input on how to handle EVA reports, as I currently do not have a good solution.

Instruments. I think certain experiments need reworking. I'm going to start with the goo canister. Rather than the warning about inoperablility, I'd like to see some more immersion which conveys the same information. Here are some examples. When the goo canister is open in the lower atmosphere, some goo could leak out. In the upper atmosphere flying, some goo could be sucked out and blown away towards retrograde. I'm going to assume that while closed, it is retaining kerbin atmospheric properties. When opened in space, some of the goo can be sucked out by explosive decompression, freeze, and fade away. This would make it more clear visually that the experiment is single use. Also, I'd like to see different effects when the data is removed. Perhaps the screen cylinder can be removed. The same thing goes for the materials lab, but more on that later.

The atmospheric analysis package seems to be flawed. I think it should be redesigned as a piece which takes an atmospheric sample rather than runs full analysis. This would make laboratory analysis later believable and profitable.

Surface sample. Give this an animation please!

Materials lab. Right now, it seems to be little more than a stackable, glorified goo canister as it functions identically. I think it needs to be somehow set apart as more valuable, or else it becomes the EVA report to the crew report that is the goo canister. My suggestion would be to make it multi-experiment capable. It has a computer bank in there, so perhaps it could run some data based experiment, which is endlessly repeatable, and it could run a physical experiment, which would be consumed on use and could be laboratory analyzed. Also, the same animation suggestions about the goo canister apply to this too.

Laboratory analysis. I keep bringing this up, but there is more to it. I think whenever science is returned to kerbin, it is automatically analyzed by R&D, and should yield full science value. I think that since local area experiments are not transmitted at 100%, they can be analyzed in the mobile lab, and then transmitted back to kerbin for full science value. This would make the mobile lab clearly practical for missions which return is not planned, at least not for a while. This however, is a contradiction, since it takes kerbals to do the processing. Most people like to return their kerbals, yet the role of the lab in the above scenario is clearly designed for missions not returning. Again, I'm not all knowing. I'd like other input on this to help figure out the best course of action.

Probes. Probes need love. For starters, I'd like to see pretty much all of the parts in RLA Stockalike go stock. This would give a lot more options of for 10 ton or less lifters and probes. It's full of great parts, is well balanced, and offers a ton of variety in an already near-perfect stock styling. For more orbital science, which probes are some of the best at, the DMagic Orbital Science pack needs to go stock. This adds two point data instruments and one visual. More point data experiments is always good to enhance probes. Since they, by this suggestion, transmit at 100%, it makes the return journey not worth the extra launch mass. Sacrificial probes would be perfect for doing lots of this, and RLA stockalike (and to a lesser extent, RLA power generation) gives them some great tools to get where they need to go to do science.

Now, point data can't be what probes are limited to. There was the Viking lander, as well as the multitude of rovers we have sent to Mars. I'm going to suggest creating a few tools for probes. First, give them a camera. I always think of a predator drone style camera as it gives good range of motion with decent resolution, as well as looking somewhat in style. This camera would let them take pictures, as many as they want. if a picture of something has already been taken, they earn diminishing returns (minus at least 50% science value off the current for each picture already taken). This gives them something to compete with crew reports. Another option I see is to make modified versions of the current experiments, particularly the atmospheric tool, in probe size. An atmosphere scoop could then feed into a probe processing unit. This tool does the same analysis as the mobile lab, allowing experiments to be transmitted for more. It requires a good ammount of power, but will not get it to 100%. I think a fair number would be 85%. There should also be a tool for probes to take surface samples, making rovers great tools. All in all, this gives probes a lot more of the respect they deserve.

Tech tree. Now I'm gonna get a bit more into that difficulty-complexity scaling crap. I think when starting a career save, you should get a popup. This window would ask you several things, some of which I'll mention in my difficulty-complexity suggestion later on. One of them would be tech tree type. It gives several choices. So far, I've thought up 3. Default, Realistic, Ground.

Default. Clearly, this is the stock tree as it is now.

Unmanned. This starts you with probes. It quickly works you into manned, but your first mission to orbit, your first Mun flyby, and your first interplanetary transfer will all likely be probe based. At least, the tree encourages that. This tree should be balanced more in line with real world technology while still retaining playability (but you wouldn't get 2.5m upper stage parts before lower, and clustering engines would be cheaper then huge mainsails).

Ground. This tree would focus on starting with aircraft and wheeled vehicles. It forces you to do some kerbin exploration before you can leave your parent body and go elsewhere. Beyond that, it's balanced realistically, much like Unmanned.

Station science. I have a whole list of ideas for this segment but I'm pretty burnt out right now, so I'll edit this in later. Be sure to check up when I do that, as it may ask for more advice. Stations are a complex animal.

If you read all that, good on you. If you read all that, you may now discuss, suggest amendments, and otherwise share your thoughts and possibly lend some solutions where I could not come up with any. And that's all I have for science.

[TADBTD]

<snip>

AMAZING. NEEEDS TO BE IMPLEMENTED. I have an idea. What if when the currency is added, that you expand the R&D center and it's supplies to gain better quality data.

Edited January 24, 2014 by Supernovy
large quote snipped

[Captain Sierra]

AMAZING. NEEEDS TO BE IMPLEMENTED. I have an idea. What if when the currency is added, that you expand the R&D center and it's supplies to gain better quality data.

Now how come I only have one response if the idea is so good? Am I scaring people off with the wall of information?

[jlrdsr]

Yes these are all good idears! Id like to see some implemented. Also mabey the goo canister and meterials bay could be refillable, where once you have used it, and the goo has floated off into space, you can refill it with more goo from a storage canister. So it would be reusable, but still finite. And for the metirials bay the same as the goo and you can also carry more than one type of sample, which give different reasults, but you have to carry the extra samples. Also MOAR INSTREMENTS! The more the better! I want soo many i cant fit them all in one space craft! And they would also fit into groups, ground, atmosphere, space, maby more, so that you need to send a mutittude of craft to truely get all the possible science. Point data should also varry inside the same biome, like if your near a rock, or at the top of a hill e.c.t. I also would like a science analasis computer, where you run say 3 experiments, and then imput them into a computer that then gives you more science based on a various conditions. It would work on the basis of comparing more than one peice of info gives you a better understanding than just the origonal peices sepratly. But if you use the same experiment more than once it wouldent give you more than a tiny bonus. And a system that can imporve the transmition quality, like an interfearence detector or a antenna power booster, and a camera that sends the kerbals at the RnD a some vedio footage rarther than just some report data, and returning kerbals back to kerbin allows them explane how something happened rarther than just the transmited report. Phew thats quite a wall of text.

[Captain Sierra]

The lab mechanic could be changed to refill rather than just reset experiments, making it more in line with what I talked about. At it's core, the mechanic remains the same as it is now. And thanks for giving that a read. I know it's long (and still not even complete!).

[jlrdsr]

I dont mind a long read, and good ideas make it worth it. Id love to see some of these implimented, if i knew more about code id do it myself!

[vexx32]

That is all awesome, but yeah, you're probably scaring people off with the wall-of-text format. Some clearly defined headings and maybe breaking it up into more clearly-defined paragraphs would work wonders.

[DMagic]

I have been mulling over some of the exact issues you discussed. I even started making up science reports for low orbit, Kerbin biomes for my magnetometer, but I realized that I don't want so many reports to be possible, I just want two or three while the others give some default result and only return science value once (limiting the actual result to a default report is simple, but they still give science value for every different biome). I've been trying to come up with ways to limit science experiments by biome with a plugin, but that's not very simple.

I think Rhidian's Science Redefined mod is a good start for all of this, and I hope some of his ideas get picked up or adapted at some point.

[Atoning Unifex]

There are some good ideas in here, although I agree with Vexx32 that it could have been made a lot easier on the eyes with some decent formatting! I particularly like the point, local and visual concepts, and the idea for different tech tree setups allowing different playing 'modes'.

Regarding lab analysis, I don't think that currently it makes sense to have a manned lab - barring the pointless transmission bonus that it produces. I think that there should be several classes of science experiment of increasing complexity. There could be probe devices and lander devices that require a certain type of command pod to be able to use, so the simple probe cores could only perform one or two basic readings, whereas the more advanced cores could perform the basic ones and intermediate ones (or maybe have it limited by number of experiments a core can perform in total?). Next there could be a class of devices that require manned capsules and, finally, a manned lab for the most complex ones (bringing meaning to the currently pointless lab).

This would allow smaller craft to only perform targeted science. whereas a manned station could perform any analysis.

Do away with the transmission losses, have all the data, as suggested, transmit at 100%, but also have some experiments produce some kind of physical resource (like the surface samples) that has to be returned to give the science points.

[Rhidian]

I think Rhidian's Science Redefined mod is a good start for all of this, and I hope some of his ideas get picked up or adapted at some point.

I have recently released an updated version of my mod (found here), which addresses some of the points in the original post; specifically the Mobile lab applications, it's application to station science, probe usefulness via the ability to recollect data effectively over long periods of time, transmission efficiency (also removing diminishing returns), and some of the points raised about experiment type differentiation.

Science can and should be better, and I have been surprised by what I see as a lack of mods geared towards 'core science' (ie not just changing ScienceDefs or cfg files or the tech tree).

[jwenting]

Now how come I only have one response if the idea is so good? Am I scaring people off with the wall of information?

ever heard of sarcasm?

stopped reading after the first line of your indiscriminate rant. Was convinced after that that it was not worth spending more time reading the rehash of rubbish ranted about before ad nauseum, all of which comes down to "let's make KSP harder for newcomers so we old hands keep our bragging rights at being king of the hill".

[vexx32]

Sarcasm isn't always obvious to all in plain text, even when you take pains to ensure it is. People are weird like that, they overlook a lot of things.

And whether or not Sierra's arguments are so easily reducible to something so simplistic is up for debate, really , but you definitely could have been a little nicer about it. (Though I do agree with your earlier statement that such an immense wall of text is a bit much, especially without proper care taken to break it up into sections, at the very least.)

[Atoning Unifex]

ever heard of sarcasm?

stopped reading after the first line of your indiscriminate rant. Was convinced after that that it was not worth spending more time reading the rehash of rubbish ranted about before ad nauseum, all of which comes down to "let's make KSP harder for newcomers so we old hands keep our bragging rights at being king of the hill".

I almost did the same to be honest, however once past the first paragraph the ranting stopped and ideas sterted flowing, even if in a somewhat haphazard way!

I would also disagree that this boils down to your 'making things harder' statement. It's just suggestions to improve the science system, which a lot of people agree needs improving. Some of the ideas may make it harder for new players, but others could make it easier. I'd suggest that before you dismiss someone's suggestions so bluntly it would be common courtesy to actually read them first.

[Captain Sierra]

Well, I'm back after a short internet blackout. It's good to see most of the comments are positive. I'll go back and do some heading editing. I think, however, I'm within my rights to leave it as it was at the time of posting as it took me about an hour to write that and it was all coming off the top of my head from ideas I'd been mulling for a few days now.

[DMagic]

It's not about making the science system harder, or less user-friendly, it's about making the system deeper, or at the very least making the tools more flexible.

The current science system and the tools available for modders don't allow for much customization or added depth. There are ways around this, but it requires a lot of work. I think this is why we're only now starting to see some different takes on the science system, rather than just alternate ways to collect default science reports (like adding crew reports to probe cores).

While it can be helpful to discuss what should be done about it, I think the best option is to try out different ideas using mods to figure out what works and what doesn't. Maybe some good ideas will be discovered and picked up for a future update, or maybe not, but I think it's worth trying out some different methods for the science system.