About the Tech tree

[TranquilTempest]

But even flying a suborbital rocket early on should be hugely meaningful. Being too ambitious should be hard.

Does orbiting a Kerbal with nothing but LV-1 engines count as too ambitious? http://i.imgur.com/xew0NYL.jpg

[thorfinn]

What Harv's post doesn't explain us is the way through which money enters this process, and I think that's an important part. For example, it could be a way to explain the connection between science and parts advancements - Science translates into Funding that pays for Engineering, flight experience directly improves Engineering, parts are unlocked by engineering points.

Flight experience does not equal flight time - even when you launch a multi year probe mission, the first stage engines fire for just 2 minutes like always.

Also, I think that flight experience with a particular component (subassembly?) should reduce its unit cost - that would give an incentive to standardization.

Edited August 16, 2013 by thorfinn

[Guest]

I fear that basing it off flight time, for instance, can have that risk. If you're not playing for fun, but playing just so you can unlock the fun, that's inarguably a grind. Lots of games have that pitfall.

HarvesteR has rejected any notions of basing an R&D system off of flight time because you can simply time warp. I don't think you have anything to worry about.

E: And regarding the rest of your post, I don't think any parts are going to be changed in stats to "balance" them against the tech tree. You should be able to do pretty much anything with your starting parts (beyond maybe the most ambitious missions, Eeloo and Jool for instance, but my Moho rovers are built entirely with 1.25m parts.), R&D should just open up more build options. Thankfully that seems to be the route that Harv is proposing.

Edited August 16, 2013 by regex

[Dave Kerbin]

So, I've been thinking long and hard here about how we're going to do the 'earning science' bit. It's a good idea, but putting it in practice poses a few challenges.

How do we rate the science value of your missions? What even defines a mission in the first place? These things are actually pretty hazy in terms of how the game deals with them, so coming up with a nice, solid way of doing science is a deceptively tough task.

I do believe I've worked out a good solution now:

The game won't award you any science automatically. That would be artificial and generally meaningless, but worse still, it would require us to make arbitrary decisions about the science value of this or that action. That's a bad road to go down. Instead, we can let you 'do science' as part of your missions, and get your science points for yourself. Here's how:

We already have a few science sensor parts, which apart from a context menu readout, are largely decorative. We can put those to some use now, along with a couple other scientific parts we're going to add.

The idea is that science parts work as one-shot experiments. That is, they're activated by action group or as part of the staging sequence, and once deployed, they to their thing. This is essentially deploying the experiment to gather data. This data isn't science yet though, because you need to get it to your resident experts over at R&D to crunch the numbers and make some sense out of it.

To do that, you can recover the experiments (or whatever is left of them). That will convert the data you gathered into scientific knowledge, provided you don't already have it. This is done by us storing where each experiment was run and what it was, and using that 'source' as a key to a multiplier value, which starts at 100% and gets progressively lower the more data on the same subject you gather. The more you spam the same type of experiment in the same place, the less science you'll get for the data it generates.

Now, if we've learned anything this far, it's that recovery is by no means guaranteed. So here's where the antennas and comm dishes finally get a purpose. Once available, you can use comms equipment to transmit science data back down and gain science immediately. Of course, you can't expect to get as much knowledge for the same experiment data if you beam it back as if you had recovered it hands-on. How efficient the data-for-science rate is depends largely on the quality of the antenna being used, as does its power requirements.

The possibilities for what the science-gathering parts can be then are pretty huge. They could be anything from experiment canisters filled with some mystery goo, a camera array, a mass spectrometer, anything really. Functionally it all works the same way, and better yet, it works with the game rather than over it.

That's about it in a very tightly packed nutshell. I'm pretty happy with this idea now, it should fit any style of playing. You could focus on running as many different experiments as possible in Kerbin's lower atmosphere, or rush into discovering the different properties of Mystery Gooâ„¢ across the farthest reaches of the solar system, and hopefully any point in between as well.

Disclaimer: Mystery Gooâ„¢ is not really a planned feature (yet).

Cheers

I previously assumed that career mode would revolve around completing specific missions like orbiting a satellite or landing on the Mun with the option to repeat them in some way for smaller gains, but what you’ve described sounds much better. It seems easier to create a wide range of goals and much more balanced for allowing sandbox like pursuit of the players own goals.

If I understand what you’re proposing it is to have instruments that each ‘collect’ a specific science ‘resource’, with instruments some having a limited use count before they are worn out. The resources are not infinite but also get used up in the process. Cities XL has a resource system like this – you need things like water and oil and each map has an overlay showing which parts of the map have it (in KSP it would be hidden). When you place a water tower it takes a circular bite out of the overlay, so you can’t just place water towers side by side. From the technical side it’s just a 1 bit map of yes/no values, and placed structures mark a point and radius where the resource is already collected. Attempting to collect anything either on a grid section marked 0 in the map, or within the radius of a previously ‘mined’ point fails. Better instruments might take a larger bite. I assume this would work even in the planetary nature of Kerbal – each planet, moon and the sun has a map for each type of science resource they contain. Either the map or the instrument has altitude parameters – you need to be right on the surface to collect rocks, while others require a certain altitude range, like between 70km and 100km to gather from a magnetic anomaly map.

Once a resource is ‘collected’ you aren’t instantly turning it into science. Instead there are path(s) by which an entity can convert units of one resource into units of another resource, until eventually you reach the Kerbal Space Center where the R&D building has the conversions to ‘Science’. For example we could land on Duna and use our remote rock collection device to collect 1x ‘Duna Rock’ resource which weighs 1.0t. We can pack that on our ship and fly it back to KSC where we can convert 1x ‘Duna Rock’ into 100x ‘Science’ to spend on the tech tree.

Alternatively if we send Bob Kerman on the mission he can be commanded to ‘sort’ ‘Duna Rock’ resources (there are things Kerbals do better then a machine), converting 1x ‘Duna Rock’ into 1x ‘Sorted Duna Rock’. Both can be returned to the KSC to be converted into 100x science, but ‘Sorted Duna Rock’ only weighs 0.1t. Finally we have a third option – we can pack a ‘mass spectrometer’ part on our probe. With this tool we can convert 1x ‘Duna Rock’ or 1x ‘Sorted Duna Rock’ into 5x ‘Duna Rock Information’. This resource weighs 0.0t (nothing), which means we can transmit it through an antenna back to the KSC. However the conversion between ‘Duna Rock Information’ and ‘Science’ is only 1x for 1x instead of 1x for 100x, so we get a total of 5x ‘Science’ for breaking down the rock and transmitting the information back by antenna, vs 100x ‘Science’ for actually bringing it back for complete study.

Finally I assume we will have antennas with different efficiencies – the rate at which they can send weightless information resources back based on the size of the antenna (or really an in game value shown in the part description) and divided by the distance (longer distance means a slower data rate to ensure everything is received). So the little antenna we have right now might be perfect for sending back just about anything from Kerbin orbit, but on Duna the distance factor would dramatically increase the time to transmit just 1x resource back to KSC. For this you would want a larger Duna base station with a big transmitting antenna. Your little rover transmits the information resources to the base station (short distance so the antenna is still fast), and then the powerful base station antenna sends it back to KSC.

The antenna efficiency becomes really important if there are practical limits on just how slow you can send, preventing the exploit of just time warping. The realistic but computationally harder approach is that transmissions require line of sight to the receiver (parts and other small scale objects don’t block LoS, just planets), and each resource unit can only be transmitted in full. That means that if the efficiency has dropped to the point that it takes 5 minutes to send 1x resource then you must have a LoS window of at least 5 minutes to send it. If LoS is broken the transfer has to start over again. A probe on the dark side of the Mun would need an orbiting relay to send information back to KSC, and a probe on Duna must transmit fast enough to complete at least one transfer before Kerbin rotates out of view.

[sjwt]

This would suck the fun from the game. Nobody like to be forced to anything that is one of the things about the game you are only limited by you design. The tech tree will allow you to go any where but it is hard at first.

You do realise the Tech part is only for the Career mode, and you *will* be forced to do things for that, things that you don't want to do, have already mastered or even don't have the skill as a player yet to do. If just want to build the ships you want? Thats the sandbox mode.

[Eric S]

Now, if we've learned anything this far, it's that recovery is by no means guaranteed. So here's where the antennas and comm dishes finally get a purpose. Once available, you can use comms equipment to transmit science data back down and gain science immediately. Of course, you can't expect to get as much knowledge for the same experiment data if you beam it back as if you had recovered it hands-on. How efficient the data-for-science rate is depends largely on the quality of the antenna being used, as does its power requirements.

How will kerbaled vs probe missions factor into that? Not that it has to, but one of the reasons to do manned missions is that an astronaut can collect more scientific data in an afternoon than a probe can in a week or two. I agree that we don't want to have time spent playing a strong factor, but I think a manned mission should have a better scientific data return than a probe mission.

It could be done as a multiplying factor, or there could be probe science packages and astronaut science packages, with the astronaut packages being different experiments that return more science data. The later would eliminate the problem of a thermometer mounted on a probe giving less science than a thermometer mounted on a munar lander. Another alternative for some types of experiments is that the apparatus for the experiment could be lighter if it doesn't need any automation.

As for where the science happens, you could have modifiers for celestial bodies, or just have diminishing returns per experiment per celestial body (or even region within a celestial body). Wait, don't like regions, diminishing returns based on distance from the closest place you ran the same experiment? So two thermometers on the same munar probe won't tell you much more than one, but using a rover to take a measurement 1km away would be better, and someplace entirely different would be better still.

I also hope to see science packages that are unlocked through science. Say you've done all the basic experiments on the Mun to the point that diminishing returns is kicking in hard, you could then decide to research more science packages that can be done on the Mun (or anywhere else), or you could research propulsion, so you can do the basic experiments somewhere else.

I'd say that I don't want to see someone be able to continue gathering scientific data conducting the exact same experiments in basically the same place on the Mun, but to be honest, I'm so excited about this whole concept that I'll be flying missions non-stop for days even if this is exactly how science works in the first release. I would expect it to get better, however :-)

Ultimately, I'd like to see science being something that encourages but doesn't require doing as many different experiment types in as many different places as possible, and having an astronaut there to do the experiment as often as possible.

I do agree on the idea of being able to radio back the basic results and then get the difference upon an actual return. Especially important for grand tour missions.

[thorfinn]

we don't want to see player unlocking new engines by studying Mun rocks, yes?

Well, no, but Kambridge university could be so thrilled by your results on Mun rocks that they will pay for the development of the engine you need for going to pick up Duna rocks

[Skyrunner27]

You do realise the Tech part is only for the Career mode, and you *will* be forced to do things for that, things that you don't want to do, have already mastered or even don't have the skill as a player yet to do. If just want to build the ships you want? Thats the sandbox mode.

This is not (at least from the information I have heard I may be wrong) the direction Squad wants. They allow you to do anything but you are limited by your design. Gaining new tech only make it easier for you to do missions. EG you don't need landing gears to land on the moon but it makes it easier, or you don't need a LV-N to reach the other planets but it make it easier. There may be contract missions which you choose to take but being forced is not (again from what I heard) what they do. Think sandbox with a tech tree, economics, and contracts that you may choose.

[Saeblundr]

[snip] Disclaimer: Mystery Gooâ„¢ is not really a planned feature (yet). Cheers

oh hell yes. i religiously attach various science parts to all satellites, never know when i will need to know the temperature of the darkside of Minmus i say!

cant wait for more structure to the campaign mode, this is going to keep getting better and better! very excited ...

[KevinTMC]

...or rush into discovering the different properties of Mystery Gooâ„¢ across the farthest reaches of the solar system...

Finally, the great "Just what is in the 'NOT FOOD' bin?" question is solved.

It's Mystery Gooâ„¢!

[e-dog]

That should give all mod-makers a chance to revise their cfg values for things that were of no consequence until now (costs mainly), and assign their parts to the tech node that fits them best.

Please, allow PartModules to change part cost. Currently, the cost is in AvailablePart, so procedural parts can't set cost based on part size etc.

[e-dog]

Heh, and I considered making a comm+science mod, kinda "RemoteTech light", without time delays and using satellite network to transmit science data, not for probe control (so you can pretend the probe was programmed or on autopilot when you control it without comm link).

[Frederf]

Please, allow PartModules to change part cost. Currently, the cost is in AvailablePart, so procedural parts can't set cost based on part size etc.

That's an astute observation. Dynamic cost based on part tweakables: resource contents, procedural dimension, etc. is going to be a necessary function.

[aeronerd]

Yea. I think you have a winner here. "go to space..... send back 'stuff'"

[Petarski]

very good ideas here

Would the distance of the craft from Kerbin count for science earned?

ex. A rover on Duna might gain more science than one on the Mun.

[leax256]

So... How long will it take to research Struts?

somewhere after You begin to have a need for them but it's not anoying to not have them.

[leax256]

That's something else to wonder about and contemplate: reliability. Will career mode include realistic reliability stats for newly-unlocked parts?

They have said before, no random disastors

[alacrity]

I am thinking that not only should R&D should increase parts available but also affect price of parts.

Ignoring inflation, part pricing could be adjusted for the usage of those parts... Mass production, material science, and part refinement could effect the prices. When you first create a part it is massively expensive, but with use and repeated builds the price would fall off.

Just a thought from the Miser Space Agency

Alacrity

[Maelztrom]

Can we have a way to manually transfer data between ships?

Let's say i have a permanent satellite gathering data, can I have a lander hand-deliver it to Kerbin?

Or, a better example would be if i make a lander / reentry vehicle can I send it to Duna, get it back to orbit, then give it's science to an interplanetary shuttle to Duna?

[triffid_hunter]

I know this departs at at least 80 degrees from the current concept of R&D, but I still have thoughts leaking all over the floor.

My vision of R&D is an evolution of parts based on player feedback. It relies heavily on being able to produce parts semi-procedurally with fine grain adjustments of attributes. At simplest example take three standard "seed" parts, engine, fuel tank, parachute. The player elects to direct the R&D department to develop new parts of these categories with directed changes. R&D "dice rolls" some new parts with more desirable stats in the preferred area with possibly some degradations in others. Given a handful of possibilities (1-10) the player accepts or rejects these potential parts into the stable of usable parts. "We need an engine with more thrust" or "this parachute needs to open higher but with less drag" or "this fuel tank needs to be shorter and stronger at the cost of capacity." Thus the player directs their own part development qualitatively and ends up with a selection of parts uniquely their own.

Gameplay restrictions could include R&D cost per role, cost to support a larger stable of actively available parts or hard limits on number.

A tech tree and R&D evolution would synergize well. One might give up an old model honed to perfection in favor of a new technology that's still raw. New techs could open up new attributes to adjust or extensions on the values allowable such that existing models could be brought up to contemporary tech.

THIS!

Sometimes I'd be quite happy to sacrifice some thrust for a bit extra ISP, or the other way.. or perhaps I want a fuel tank with a size that's right in between two stock tanks.

Procedural wings and fairings seem to be quite popular at the moment, making _everything_ procedural as part of the tech tree implementation would make me a very happy clam

What will you do with plugins in career mode? Using MechJeb in career mode is definitely cheating.

I disagree most emphatically. Current day space missions are flown with VASTLY more instrumentation and automation than stock KSP currently provides, I'm sure they have something a fair bit more advanced than MechJeb tucked away in mission control and duplicated on the craft

I'd love to see MechJeb integrated into the tech tree in a sensible fashion.. perhaps at the start you only get the instrumentation (delta-V in the VAB, vessel, orbit and surface info during flight) and as you fly more and more missions you slowly open up the various autopilots- Smart A.S.S. first of course, then perhaps maneuver planner, ascent, rendezvous/docking, landing.

For best results, I'd love to see these opened up by player demonstrating ability to perform these maneuvers manually- perhaps something like

* "12 successful launches, ascent autopilot level 1 unlocked!"

* "40 staging events on successfully returned craft, auto-stage unlocked!"

* "20 degree plane change completed after gaining electronics level 4, Smart A.S.S. level 3 (normal/antinormal) unlocked!"

Alternatively, the text could read "MechJeb has learnt ..." (ala Black & White) instead of "... unlocked"

Hopefully the devs will create an API which allows MechJeb's author to implement exactly this!

How do we rate the science value of your missions? What even defines a mission in the first place? These things are actually pretty hazy in terms of how the game deals with them, so coming up with a nice, solid way of doing science is a deceptively tough task.

I do believe I've worked out a good solution now:

The game won't award you any science automatically. That would be artificial and generally meaningless, but worse still, it would require us to make arbitrary decisions about the science value of this or that action. That's a bad road to go down. Instead, we can let you 'do science' as part of your missions, and get your science points for yourself. Here's how:

We already have a few science sensor parts, which apart from a context menu readout, are largely decorative. We can put those to some use now, along with a couple other scientific parts we're going to add.

The idea is that science parts work as one-shot experiments. That is, they're activated by action group or as part of the staging sequence, and once deployed, they to their thing. This is essentially deploying the experiment to gather data. This data isn't science yet though, because you need to get it to your resident experts over at R&D to crunch the numbers and make some sense out of it.

I like where you're taking this

I think a strong driving force during early career should be commissions; ie "take sensor X to position Y, return within 7 days and gain a 350% bonus in 'science'."

As career progresses, I think these commissions should become less important, allowing the player to perform their own science. I'd love to be able to find anomalies in space, eg take a gravity sensor to L3/L4/L5 for an unannounced bonus, or drop Magic Goo on the moon arch- essentially silent commissions.

Later in one's career I'd love to see some really massive commissions like a Moons of Jool grand tour- "land Billionaire Richie Kerman on 2 of Jool's moons and complete a full orbit of at least one other then land him back at KSC within 5 years to receive 60 million Kubles and 5000 science points."

Lastly, I'd be interested what the devs think of an exchange rate between science points and funds which varies based on events such as disasters (slightly poorer rate), deaths (vastly poorer rate) and surprise commissions (rate increase).

[EchoFiveOne]

THIS!

I disagree most emphatically. Current day space missions are flown with VASTLY more instrumentation and automation than stock KSP currently provides, I'm sure they have something a fair bit more advanced than MechJeb tucked away in mission control and duplicated on the craft

I'd love to see MechJeb integrated into the tech tree in a sensible fashion.. perhaps at the start you only get the instrumentation (delta-V in the VAB, vessel, orbit and surface info during flight) and as you fly more and more missions you slowly open up the various autopilots- Smart A.S.S. first of course, then perhaps maneuver planner, ascent, rendezvous/docking, landing.

For best results, I'd love to see these opened up by player demonstrating ability to perform these maneuvers manually- perhaps something like

* "12 successful launches, ascent autopilot level 1 unlocked!"

* "40 staging events on successfully returned craft, auto-stage unlocked!"

* "20 degree plane change completed after gaining electronics level 4, Smart A.S.S. level 3 (normal/antinormal) unlocked!"

Alternatively, the text could read "MechJeb has learnt ..." (ala Black & White) instead of "... unlocked"

Hopefully the devs will create an API which allows MechJeb's author to implement exactly this!

I really like this idea. Personally I use mechjeb to automate tasks that I've already accomplished but are tedious to repeat, e.g. conducting multiple launches to loft space station parts into orbit. As you say something like having the ability to point prograde or retrograde automatically would be a low level unlock, then later on perhaps 'surf' mode, then perhaps the final unlock would be the ability to automatically calculate and execute a Hohmann transfer or similar.

[lazarus1024]

Harv I am sure will clear it up at some point, but reading his latest:

It sounds like there will be no time component. You launch a rocket around Kerbin with a grav sensor on it. When you reach orbit, you turn the grav sensor on and you have officially completed "Gravity research in Kerbin orbit". You re-enter your ship, land it and then recover it and you get 10 Gravity research points at that point. No "time" required other than what it takes to get to orbit, hit activate and re-enter. Or you launch the ship with the grav sensor in to Kerbin orbit with a radio antenna, activate the grav sensor and then activate the antenna, you immediate get 5 Gravity research points for having conducted gravity research and beamed the information back.

Next time you do it, maybe you only get 7 gravity research points since you already conducted the experiment, and so on.

Or you fly to the Mun with a soil sampler, return and get the 10 soil sample points from collecting on the Mun. Then go to Minimums and do the same, get 10 points. Go back to Minimus again and only get 7 points this time since you were already there. Rinse and repeat.

That is my take on what Harv was saying. I think it makes sense, I think it is great that is the direction it seems to be going.

A couple of questions/suggestions. Since recovery is going to be worth less than com reportings experiments, I'd suggest that for a mission you have the ability to both "com back" experiment data to earn research points AND be able to return the mission and recover it for the rest of the points you would have been able to get if you had just launched the mission normally and recovered it. This provides a bit of a safety net for missions. So if I launch my Apollo mission to the Mun, land with some instruments I can beam back my experiment data to get half credit and then return the mission to get the rest of the credit, but if I lose the mission in the process of returning it, I still got half credit for beaming the information back.

One other "lastly" not, I don't think money is going to make it in to this one. Harv made it sound like that wasn't super likely for this release, though it sounds like they are going to work some of the mechanics in to the back end so that KSP is a bit more ready for money in career mode. Earning money and using it sounds like that is another big dynamic which might be better off waiting for .23 instead of .22. Though maybe this is in part where resource collection is going to come in. Either in part to earn money, or maybe at least make missions less expensive by producing resources that would allow you to better re-use parts (IE you can refuel your interplanetary tug in Mun orbit from your mining/refining base and fuel station rather than returning to Kerbin, and then having to spend money on building a rocket to ferry fuel up to your tug).

[HarvesteR]

It sounds like there will be no time component. [snip]

One other "lastly" not, I don't think money is going to make it in to this one. [snip]

If you mean a time component referring to requiring some time to elapse to research a new tech node, then no, that's something you can work around by simply having a ship at the pad to wait it out. These time constraints usually only work for real-time games, where time management is big.

Now, as for the experiments themselves, some experiments could require a running time. Like you said, some gravity measurement in Kerbin orbit could be picking up data over the course of several orbits. But then we get into the "you can jsut time warp" issue, so I think such mechanics have to be worked in carefully.

The cool thing about this is that it doesn't really matter much as far as the science is concerned. This becomes a front-end feature, something that the experiment module itself worries about. It's up to how each experiment works then.

As for the money, you are correct. A budget implementation is something we're only planning for later, so it's not going to be on this update. But in keeping with our ways of always keeping the game at a playable state, the R&D system is able to function without it, as a standalone mechanic. Hence the whole science thing.

Money does come into R&D eventually, but it's much too soon to be talking about that right now.

Cheers

[rdb78]

I would love a landing radar that told you the incline of the surface below the spacecraft. That's more of a piloting thing, but it sure would be handy. Maybe it could also allow topographic map construction if it's in orbit or something.

[lazarus1024]

Cool. Thanks for the information!