A more intuitive tech tree

[DuoDex]

Why don't you describe it for those who haven't played it?

It's web style, but techs/parts aren't necessarily "better" as you explore it. They just become more and more specialized. So an example of this would be the LV-909 being on the same "group" as the LVN, but the LVN would be farther away from the center and not on the same strand as the Poodle, for example. If people show enough interest I might make an example.

[CaptainKipard]

It's web style, but techs/parts aren't necessarily "better" as you explore it. They just become more and more specialized. So an example of this would be the LV-909 being on the same "group" as the LVN, but the LVN would be farther away from the center and not on the same strand as the Poodle, for example. If people show enough interest I might make an example.

I think that's understandable but do it anyway. Grab Dia and draw it. We're all throwing things out there, and people respond better to visual communication.

I like the Engineering based tech tree he did a really good job on it.

I think Kerbals should start with at least a basic plane or something like that then move onto rockets and probes and a whole separate portion for life support needs to be added.

There needs to be a branch for life support, space stations, colonization, rovers, and drones.

The problem with grouping according to craft design is that it's not easy to do. Parts which go well in stations go equally well in other types of craft. I'd say keeping them grouped based on function is easier and more intuitive.

Edited November 17, 2014 by Cpt. Kipard

[Nertea]

My favorite tech tree tree I've ever used was the Blind Research option in Alpha Centauri. It's broadly similar to the categorical research suggestion here. There was a tech "tree" of stupid complexity (here), but you put your research points into one of four general areas - Build (expansion and base stuff), Conquer (guns and armour), Explore (mobility and ecology), Discover(pure science). You could focus on one, two, three or distribute points equally to all four. Technologies belonged to each of the groups and groups were scattered everywhere. When you got enough research points, you'd get a tech that would be roughly based on your available links in the tree and your selection of areas - full focus on Conquer would usually get a Conquer tech, or a prereq to a next Conquer tech.

Techs were relatively fluid in terms of categorization, sometimes a Discover tech might give you a weapon, or be required for a military discovery later on. It was very flexible, there were never dead ends, and most full games I played didn't have exactly the same route.

I would love to see something like this for KSP! It would combine a structured tree-based approach with flexible path, and add the "no career is quite the same" thing that I think is lacking right now.

[CaptainKipard]

Would you like to sketch out a rough outline in Dia? The image you provided makes my head hurt.

[ximrm]

I would be for two starting nodes:

One for early space missions and another for aircrafts.

Those could intersect a various points and make the progression a lot more dynamic.

It's just a personal preference.

Edit: I would really like to be able to upgrade a node got get better versions of the parts that can be found withing that node.(Engine has slightly better ISP, thrust or size. Struts slightly stronger etc.)

Edited November 24, 2014 by ximrm

[Rakaydos]

What about three starting "Tier 0" nodes, that cost significantly more than the tier 1 nodes? So you start with enough science to start with -EITHER-

1)basic Wings, Landing gear, basic jets, lateral intakes, jet fuel tank and the Mk 1 cockpit

2)Probe core, thermmoter, baromiter, antenna, probe tankage and the roccomax 48

3)Mk1 capsule, RT10, Lv30, parachute and small tankage.

The tier 1 would then overlap significntly, letting probes and spaceplanes access the LV45, for instance.

[Kegereneku]

I think we need a more fundamental change Rakaydos.

Your proposal would give some much needed choice, but it would still be using rigid bundle containing the technology you want for another "route".

Your probe core route for example don't give much propulsion, so since (3) is costly you would skip to the LV45 and whatever has tank and decoupler cheaper

If you wanted more science part you will grind for (2) probes cores because it's more cost efficient.

And the jet route, although it would give you the ability to get science all over Kerbin is fairly limited to atmosphere only (no ladder) and it take longer to fly over a specific zone to make report than it does to send a capsule sub-orbital.

ximrm,

The main problem with upgradeable parts is that you'd need something to tell them apart on screenshot to make sharing and helping easier.

[Laie]

My favorite tech tree tree I've ever used was the Blind Research option in Alpha Centauri.

Would you like to sketch out a rough outline in Dia? The image you provided makes my head hurt.

Now way -- and that's the point. The Alpha Centauri Tech "Tree" actually is a tangle, a thicket. There's a lot of interdepending connections crisscrossing all over the place. Which makes a lot of sense in a civilization-like game. I'm not sure if the concept would carry over well into KSP.

By the way, I can heartily recommend the game; in my opinion, it's the best civilization that ever was.

[Badsector]

I want see tech-tree unlocked by missions and parts unlocked by science points with time and queque for research them, that can make more challenging to the game.

[Wanderfound]

A key point here: why are nodes bundled at all? Why not have each tech advance unlock a single part?

[Badsector]

If you watch a tech tree like technology stage, at every unlock you have parts bundled on this stage and after you have to spent time and resources for research them, you can start with small probes and after some months-years sent the first kerbal to the mun, and that need a totally rebalance relative to gameplay on techtree and contracts rules

Edited November 24, 2014 by Badsector

[CaptainKipard]

A key point here: why are nodes bundled at all? Why not have each tech advance unlock a single part?

I like that idea. Maybe something like this:

Each node which starts a new technology branch could have a higher cost associated with it (simulating RnD), with each subsequent variant being cheaper (simulating minor improvements). The exact numbers are a matter of gameplay balance.

You'd still need to bundle some of the most basic parts in a root node of all branches, otherwise they'd have to be empty.

Edited November 24, 2014 by Cpt. Kipard

[John FX]

Whatever the final tree thing ends up being (maybe not even a tree, maybe a node web) the only thing I care about is that there is replayability in it which means I`m not forced to make the same choices over and over due to restrictions in the tree structure and I can adjust the structure myself.

To clarify, on one save I might want to follow the earth tech path closely. I`d want to have wings and landing gear first with very small rockets (almost toys) and jet engines followed by unmanned craft to LKO then manned from then on.

On another save I may want rockets first but unmanned so I`d need a pod, chutes and some tanks and engines etc

On another save I may want to follow the current tech tree.

And so on.

The best way forward IMHO whatever the final structure will be is to allow the player to adjust, load and save their own trees edited in an easy way in game. To place parts wherever they like, make nodes, delete nodes, set science and funds needed, make and break node connections.

Then everyone gets the tree they want and everyone wins. All the suggestions above could be made within that framework. The `exploding star` tree, the `micro purchases unlock individual parts` tree, the current tree and any tree you can imagine could be constructed.

No hardcoded anything.

this is my 1/50 of a currency unit.

EDIT :

Yes. Easy tree modability is a must, but what would that backend look like? However hard I try I can't think of a simple way for stock and mod parts to be automatically placed in appropriate places for all potential custom trees, without a ridiculously long MM configs.

One partial solution could be adding the ability for mods to append nodes to existing trees.

Say for example you play with the tree in my diagram. It would be the best tree for this because it's intuitive and tech oriented. A mod like Interstellar for example could use some of the "stock" nodes, and dynamically add new nodes for other technologies. The player would be able to easily install the mod and uninstall it, without having to use a custom tree.

The nodes then would have to be moved around dynamically in the tree view.

I get a feeling like the only reason this might be true is because the current tree is such an unintuitive mess.

I think the opposite. I think an intuitive tech-oriented tree would be much easier to navigate. Players would be able to predict to a greater degree which direction to research for their needs, compared to the current tree.

Mods are something that would need some thought in an unfixed tree. Where to place? What tech should already be unlocked? What science and funds cost? etc

I like the idea of mods dynamically adding nodes but a good framework would be needed for that. Maybe mods could be placed roughly by a sort of faux tech level? For example, Mechjeb info displays could be `tech 1` level and so would be placed in the appropriate place automatically by the tree manager in KSP (mod, control, tech 1) and the docking autopilot might be tech 4 so would be placed into a `mod, control, tech 4` node?

i would imagine mods could have a sub tree (like a subassembly) if they have multiple nodes so only the first node would need placing and all the rest would have a relative placement.

I can see validity in the intuitive tech tree for new users, as you say the current one is a bit of a mess. New users should all have the same tree though so everyone has the same `out of the box` experience.

Edited November 24, 2014 by John FX

[sherkaner]

I like that idea. Maybe something like this:

Each node which starts a new technology branch could have a higher cost associated with it (simulating RnD), with each subsequent variant being cheaper (simulating minor improvements). The exact numbers are a matter of gameplay balance.

You'd still need to bundle some of the most basic parts in a root node of all branches, otherwise they'd have to be empty.

http://i.imgur.com/yHqqR0n.png

I'm a newbie jumping into this thread, but in concept I like this idea. It seems really odd to me that there are these artificial groupings of parts, which inevitably leads you to researching a potentially expensive node for a simple utility part contained with in it. Plus I think there is more satisfaction of feeling like you are really running an R&D program when you have that sense of learning from the parts you have, enabling building of similar but bigger/more complex versions.

And there could still be cross-linkings that enable or reduce the cost of similar parts in other types. For example, if you push far into mkII spaceplane parts, including jet fuel fuselage parts, that should at least make it easier for you to get similarly-sized rocket fuel tank parts.

That also opens options for allowing players to make long-vs-short-term decisions. For example, one might spend a lot of time/resources going directly for a bigger rocket, after which building a smaller one would be trivial/cheap. However, it would be much more expensive to go that way than going through a more stepwise progression -- like a crash Saturn V program vs. a progression of R-7 rocket derivatives, or shooting straight for a Space Shuttle rather than building an X-15 and then a Dyna-soar. Although perhaps there might be some very large-scale tech "eras" -- maybe just 3 or so -- where you can't see to the next level until you've completed enough of the one you're on; you don't even know to try to build a scramjet until you've worked out a good bit of jet power, aerodynamics, and materials.

On the surface it seems like having a per-part "tree" would lead to an immensely complicated UI, but I agree with those here that suggest that it doesn't have to be so. In fact I think it could be much more intuitive for the player to just have parts -- in combination with good basic contract/missions that give a player good starting points to learn the function of any new class of part they're looking at.

An by the way, I'd generally like to give this entire thread a big +1 -- great discussion here, and I very much agree with the overall goals and ideas being discussed.

[sherkaner]

By the way, for those who didn't see my thread over here (which perhaps I should have posted here in the first place), I'll post this little aerospace timeline I made that might be an interesting reference resource in this discussion. I definitely DO NOT believe we should be trying to force the player to duplicate this progression in any sense, but I would tend to evaluate tech tree ideas on whether a player would have the ability to follow this kind of progression through the tech tree if they so desired. But if somebody wants to run a program that goes straight for a Mars colony at the exclusion of all else (with all the challenges that would entail), I'm all for it!

[CaptainKipard]

I'm a newbie jumping into this thread ...

A newbies perspective is equally valid to that of someone who's played well over a thousand hours like me. It's possible we may have overlooked something or simply forgot what it's like starting out. This time though it looks like we agree.

[CaptainKipard]

Summary

The points may conflict. I tried adding all of the most discussed or supported ideas.

1. Rework the tree to give players an open experience. Rough outline below. Let players follow their own progression.
   
2. Branches divided by technology.
   
3. Try to put a single part per node, or only strongly related parts. Keep it intuitive.
   
4. Have successive nodes with larger/more advanced parts. Example below. more
   
5. Allow modders to dynamically append nodes. more
   
6. Create new part variable for determining part type/function. Useful for filtering. more
   
7. Little or no node interdependency.
   
8. Make really low-tech parts like ladders or thermometers available early. more
   

Example of a single technology branch

Rough tree outline. Nodes need breaking up.

Edited November 29, 2014 by Cpt. Kipard

[sherkaner]

I think that looks fantastic; so much more intuitive and really seems to eliminate problems where you're stuck buying into a lot of R&D when all you need is a little on a particular thing. The devil may be in the details once you really start dropping in individual parts, but it sure feels like a massively better way of doing it.

Layering on top of this, it still feels like there might need to be some concept of technology eras (perhaps linked to KSC states or something) that would prevent a player from *seriously* minmaxxing through one branch, but maybe it wouldn't really end up being a problem in practice.

Even with the initial small amount of science provided, I think you'd also have to carefully select a good basic set of starting parts to automatically give the player, along with a set of basic initial contracts (which can still be addressed in different ways ideally). I just imagine some newbie situations where they make a couple of bad choices (blow all their science on tanks or something) and are unable to make any progress. Although I liked the idea somebody else had here to maybe have a small selection of initial states with starting parts and contracts: Maybe the Air Force option where you've got wings and such, but only the most basic rockets -- or the OKB-1 option where you've got some nice liquid boosters and tanks, but that's about it -- or the NACA/ORDCIT option where you've got nice guidance and instrumentation and some middling rocketry stuff.

[CaptainKipard]

Even with the initial small amount of science provided, I think you'd also have to carefully select a good basic set of starting parts to automatically give the player .... I just imagine some newbie situations where they make a couple of bad choices (blow all their science on tanks or something) and are unable to make any progress.

Like maybe a thermometer from the start? So you have a way of fixing your mistake. Realistically tho that's a danger with any game. I doubt that anyone would be silly enough to not research some kind of mibility at the start. I just don't think that a tank and an engine needs to be bundled in the start node, because some people might want to drive around the KSC at the start, instead of flying, or they might want to create some science planes instead of rockets.

[sherkaner]

Like maybe a thermometer from the start? So you have a way of fixing your mistake. Realistically tho that's a danger with any game. I doubt that anyone would be silly enough to not research some kind of mibility at the start. I just don't think that a tank and an engine needs to be bundled in the start node, because some people might want to drive around the KSC at the start, instead of flying, or they might want to create some science planes instead of rockets.

Yeah, I was just thinking like an RT-10, nosecone, very basic probe core, and a thermometer maybe. Then no matter what other choices you make, you can at least play with some sounding rockets -- I mean, you're not going to be playing KSP very long without at least some rocketry anyway. But sure, you might be right that that might not even be necessary. Just a small point -- overall, this system seems certain to be a good thing.

[Kegereneku]

I just imagine some newbie situations where they make a couple of bad choices (blow all their science on tanks or something) and are unable to make any progress.

That's a very good concern.

As you say we could solve this with one one of the earlier proposal :

- ask the (supposedly new) player to select a (manned/unmanned/plane) bundle before he encounter the tech-tree for the first time.

- script the tech-tree to force the choice of "minimal propulsion/control/tank/science" choice. (it's not my preferred option as it could make confusing how free is the tech-tree)

- or let them reboot the scenario, they are still at the start anyway

About your thoughts on "technological Era", I'm not really against but it might not be needed depending of the Gain/Cost balance in science-point.

- Of part cost increase non-linearly for example. A skilled player might try to rush "his best part" but he wouldn't gain as much science-point from each outer planet, encouraging him to spend money on other techs.

- There's also cross-branch Prerequired technology

- I also see contract diversity ensuring a player would prefer keep some diversity.

After all, you can't take "satellite" or rover contract if you tried to avoid those part do you ?

[sherkaner]

A separate but related topic: Has there been discussion of other ways of "funding" the tech tree other than general "science"? The science system isn't awful by any means, but it does feel a little... arbitrary. I'm curious if there are any other ideas for something that feels a little more like R&D, where you wouldn't necessarily have to be gathering atmospheric data in order to get bigger rocket motors.

[Superfluous J]

A separate but related topic: Has there been discussion of other ways of "funding" the tech tree other than general "science"? The science system isn't awful by any means, but it does feel a little... arbitrary. I'm curious if there are any other ideas for something that feels a little more like R&D, where you wouldn't necessarily have to be gathering atmospheric data in order to get bigger rocket motors.

This is getting outside the thrust of this topic, but I had an idea where you can use all the parts in nodes you've unlocked, and a single part from a single node you've not unlocked. Some handwavey suff happens and then you earn the ability to unlock that node. I never got past that 2nd part, but the third step is definitely profit.

One idea is to completely decouple science from the tech tree and you simply cannot use new parts until you've completed a contract with them. Also, it would completely eradicate the tech tree now that I think about it. Just an internal tree to decide what parts are eligible for part contracts. I'm not sure if this is something that I'd like in stock but it'd be a really fun mod to try.

Edited November 30, 2014 by 5thHorseman

[Kegereneku]

There was various discussion around that.

As said by 5thHorseman research could be made entirely contract based

But I also read suggestion to make it :

- based on testing & recovering unique-prototype to allow to buy more

- based on how much money you pour into it (if you had regular income)

Those are the ones I remember.

Myself I would be okay with the current gather/spend points if parts weren't bundled and cost didn't followed a cubic curve.

[sherkaner]

There was various discussion around that.

As said by 5thHorseman research could be made entirely contract based

But I also read suggestion to make it :

- based on testing & recovering unique-prototype to allow to buy more

- based on how much money you pour into it (if you had regular income)

Those are the ones I remember.

Myself I would be okay with the current gather/spend points if parts weren't bundled and cost didn't followed a cubic curve.

Yeah, I can think of those and a lot of other ways of doing it. A hybrid system occurs to me, where you unlock the part with science/money/time, but then the part is initially very unreliable and the next level of that part is unavailable to unlock until you've used it enough to make it reliable.

Anyway, don't want to distract from the main tech tree topic at hand, but since we're talking about this per-component "tree" idea, it seems like it might suggest interplay with the way in which you unlock those components.