FreeOrion

O.k. I stand corrected, there are good arguments on both sides. My main worries are the gameplay issue of how exactly we balance a tech tree that we are randomly deleting parts of, and the practical issue of how we randomly delete stuff from a tree with relatively complex interdependencies (i.e. prerequisite techs) without leaving it hopelessly effed up. If we can solve both of those without too much effort then I'd probably be much happier about the idea.

Sorry, could you explain this a bit further, I get the feeling I'm not quite understanding what you mean. (Probably a problem on my part to do with me getting confused about prerequisite and allowable).

Oh, and one thing w.r.t. recursive tech trees...

One thing that could work quite nicely is the idea of selective refinement.

Say you've researched lasers. You could then choose to spend more and more research points on trying to improve their performance. You could iteratively refine size, damage, range, accuracy - by, say, 10% each time.

An exponential cost scale could apply. So, say, to initially research lasers costs 1000 RP. To add one level of refinement costs another 1000. The next level of refinement costs 1500, the next 2000 and so on. Or just have it the same cost every time... whatever.

Have each level of refinement produce a random effect (say, 0% to 20%) - so the player never knows what they will get - they may get a big improvement in laser range for one refinement, and an insignificant reduction in size for the next. Maybe let the player choose what type of refinement they want, e.g. if they want to improve accuracy then they can - but in this case give a significant (25%?) chance of a zero improvement.

OR : just have every level of refinement give the same (say 10%) improvement - but make it harder and harder to achieve (increasing costs/time, or increased chance to fail).

Ah... well... having variable techs within a tree structure becomes a bit more of a problem. The interdependencies tend to be a bit of a killer. Moo3 sidestepped this nicely by having the theoretical tech levels as the prerequisites rather than the applied techs themselves. Thats not really a proper interdependent tech tree (in the Civ/SMAC style) though. Its more just a set of linear advances (the theoretical research fields) with applied techs hanging off of them.

You COULD, however, have the theoretical fields organised in a proper Civ/SMAC tech tree, with proper interdependencies, and then hang various applied techs off of each theoretical tech advance. All theoretical advancements would be visible to every player, but the visibility of actual applied techs would change from game to game.

I think he means giving a tech (say) four prerequisites, but only requiring 2 of them to be met before the tech can be researched. This would help, but it would still be possible (especially if youre applying interdependencies at random) to get stuck in a dead end.

And assigning interdependencies at random will just make things silly. "I'm sorry, i need to have researched cloning and spaceports in order to build stellar converters?". I'm not a massive slave to realism, but you need to maintain some suspension of disbelief.

Sorry for the spamming

So what we see in HOI is:

Possibility of quite different playing strategies focussing on different unit types.

This is possibly because of terrain differentiation, i.e. islands (britain, leading to bombers/navy) continental (germany, leading to armies) and isolated (US, leading to navy/nuclear).

A possible problem in Moo is that all races MUST follow essentially the same research strategy as it is not possible to heavily emphasise one area other another.


So perhaps it would be a good idea to try and create similar options in Moo. And I have the germ of an idea for this.

Jump Gates versus hyperspace engines.

We have two fundamentally different engine/ship types, one (jump gates) only allows you to move between immediately adjacent stars thus producing 'front lines' in the combat, but allows you to move ships large enough to carry significant numbers of troops, so you can do ground invasions this way and conquer your opponent. (i.e. these ships are the gameplay equivalent of 'army' in HOI).

These would obviously lead you to also develop techs for conquest and industry (to keep churning out those heavy ships and troops).

The other (hyperspace engines) allows you to move directly to any star within a reasonable radius of the origin star, but only with relatively small/light (or maybe just unmanned) ships. So you can't carry troops, so no ground invasions. So these allow you to project force and keep your enemy down, by raiding ship yards and trade and bombing planets, but not to knock him out. (i.e. the equivalent of airforce in HOI).

This path would lead you to emphasise research to keep those light/automated ships competative with the big lumbering jump ships, and spying to discover suitable raiding targets and add to the grinding down effect of your bombing.

This way a player could choice between distinct gameplay styles, and so actually follow different research paths.

Extending this I think it would be a good idea if most of the main categories of research had two or more parallel 'threads' running through them which are equally viable alternative strategies, with some tech firmly in one thread or the other, but with many overlapping. Each time you played the game you'd then be able to choose a different set of threads to follow, and try and build a gameplay style that fitted them.

The idea obviously needs fleshing out, but what do people think?

p.s. @Davebaby, As you've probably noticed I gave up worrying about spamming this board ages ago... Seriously, I know double posts are normally against etiquete on boards, but I think the idea here is to get the ideas down and mulled over, so I treat having to wait until someone else posts before putting down the next idea or comment as a bit pointless. Plus it'd just mean I was constantly editing one post until it filled five screens by itself which is probably as bad as double posting. (this post should probably be three as it is...)

I have fairly strong opinions on some of what's come up, so pardon me if this comes across too harshly.

I consider removing at random any number of techs at the beginning of the game in the name of variety to be excusing us from out duties as game designers. The -only- way to balance a tech tree properly is to go over every single part of it very meticulously. Removing techs from the tree, causing them not to be researched at random (and only retrieved via spies, etc.) or randomly generating pre-requisites or interdependencies is putting into the hands of random chance the job we are supposed to be doing.

Emrys has a good point about terrain differentiation; I have two rebuttal points for that:

1) Although we will of course have the classic FO setup of one planet, one colony ship at game start, we also plan on creating 'scenarios' -- creating a universe that is as 'full' as, say, EU2 or HoI. So there will be several different angles of attack, and pre-defined empirs with different strengths in different tactical areas. This doesn't answer your question of 'all terrain is space,' though, so:

2) The German predilection for tanks or the British preference for naval power could be transferred to similar technology preference here; one race could prefer corvettes and fighters en masse while another prefers a small number of larger ships; the same sort of comparison you'd see between the Spanish Armada and the British Navy a few hundred years back. More importantly, though, at this point we can only figure out whether or not we'd like to even try for this sort of disparity between races, and build the tech tree appropriatley; we are not actually going to do many of the techs until we work on the relevant systems, so all we're talking about is structure and means to balance the tree. My point is that the tree is best balanced by making a game that does not encourage the same path through it, rather than artifically altering the tree to force you to take some arbitrary path through it; that, in my mind, gives you even less 'choice' than if there was a single best path through it.

As for Jump Gates vs. Hyeprspace, that's really well beyond the scope of what we're doing, and is also contrary to some things that have already been passed. However, it is an intruiging idea; that's the sort of thing that I could see coming out as a mod that could be tested and possibly incorporated after v1.0, but right now I'm trying to force everything to be as straightforward and streamlined as possible before we start introducing complex, game-defining structures like that (to say nothing of the fact that it would require double the amount of AI work). Remember that our 'Keep it simple' rule is not just because it's good design practice, but because we have to teach the computer how to play by these rules as well.

As for the recursive techs, that seems too artifical to me as well and provides no benefit that I can see. I would rather balance and design everything by hand than enact some over-arching rule that informs the whole system based on an arbitrary formula.

_________________
Surprise and Terror! I am greeted by the smooth and hostile face of our old enemy, the Hootmans! No... the Huge-glands, no, I remember, the Hunams!

I think it's important here to comment that a split between small/large ships in moo does not achieve the same type of thing as that between land/naval power. The small/large ships still perform the same task, they move in the same way and fight on the same terms. The land / navy units move differently, perform different tasks and can't fight each other on equivalent terms, thus the choice between them is a real strategic choice. The strategic element of the small/large ship choice is only really one between concentrated versus dispersed firepower.

That's in MOO. This is FO. We can design our combat engine to look more like a Total War series engine if we want. MOO2 and MOO3 were unable to make the kind of distinction you're talking about, but I don't even know how hard they tried.

_________________
Surprise and Terror! I am greeted by the smooth and hostile face of our old enemy, the Hootmans! No... the Huge-glands, no, I remember, the Hunams!

Two points about this, which are slightly self-contradictory.

I agree that we can't really do the techs till we have a fairly coherent game to alter with them, so all we can really discuss is the general shape of the tech tree, and in fact anything much beyond "There will be a tech tree, which allows you to research stuff in some kind of progression." is going to be a slightly pointless discussion.

On the other hand, the second part of the paragraph is heading dangerously into "There will be no problems in this game's design because we'll balance it so well...". Which is pretty much the line the Moo3 developers took when people said their design was overcomplex and unlikely to work. (This sounds rather unpleasent, it's not supposed to, but my drafting skills are not up to improving the tone today.)

Fundamentally, the idea of a broad, balanced tech tree with multiple viable routes is only going to work if there are really multiple viable playstyles, other wise everyone will have to take the same route, because they are all essentially persuing the same strategy.

Envisage the tech tree as a river valley falling from the start of the game to the end, the more useful a tech the lower the ground there. Players will try to follow the lowest route along this valley, by trying to gain the most useful/best value techs as they go.

If you design a game with really only one good strategy, or where all strategies require so many similar parts that they differ by little in tech needs, then you are going to have one river course which is clearly deeper than all the others, the bottom of the valley, and the only way to get players to choose different path is to micro-balance the benefits of all the techs to try and 'flatten the valley floor' i.e. to make them indifferent between techs. This is going to be very hard to do, if not impossible, and almost improssible to mod without breaking that balance.

If you design the game to allow several distinct strategies on the other hand, then you have a valley with multiple streams through it, with little hills of not so valuable techs between, the streams crossing each other (at the really good/key techs) as they go. Players following one strategy or another find different techs the more useful depending on where they have come from. There may well be one route that is the best (deepest), but all of them are good choices, and the player has a real strategy choice at several points where they choose the route they want to take, rather than just the opportunity to pick a tech that isn't the best choice.

Now obviously this needs a fairly broad tech tree to allow several coherent parallel routes, but I think it gives you more chance of a fun tech game without needing enormous playtesting and tweaking.

(eek, a river analogy, too much coffee for me today )

Ah, there's the point, we can design our combat model that way, but unless we realise we want to, we won't. If we design a research system based on one from a game with these distinctions, because we like the way it works out in practice (which I think I'm beginning to see), but the system we're basing on works because of the distinctions, and we don't have those distinctions, will the system still work? What do we need in order for it to work? I think those are the kind of questions that need to be thought out, partly because they're relevant for the design of the research model, but also because they feed into the general design concept.

A question, in HOI do the Germans have a prediliction (i.e. national advantage) for tanks and the British for navy, or is it a choice the players make based on the strategic position in which they find themselves, which would be reversed if you moved the British player to Germany and the German to Britain? (I had the impression you were saying it was the latter, if the former then I'm less impressed with the design. I think we should aim for a system which produces the second effect if possible.)

Sorry about the triple post but I'm off out and wanted to get the ideas and questions down.

I'll try and answer your question in reverse order:



It is some of each, but really the latter. The only way in which it is a 'national advantage' is that they start with more of that technology. One thing I am wrestling with is the controversial notion of not focusing exclusively or even mostly on the classic Orion setup of 'everyone starts with one planet.' This has produced, in all three MOO games, a situation where there is really very little diversity in playing styles. You cannot play as a pacifist; you must conquer, and you must build the most number of the largest ships (except in MOO1, I seem to recall, where having hordes of smaller ones was useful). There is never an advantage to building a corvette instead of a battleship. And while I'm leery of making decisions based on a combat system we haven't designed yet, I know that one of the cornerstones of what I intend to propose is that we -must- somehow emulate aspects of combat engines like Total War or Homeworld (more the former) that give actual, strategic advantages to each unit; light cavalry (the corvette in this case) is useful because of its speed, its ability to pursue routing units, and it is sometimes used as a long-range missile unit rather than for cavalry charges. This is the kind of distinction I think we need to really make a great game. I cannot admit to having the whole thing designed or even say that FO will choose to go that route, since I am only one voice of many; but it is a defensible argument, anyway.

Back to your question: It's really based on strategy. The British player could choose to forego their navy entirely; they risk losing their many convoys from their colonies that bring in necessary raw materials, but the gamble is that they could reinforce Belgium and the Maginot line enough to keep the Germans from ever getting to the point where Vichy France was created. There are several ways to play as all of the nations; in one game as the US, I spent a lot of time influencing Italy away from Germany, so they never joined the Axis. That is also quite a gamble, but it happened to work for me in that occasion; it quite easily might not have.

If you reversed the British and German players, their strategies would change quite a lot, yes.



This is really the heart of this entire discussion. It is difficult for a project like this to plan out the part of the game we're in without also designing something that isn't slated until later on, but we really cannot do it in total isolation. I am arguing as though a model for space combat looks somewhat like what I'm planning will turn out; it may not, and it may be that we will come back and revisit the tech tree in 0.4 and 0.5 and change the whole structure. I think the tech tree has to be mutable all the way through the game, that we have to evaluate it at every step. But this is also dangerous, because the people we have here and contributing now are very different from the people we had when we were designing v0.1; and part of my job is to maintain some consistency, if, for example, everyone buys into some of what we've been discussing now and then the crew hates it for v0.4.

The core of my argument so far:
- A working research tree can be defined as one without (m)any easy exploits, in which the same player would follow different paths through it depending on their strategic situation;
- None of the MOO series ever had a working research tree, as defined;
- While we must be careful of designing anything in this early phase based on assumptions of what will happen when we're building the space combat engine, we must also guard against making assumptions based on how MOO chose to solve these problems; I consider it more likely that we will develop a system that has more in common with games that influence us now than we would reproduce most of features of any of the MOO combat engines; that is to say, I do not give additional credit to a solution to a problem simply because that is what MOO did.



That's actually the opposite of what I meant. MOO3's solution to avoiding problems in the game's design was to hide most of the design from the player, and to change random pieces of it every time you played. Also, though, MOO3's 'design' changed 6 months before it was released. I have spoken fairly extensively with Alan Emrich about their process in the interests of dispelling rumors and seeing precisely what lessons could be learned, and it is from him that we got our mantra of #1) KISS (no rule that could not be calculated or explained by a human player) and #2) Steal what works before building a new system; there are very few 'new' systems and the real danger of pride is thinking that we can, from scratch, do better in terms of raw, fun-game design than people who have been doing this for decades and not borrowing what works on those grounds. Perhaps we can do better on some things, but I subscribe to those two rules.



This paragraph contains our goal and our challenge. The idea of a broad, balanced tree with multiple routes only works with multiple playstyles; therefore, we must labor to support multiple playstyles in everything we do, and this is a prime example.

my brain hurts now. good discussion!

_________________
Surprise and Terror! I am greeted by the smooth and hostile face of our old enemy, the Hootmans! No... the Huge-glands, no, I remember, the Hunams!

Right, so where have we got to. A first bash at a summary of the current jumping off point (this post is likely to be reedited as people complain I'm misrepresenting them.)

FO is going to have a tech model based mostly on HOI. So far this means things like

Research projects will take a number of research points for a number of turns, not just a total number of RP's.
More than one project will be able to be researched at one time.
You will (probably) have to have the required number of RP/turn available to start a project.
We need to think what to do about the spare RP's left over each turn (either another use for them (is this the place for refinement?) or bear in mind that they will need to be easily redirected back into the economy somehow.)
Tech will be split up into theoretical advances (mostly serving as level markers with limited practical benefit) and applications (and refinements).
Theoretical advances will unlock groups of applications (i.e. each level has a bunch of applications).
Theoretical advances mostly depend on the theory before them.
Practical applications will have prerequisite applications which must be developed before them, often from different branches.
We would prefer a broad tech tree with multiple viable routes through.
Categories should be as directly related to gameplay and particularly strategy as possible (i.e. no 'mathematics' category, but 'Ships' might be o.k.)
Research cost will scale with galaxy size.

Open questions:

Randomly block out different techs each game or not?
Many categories (10+) or fewer (6-8 )?
Will it be possible to develop applications without knowing the theory, if all the prerequisites have been met, maybe at some kind of penalty?
Should we have techs that reduce the cost or time of research generally (or for particular categories). ?
Should developing related (but not prerequisite) techs reduce the cost /time of developing a tech?
Do we want to repeat the outline of the tech tree to extend the research game?
Should we have separate practical and doctrine categories a la HOI, or merge them together? (is this important?)
Should we try to hardwire choices with race bonuses to particular techs, or try and make it dependant only on strategic position.
Do we want to achieve multiple paths by balancing all techs perfectly, or by deliberately introducing several 'prefered paths' or tech families?
How can we produce genuinely alternative gameplay styles and support this with the tech system?
How do we do this whilst keeping things relatively simple and fun?
how big/small do we want the steps of research to be, ie.e how much of adifference will each discovery make to your strategy and tactics?
Do we want to design in 'bursts' of discovery?
What happens to research on projects you can no longer afford, or want to cancel?
Do we want more research points to mean more creativity (wider research) or highwer tech (more advanced research)? If the latter, how do we get it with this model?
How much do we want to change the relative costs between techs as a result of previous decisions? Will this just complicate things?
Should available refinements be strongly based on the prior decisions?
Would developing fundamentally distinct unit types (which serve separate purposes and are not strongly substitutable) and differentiated terrain encourage varied strategy?
Should we design around the classic moo start (one planet), or a predeveloped empire start?

And the Guiding priciple


p.s. i know this is biased towards my ideas and questions, because those are the ones I can remember when writing this, I'll go back through the thread and pick up the other points tomorrow (i.e. in about 12 hours), I know there are several good points I need to include.

One point regarding the balance of the tech tree is : this is an open source game. If it turns out there is one 'uber' path through the tech tree that everyone takes, every time, then the game can (and should) be reworked to solve the problem.

I'm not saying that you should plan for failure... but IMO its pretty much a law that you cannot predict all of the strategies and exploits that people will come up with in a game with this many complex interrelating factors.

Okay then, yes... i guess i am saying that you should plan for failure.

This is, IMO, one of the greatest potential strengths of freeorion. Imagine what Moo2 or Moo3 could have been if we'd has access to the source code and been able to fix all of their shortcomings for ourselves.