Check this new ideas here here and here What if the game would be –let's just say it- re-playable. What if one could and would play it again, being assured the game will keep its intrigue intact, but the path to success- in particular, discovering and applying the new alien techs- would be different each time he or she endeavours to such a task? How can that be done? Here is a picture of what a simplified function might look like(2 inputs=x,z,one output=y) Study First of all what would we like to be different in the technology? We would still like to see one technology discovered one after another, graphics depicting each technology and details both fictional and technical describing that technology. And of course, maybe real-life (read in-game real-life) improvements brought by that technology (such as possibility of using laser rifles after laser rifles have been discovered) would be desirable sometimes. So,we concentrate first on real-life (see above) improvements. That said we have a couple of weapon types and their technical attributes. What we want is a new set of attributes. We may write in this regard: F ([object])= Attribute 1 , Attribute2 ,.. , Attribute n Let [ingredients for object] be the resources for an object depends on its class and we can choose H ([object], [ingredients for object])=F ([object]). Therefore, we have now, simplifying a bit: H ([object], [ingredients for object]) k =Attribute k And now a slight twist… instead o f leaving the [ingredients for object] to be strictly depending on the object, we can accept a range of values ([0…1] for the sake of simplicity) and include here a certain set of imaginary ingredients (like how many drafts were made, what cooling strategy was used) situated in the [0…1] range as well. The only observation we make is that the function should be the same over new iterations. A suggested function would be one with hills and valleys, letting local minima and maxims collide over different sets of attributes, so that one different projects make for different attraction points, and trade offs are inherently the main coin here. A suggested form for the function would be: H ([object],[i1,i2,i3…im])=Sum(gaussian functions*(random(2)-1)) An interesting observation is that if we let the universe (e.g. the mappings from inputs to outputs) be the same, the item-production strategy will become predictable once again. In this regard, we can choose to hide the mappings and, at the same time, generate a new technological universe (read randomise the mappings) each time a new game is generated. This allows for a direct link between technologies and game discoveries. We can view the discoveries as very fruitful projects. As long as H are continuous and slow-paced in i1…im (and in our example they do), each new technology would yield a class of projects, behaving in almost the same manner. A small example would be: Hk(object,i1…im)= Sum Pj*e-d(i,pointj)* d(i,pointj)/2 Where pointj are randomly generated when creating a new technological universe and d() refers to distance. P is a number between [–1..1]. Obviously, a filter must be applied so that Hk remains in its range (could be [0…1]). A further refinement would be to add a monotonous function depending on the actual game resources involved in computing the project's value (some of the i1…im as previously established). [Edited by - vallentin on February 17, 2008 10:16:24 AM]

That all really goes over my head. Could you give me the gist of the details?

I can..it goes like this
Randomizing technologies each game.
Technology system means coupling a set of inputs(the raw materials resources needed+some imaginary factors) with outputs(object's in-game stats).
This yield a function f
f(gold,wood,food)=>[Attack,Defense,HP]
that means for different gold,wood,food quantities we get different outputs(so we may have say,different types of tanks[=different ingredients]=>different types:M1A1,missile tank,Abrams,Churchill etc.)

The twist is every new game you have different mappings(technology is new for each new game so prior knowledge of a presumed tech tree does not apply).

So an example of this in action would be a game with random elements to only the tech tree or the items or both? Diablo 1-2 are games just of the top of my head that had random elements to do with drops and item stats.
But beside from that you have to make a game as balanced as possible to be as fun as possible right? Well what if your random tech three gave unfair advantages to the player? Make him invincible the whole game? Bad idea. Make the enemies harder to match the player? Even worse.
I think I am missing the point of your article.

No, something slightly different.
The main classes of the objects (e.g.mobile unit) stay there.We randomize only his attributes(whether it can fly,how hi,hp attack etc.)
The twist is there is no tech tree.Instead you have to discover a vehicle that fits your needs(of course the attributes have some boundaries).The tech universe is the same for all players(in the same game I mean).The difference is who finds out what he needs.It is not so unbalanced if you think espionage exists.And it certainly enriches the gameplay(go to the google thread).


[Edited by - vallentin on June 15, 2005 9:08:39 AM]

Ah, now that you add things like espionage I can see how this can enrich the gameplay. Good call.

Hey some opinions..I am crying desperate for a debate..


[Edited by - vallentin on June 16, 2005 2:57:43 AM]

Could you take difference between a chat and a forum? It doesn't look like there are high cost connections to internet in Rumunia.
I remember on answer on post one year after it. It was on purpose, but 2 days might be norm.