Return to “Suggestions”

Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#242
davdav wrote:
Cornflakes_91 wrote: your "lack of discovery" assumes that theres information already available somewhere and if its available to someone you know and can contact.
I'm a little bit desapointed by that. Is there a topic about discovery in LT? I can't find it.
With speed of light limit disabled for every points of the game, the player could buy an universe map at the first station and there is nothing to discover, it's like google maps.

Sorry for the HS.
This assume that the universe has universal trade and is a friendly peaceful place, and every society is at their technological best. :monkey:

Oh sure you could get a star chart, but it probably wouldn't reveal where every wormhole went. Since hostile systems probably don't welcome explorers.
"A sufficiently advanced technology is indistinguishable from magic."
- Arthur C. Clarke
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#243
davdav wrote:
Cornflakes_91 wrote: your "lack of discovery" assumes that theres information already available somewhere and if its available to someone you know and can contact.
I'm a little bit desapointed by that. Is there a topic about discovery in LT? I can't find it.
With speed of light limit disabled for every points of the game, the player could buy an universe map at the first station and there is nothing to discover, it's like google maps.

Sorry for the HS.
You can't buy map of the universe right from the start. Data will spread like viruses - through nearby hosts and at some point data will become so common it won't be profitable enough to sell it. To be "infected" with "knowledge" you need to: a) meet someone who has it and b) be able to pay for it. At some point you might get large map, but given that map will grow and grow, then there will still be blank points left to explore on the fringes, reaching ever far away. Explorer-like NPC can sell you larger map than an NPC-trader who never left his home system, but he won't have full map anyway, because trader doesn't need it and explorer can't be everywhere at once. And what happens when said explorer dies by hand of a pirate? All his data is lost.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#244
I have read over this thread several times, and think that one of the barriers to a satisfactory solution is the idea that research in LT should correspond in nature to research in reality, especially in the sense of progress and technical superiority we feel in comparison to the 17th century. I agree that there should be a loose connection, but I think it is the feeling of progress which is more important than actually making progress. I think LT needs a healthy dose of eternalism. My idea is similar to one's proposed before, and also combines a bit from others.

What I propose is that the “breakthrough rate” is not a fixed number, but instead is an oscillating variable unique to each area which can be researched (weapons, shields, power generation, thrusters, etc) ; that these oscillations have VERY long intervals on the order of 100-500 hours of unpaused playtime, the specific interval for each area being determined by the seed.
Image By areas, I am not talking about research on any particular blueprint, but as a global variable for each branch of the research tree distributed over all research in that branch by any and all players. Already we know that ~90% of all research will just shuffle stats around, but every so often there will be a breakthrough. My proposal would say that instead of X% of your individual research resulting in progress, X% of all research in that area will show some progress. I am of course drawing on McDuff’s notion that any tech’s value is the average of its component stats. My suggestion for a variable “breakthrough rate” is simply saying that the sum “Value” of a newly researched technology should vary; each tech however can of course allocate this “value” in a myriad of ways.

Here, I divide tech into two branches, Limited Tech and Unlimited tech. In actual gameplay, these should be indistinguishable, but in their macro structure, they are very different. Limited tech are those areas which warrant research, but are limited by the engine itself (I think the only hard limits should be engine limits). Unlimited tech is everything else, where the only thing that grows is numbers; if you want to play long enough to have 10e40 dps against 10e43 shield strength, you should absolutely be able to do that. With unlimited Tech, only the ratios matter.
Image What you would predictably see is that as each area has it’s positive breakthrough rate in a parabolic arc is an S curved technology level, I call this half of the interval the Age of Growth. However for the Age of Decay, I’m not quite sure how to go about that. Perhaps it would be through the regular destruction of blueprints, but I have my doubts. What I propose is that blueprints themselves have a lifespan… a very long lifespan, but ultimately finite, perhaps 1/20-1/10 of an interval. At any rate, in the age of decay, the “sum Value” of new technologies in the given area of research needs to go down. Again, this is why this idea is on such a macroscopic scale, so that pretty much no one notices that no one seems to be making any progress in a particular area. That is also why I have split up the variables into the various areas, so that even if no progress can be made in shields, some progress can be made elsewhere (areas where no one can make any progress and are often losing sum tech value in all areas would be true dark ages).

For tech which faces engine limitations, there is the additional fact that you simply cannot exceed certain values for particular stats. Taking absolute speed for example, you wouldn't ever actually research on absolute speed, you would research thrust per unit of energy. I'm not sure exactly how, but I suspect you could tweak the research here so that a super powerful thruster can never be below a certain mass so that X= power * mass + other stats ; X<= max... additionally, if the decay in technology does not lower technology values to their original point, in the next interval, breakthroughs must either be smaller or less frequent in proportion to the distance from maximum Tech value, however the theoretical maximum at the end of any age of growth remains the same.

For unlimited tech, the absolute best thing you can hope for is that you get to keep the best tech of the previous age of growth and wait it out until the new age, and then begin making breakthroughs again. The worst possible thing is a total collapse, where an entire area has lost more than it gained in the whole of the previous age. What you would expect though is that a moderate Sum Value has been lost in this area, and a renaissance of sorts will bring new advances, bringing them beyond the capacity they once had.

Because these are very macro in scale, neither the player nor the AI will really know what sort of age they are in, only the biggest players, who dominate the research being done in the simulated world will even know the rate at which progress is made, for small scale researchers research is very stochastic, but for players who dedicate themselves to it and have massive R&D centers, it will ultimately be a bit more predictable.

What you can expect over a very long time is that at different points in time, different areas of research will be more or less effective, that different areas of the game will dominate play. Image Some of the benefits of this idea that tech exists in eternal oscillations include the fact that you can have tech levels start at pretty much any point you like. Long term LOD could be fairly simple, by applying however many intervals you like and then randomizing how much was lost in the age of decay; tech would never break the engine because it is impossible to advance further than the engine allows. Also, because each area of technology has a different interval and breakthrough rate, no single strategy will work forever; sometimes shields will be highly developed, other times weapons, sometimes both, sometimes neither.

The most notable con though I that in gigantic scales of time, where the game has been running for thousands of hours, the different intervals will likely never be as close as at the beginning. shields may eventually get to a level where they ALWAYS far outpace weapons, engines may never be able to fully power everything at full capacity or hardpoints may never be able to use all of the energy... But I think that to anyone looking for such a long game, this might be more a feature than a bug...

So, thoughts?
Image
When you're trying to fill an infinite multiverse, if you're not willing to consider the entire creative output of humanity as a starting point, you're wasting your time.
User: JoshParnell is accountable for this user's actions.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#245
That cycle strikes me as pretty artificial.

Varying research progress was the result of different organisations being in power, wars, famines etc.

Some cyclic behavior is exhibited from the economic cycle, not from within any mechanism inside the "economics of science"

Tldr: as far as i know there never was any internal reason for scientific stagnation, only external ones.


And i'd also not time out blueprints, if at all i'd only let time out blueprints which have no utilised children or are used themself.

A technology which is in usage wont be forgotten as you are using it and know how it works.

A separation between research data and production data in this respect would be useful.

You have a single instance of the data used to create the device which you can use to create stores with production data, which only contain the actual data needed for building the device (and from which you can mint assembly chips).

So you can loose the knowledge about its inner workings independently from the knowledge of hoe to build it.

"We dont understand the jumpgate technology of the old empire, but we can still build them"
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#246
Hyperion wrote: ...
What I propose is that the “breakthrough rate” is not a fixed number, but instead is an oscillating variable unique to each area which can be researched (weapons, shields, power generation, thrusters, etc) ; that these oscillations have VERY long intervals on the order of 100-500 hours of unpaused playtime, the specific interval for each area being determined by the seed.
...
I hope all this kind of parameters will be parameters in the game. Not every body will be able to spend 500h in the game, even if it's the best game of the century !
Actually, I would like a boredom parameter : when you began to be bored by the game, you can increase the parameter and usually long actions (research, construction, ...) are shortened by the AI and the player gains more credits for missions. It could enable even short timed players to see almost all aspects of the game.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#247
hyper and i have cooked up some advanced version of his "cycle of research".

basically every blueprint has a timer on it.

this (long) timer starts to go down once you stop researching on it, and you can increase the timer again to its maximum by performing research on the blueprint.

when the timer runs out, the blueprint becomes "unknown knowledge", which prevents the blueprint from being used for research but you can still mint assembly chips from it.
so you may forget how the technology works, but you can still replicate it.

a "timed out" blueprint can be made known again by performing research on it.
this research serves only to understand the technology contained in the blueprint and not to design some new technology.
the research power needed to understand a blueprint should be detetermined by your relative tech level.
if a blueprint is far above your technological knowledge, you may not be able to understand it, and thus not able to improve on it.
(this should also apply to general reverse engineering btw)

the timer is specific to every player, so someone who steals your blueprints isnt instantly able to understand it, but first has to perform some analysis on the data he has stolen.

this sheme allows knowledge/research progress to be lost, while preserving the capability to build the contained objects.
"we have forgotten how the ancients jumpgates work, but we have retained their construction plans"

this opens up for economic/social effects to harm research without losing all the capabilities that research gave.
maybe an economic low forces the empire to stop researching, causing it to forget many of its technologies
etc.

this also opens up the possibility of "lost knowledge" from old empires.
and we dont even need special mechanics to prevent you from reverse engineering it and skipping many rungs of the research ladder.
some funky elder race tech, which is far above your own, which can make you much more powerful, but you dont understand it and can only copy it.

Code: Select all

<+Hyperion> Hello
<+Cornflakes> ohaiperion
<+Hyperion> how are you
<+Cornflakes> having done a test in university today, and currently writing more in "reinterpretation of research"
<+Hyperion> hmm, i havent read the latest in that thread
<+Cornflakes> its about armor
<+Hyperion> saw your reply to my idea though
<+Hyperion> its an odd comment you made
<+Cornflakes> i dont know
<+Cornflakes> im bad with words
<+Hyperion> that my cycles seem artificial
<+Cornflakes> i basically just disagree with the cycles coming from inside research itself
<+Hyperion> why?
<+Hyperion> I know its not realistic, but neither is friction in space
<+Cornflakes> it just rubs me wrong
<+Cornflakes> this forced cycling
<+Hyperion> Well i suggested that because to me its about the feeling of progress being made, not actually making it
<+Cornflakes> we could design some economics aspects that create such an effect in research
<+Cornflakes> some factions/governments banning or strictly regulating research
<+Hyperion> Yes, Mynamewuztaken suggested something like that
<+Cornflakes> note: i actually like the basic idea of cyclic research
<+Cornflakes> but i dont like the fact that it appears "just so"
<+Hyperion> well as i said, the cycles are very large in scale, you would never really notice them at all
<+Hyperion> if you notice them, they are too small
<+Cornflakes> then comes the question: why bother with them?
<+Hyperion> well so that people looking to play very long games have interesting variation
<+Hyperion> I dont expect too many individual universe playthroughs will last several hundred hours
<+Cornflakes> we could archieve the same variation without artificially destroying research
<+Cornflakes> but by cyclically varying the relative research speeds of different aspects/parts
<+Hyperion> Well, I got that idea from gazz actually
<+Cornflakes> that temporary research stuff?
<+Hyperion> yes... i didn't like his idea that it would be a short term buff though
<+Cornflakes> that would be really frustrating, yes
<+Hyperion> I figure that if you give blueprints a lifetime of say 50 hours, you could certainly use them quite extensively, and even then, they would likely become obsolete well before they expire
<+Cornflakes> i'd not use an explicit timeout
<+Cornflakes> as i already said in my reply
<+Cornflakes> unused branches could be pruned
<+Cornflakes> or things that are below a certain treshhold below your tech level
<+Cornflakes> so you "remember" the last, say, 10 levels or so
<+Hyperion> well remember corn, there arent explicit "levels"
<+Cornflakes> and those which are definitely outdated get pruned
<+Cornflakes> but you have research nodes and those have child node
<+Cornflakes> s
<+Cornflakes> so a node which's youngest child is removed 10 generations from it ould be pruned
<+Hyperion> ah, i see what you mean
<+Hyperion> well how would you lower the overall tech level in the universe?
<+Cornflakes> first idea: destruction
<+Cornflakes> second idea: in addition to the pruning based on descendancy we could prune based on "last used" time...
<+Cornflakes> wait...
<+Cornflakes> that archieves nothing in that direction...
<+Hyperion> yes, i had suggested that as a possibility, but im not sure that top tier tech blueprints would really be destroyed all that often
<+Hyperion> they would be at best captured and tech stength would be consolidated
<+Cornflakes> mhm
<+Cornflakes> we have a certain "empires cannot fail" problem here
<+Hyperion> thats why i suggested blueprint lifetimes
<+Hyperion> 50 hours or so of gameplay is likely a long time
<+Cornflakes> i see now where you came from
<+Cornflakes> and i just came up with something that would do basically the same...
<+Hyperion> do tell
<+Cornflakes> every research has a timeout
<+Hyperion> Hmm?
<+Cornflakes> so when you are not researching on it the timer runs down
<+Cornflakes> but this timer can be halted temporarily
<+Cornflakes> by doing research on the nodw
<+Cornflakes> node
<+Cornflakes> (im real-time compressing ideas, so do not wonder)
<+Cornflakes> :lol:
<+Hyperion> ok, so what happens when the time reaches 0?
<+Cornflakes> so when theres no resources to do research, you lose slowly all your knowledfe
<+Cornflakes> its lost
<+Cornflakes> the research
<+Cornflakes> it could be propablilistic, with increasing chances with lower timers
<+Hyperion> but nodes are physical objects... if i steal a blueprint, how does the timer work for me?
<+Cornflakes> you came up with that :P
<+Cornflakes> im just adapting the stuff
<+Hyperion> that blueprints are physical objects? its been like that for a long time
<+Cornflakes> i guess the timer could continue running
<+Cornflakes> nope, with the timeout idea
<+Hyperion> oh
<+Cornflakes> so you are bound to have an idea on that yourself
<+Cornflakes> :P
<+Cornflakes> i guess the timer should just run down (slowly, again in your timeframe)
<+Cornflakes> as usual
<+Hyperion> i see, the way you said it confused me
<+Cornflakes> hmm... better idea
<+Hyperion> you are basically just giving blueprints lifetimes which can be extended if you continue to do research on them
<+Cornflakes> yep
<+Hyperion> ok, i can accept that
<+Cornflakes> on second thought, the blueprints themself could not just vanish, but get into a state of "unknown knowledge"
<+Cornflakes> so you have to reapply research power to analyse it
<+Cornflakes> and make it usable again
<+Hyperion> hmm
<+Cornflakes> caches of lost knowledge: check
<+Hyperion> interesting, i like it
<+Cornflakes> blueprints not magically vanishing: check
<+Cornflakes> this unknown knowledge could also apply if you just steal the blueprint
<+Hyperion> ok, that also makes sense
<+Cornflakes> "unknown knowledge" could also be limited to be only usable for building stuff
<+Hyperion> i would think modifiers would be applied depending on how similar your "known" knowledge is to the unknown stuff
<+Cornflakes> so you can replicate the stuff thats in there
<+Cornflakes> but you dont know how it works
<+Cornflakes> and cannot improve on it
<+Cornflakes> maybe some modificator on the "relearning time"
<+Cornflakes> the closer it is to your knowledge, the faster you can understand it
<+Hyperion> so unknown knowledge is a blueprint you cant do research upon until you analyze it, and known knowledge is a blueprint you can do research upon
<+Cornflakes> yep
<+Hyperion> Ok, this does make sense, and I am definitely getting an Aristotelian treatise coppied by an 11th century monk vibe here, which i like
<+Cornflakes> i think we can make that idea "public"
<+Hyperion> ok
Last edited by Cornflakes_91 on Tue Mar 01, 2016 1:56 pm, edited 1 time in total.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#249
I was initially thinking "there's no way a system is going to be able to handle hundreds of thousands of timers at once.." then realized that all you need is for each blueprint to have a timestamp that either gets updated due to research, or doesn't. Then, next time you want to run the blueprint, you check the timestamp against the current time, and see what results.
Spacecredentials: looks at stars sometimes, cheated at X-Wing vs TIE Fighter, killed a titan once.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#250
Interesting. If we leave aside reverse engineering for a second, which to my mind still raises questions over the potential for floating nodes on your tree and unknown technologies and materials, there's one immediate thing I may suggest for this timer business.

The timer applies to all nodes on the branch of the tree, rather than just a single node. Thus if I have a Y-shaped tree, and I research on the left-hand branch, all nodes back to the root node are "live" - i.e. they can be immediately used to create a new branch. All nodes on the right-hand branch however are on the timer and can potentially be forgotten.

If I reselect one of the nodes on the right-hand branch and spend the time to reanalyse, then all nodes beneath it are immediately live again (nodes above it would not be).

This would mean that if you never bifurcated your tech tree then you would never run into this problem. Conversely, it would be a reason to avoid splitting your branches too often.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#251
One thing about "unknown knowledge" which Cornflakes forgot to mention was that analyzing these pieces of technology would also be affected by the breakthrough rate, so if that area of tech had a negative breakthrough rate in place, it would take significantly longer or might even be impossible to analyse the unknown knowledge in your possession.



Another thing discussed in IRC with cornflakes and others is how you would get an AI to choose an objectively worse technology.

We don't really know how the AI evaluates whether what their research produces is better or not, so this might already be taken care of.

What we can assume though is the AI will look at the sum "value" of the technology, they will look at the specific stats, and that they will look at the difference between the old and new blueprints.

The way we hammered out is that an AI will sometimes be willing to go with a worse overall piece of technology, if the stats it cares about are high enough above what it has presently.

So for example, you have a given blueprint with stats: 2A, 2B, 2C, 2D, 2E. This blueprint has a sum value of 10, meaning that any research done on this blueprint can shuffle a maximum of 10 points around, unless you have an advancement or a depreciation. If your research is specializing to raise stat E, you can have a blueprint with stats 1A, 1B, 1C, 1D, 6E. and the value of this technology is still 10. But, if you had a result with stats 1A, 1B, 1C, 1D, 5E then the value of this technology is only 9.

Now, if the AI, is first looking to see whether their research has resulted in an advance, a specialization or a depreciation, this second result would obviously be a bad choice to go with, because any further research on that technology would have only 9 points to work with. However, if the AI is primarily considering the difference in stat E between the old and new blueprint, the worse overall technology is in fact better for their purposes.

By considering first and foremost the differences in a particular stat between what they already have and what their research provided, AI can in some cases be led down paths to ultimately pretty crappy technology. If combined with blueprint timers and a negative breakthrough rate in place, they can get stuck with it until the breakthrough rate returns to positive numbers. But to give the AI a bit more sense about this, they should in fact have some consideration as to the absolute tech value of their new blueprints.

Lets say that when an AI evaluates their new technology, they first look at the difference in the stat they want, but then they look at the absolute value of that new technology; if the technology's value is below a certain threshold, even if it is superior in the stat they are looking for, they will discard it (or sell it). Here, I think we could bring in the "Intellectual" AI trait, making it a bit more useful. An NPC with Max Intellectual would have 0 tolerance for worse overall technologies, and will refuse to pursue any research on such a blueprint. An NPC with moderate Intellectual would have some tolerance and would be willing to do further research on a blueprint that is for the sake of argument 80% as good as the previous one, so they would be fine with the 1A, 1B, 1C, 1D, 5E blueprint, but wouldn't be fine with a 1A, 1B, 1C, 1D, 3E blueprint. And Finally an NPC with little or no Intellectual would barely if at all consider the absolute value of the technology, and would only look at the particular stat they care for.

Going back to my original Idea, the only way you would realistically expect the overall level of technology to even stay stagnant in an interval's Age of Decay is if NPC's with Max Intellectual continually do research on the best technologies for the entire time. If they stop doing research, the tech would soon lapse into being unknown knowledge and they may not be able to recover it. If the NPC's do not have Max Intellectual, then they could be lead down paths which give them worse and worse overall technology because it was better in the stats they were looking for.
Image
When you're trying to fill an infinite multiverse, if you're not willing to consider the entire creative output of humanity as a starting point, you're wasting your time.
User: JoshParnell is accountable for this user's actions.
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#253
Sorry for the semi-incoherent look to this, I can't explain it better while my brainstate is at zero (which it seems to be at today)

Okay... see I have always seen this as an infinite progression personally.
Where research towards a thing would slowly increment all values, asymptotically, towards infinity. (defined as the max value of the int size of the value) ( 255, 65535, 4294967295 )

The loss of progression and gain of it would be emergent properties of the core game-play mechanics around this.

First, a Research Lab module would be assigned a technology, and they would produce Theory Documents. (Invention)
They could also be assigned a Prototype and they would produce improved Theory Documents based on the ones used to create the Prototype. (Iterative Design)
They could additionally be assigned a Device and they would produce an inferior set of Theory Documents based on the ones used to create Device. (Reverse Engineering)

Theory Documents would contain some raw numbers for each of the stats for a module. Things like, fire rate, range, damage. And would display these relative to the average score on each one.
The scores would be simple numbers between 0 and 65535. Defining only ratios for bonuses.
The total tech level would be based on the average of all the stats, and would give slight improvements across the board to the technology.

Example: Laser Theory Documents might contain Range +25% and Damage +5%, along with Rate of Fire -30%
At a stat average of 1337, this might add a +17% bonus from the base version of the item, BEFORE the other bonuses above are applied. ( leading to a Range of +46.25%, Damage of +22.85% and Rate of Fire of -18.1% compared to the basic version )



Theory Documents would then be shipped to a Design Lab which would consume the Theory Documents and produce a Prototype and an Original Blueprint.
The Prototype would be an exceptional version of the item that could be produced from the Original Blueprint, with higher mass and volume. These Prototypes
would primarily be useful for researching at a Research Lab to produce the next generation of technology.

Example: Laser Mk2 Prototype and Laser Mk2 Original Blueprint


The Original Blueprint could be copied into Limited Blueprint Copies, or used to manufacture the Device inscribed in it. Each copy created, or each run of manufacturing would damage the Original Blueprint and it would eventually be destroyed.
The Limited Blueprint Copies could only be used for manufacturing. However they can only be used a set number of times before self-destructing.

Example: Laser Mk2 or Laser Mk2 100 Run Blueprint Copy


The Devices created could be used in a Research Lab to reverse engineer them, giving you a lower quality version of the Theory Documents which you could then use to produce a Prototype allowing you to research this and start getting a leg up on the research.




This cycle would allow players to interrupt it at any stage, attacking the courier with the Original Blueprint would prevent that faction from creating new copies of the blueprint, or new instances of the item it codes for entirely.

Capturing a foe's Prototype allows you to research it instead of them, setting back their research (as the will need to reverse engineer their own Deviceto get back to where they were, or research the last Prototype they had again)

Selling Blueprint Copies is a valid economic tactic, allowing a research heavy faction to get instances of their designs out in the wild.

Capturing high quality Devices from super-high tech would allow you to get into the research game at a lower level.

With every item destroyed, a player could start a research lab on a task with no prototype and get the most basic of devices created, in case of a massive disaster.

I would also suggest that the most basic form of every item can just be created without needing to go through the full process. (minimum positive stats, maximum bad ones) Which means that so long as you have manufacturing you can restart from scratch. (build a research lab, and a design lab, and you can start to produce better items to replace the initial ones)
°˖◝(ಠ‸ಠ)◜˖°
Toba - A Development Dump
Post

Re: Squaring the "Vertical Progression" Circle (maybe)

#254
I remain of the theory that it's a design error to include a research mechanic in a game that has no defined ending. My solution to the vertical progression problem is: don't do progression.

Play long enough, and one of three things must become true:

  • Closed: Technologies are capped at a reasonable maximum power. After that, the research mechanic is useless to the player.
  • Open, Unbounded: Every advance is valuable; eventually some character can destroy the universe.
  • Open, Bounded: Each advance in a field is less powerful; eventually research becomes pointless.

I don't want this to be true; I'm persuaded that it is true. But I might be wrong in this view that research progression eventually must become either overpowered or irrelevant in an non-ending game. So I'm open to friendly thoughts on where this line of reasoning might be mistaken, and on how research might be designed to avoid either of these fates.



That said, if research does get included in LT anyway, there's another way to think about how the AI might go sub-optimal, and that's over time.

A good AI will have a long-term/global goal, for which researching various technologies is one component in a top-level project. This AI will accept research paths that are not great individually as long as the odds are good that they'll lead in the shortest time to the overall desired research goal(s).

An OK AI will have a medium-term/regional goal, and will try to go in a straight line for a target technology even if that winds up missing out on other techs and is less than advantageous over the long term.

And a poor AI will have no goal beyond the immediate and local, and will go for whatever research seems to deliver the most value for the AI's current problem.

Online Now

Users browsing this forum: No registered users and 1 guest

cron