I had to Google a lot of the terms you guys used. Need to go back to the theory, and brush up on my calculus and all that jazz.

How does the research system work currently in detail? If you could link me to a post with the info that would help. A lot of this depends on whether research is done by purely probability, or if there are definite rewards for certain investments of time/money/labor/goods. I'm assuming probability.

I really don't think there is a way around this; Someone has to establish a time line for how many hours constitutes a full game on average. Research being driven by probability will cause that to vary, but there will have to be a target in mind, and there will have to be a cap on technology because lets face it: at some point, you can't make it any better. maybe the player should be able to choose the game type by selecting quick advancement or slow advancement on a slider of some kind?

Using some kind of probability modifier for an entire universe could help control the pace of the game. Each research center has its own probability for creating an innovation, but what if it was also affected by a global modifier based on the number of innovations made in a category in a certain time?

Lets call a type of ship a category and mass a modifier. If the universe sees an advancement in fighter mass, then the probability for any research center to make another advancement in fighter mass (that exact category and modifier combination) beyond the one previously made is reduced by a factor based on the size of the advancement. So an increase of X reduces the innovation modifier by a factor of X^C with C being a constant. This reduces the probability that a new better technology will be discovered, but not the probability that other companies will research a lesser or equivalent technology. That's very realistic. This way players wont have to chase the dragon as far as upgrades go because its likely that the best thing out there will move a little more slowly, followed by other companies catching up and increasing the availability of similar goods.

Keep in mind that this modifies the % chance a research facility has to begin with. So if one facility has a chance of 1% per tick and another has a chance of 2% per tick, and the modifier is -1.5%, then it is IMPOSSIBLE for facility one to research a cutting edge technology, and facility 2 only has a .5% chance per tick of researching a technology that is better than the current market leader. The -1.5% modifier is decaying every tick, increasing the chance of a cutting edge technology from all researchers. The longer without an innovation, the more likely it is a smaller, less funded operation will come up with something cutting edge.

Now say chance has it that research facility 2 has a break through and creates an industry leading technology with its .5%. This upgrade is just a 2% decrease in mass of the fighter. Then the current -1.5% global modifier is increased by a factor of 2^C and becomes -3.5%. Now neither of the research facilities can create a new leading technology in fighter mass until the global modifier decays to a manageable level, or until they increase the inputs of their research to increase their personal innovation %.

What if a long amount of time passes and no innovation takes place? The modifier decays from the positives into the negatives, and begins increasing the probability of an innovation in a certain category/modifier combination. This helps maintain a linear progression of technology that doesn't feel linear. There will be periods of time when the market is flooded with technological advances, and periods afterwards where very little changes in the markets. This should diminish or remove the possibility of innovations snowballing or stalling.

There could also be a way to reduce the global modifier based on how much global research is placed on developing a certain category/modifier. It is more likely innovation will take place if 50% of all research is on the same category/modifier, than if 1% of all research is on that category/modifier.

The slider I mentioned at the beginning could effect the rate at which the global modifier decays, leading to slower and faster games due to the speed at which research is able to innovate. The rate at which the global modifier % decays will be based, through some mathematical wizardry, on the estimated length of 1 game, adjusted by the preference of the player. so much power in 1 slider bar.
Once you reach the length of time a game will last? That's when things would start to get stupid.

Essentially:
1) Let the player have some say in how fast technology innovates.
2) Innovations of a certain type should reduce the probability of similar Innovations based on a normal distribution.
3) Lack of Innovations should increase the probability of innovations of a certain type, based on a normal distribution.
4) Increased research in a certain type should increase the chance of innovation, just the way a lack of research would allow for no innovation.

Long post. As for the diminishing returns? Just let those ridiculous innovations take place and let the player/AI decide. Give them an efficiency ratio of some kind to look at. A base ratio helps but it does come down to player preference on exactly what they need. Hopefully the AI will stop researching new and larger drone types once it realizes that the new types aren't efficient, no one is buying them, maybe its the best its ever going to get. You don't have to cap what is possible, you just have to give the AI and the player a way to realize that drones at a magnitude of 1000x aren't efficient.

I think I like this suggestion. It solves the issue of technological disparity nicely, which should keep a level playing field for the game.

The innovation inhibitor should have some distance based falloff, because noone cares or even knows if the civilisation 50 jumps away made fighters 0.5% lighter, so why should it inhibit my research?

Nice thoughts, Mr Taken. I propose my own thoughts on research in Semi-Markov Research Trees, where I model researching as a generalised semi-Markov process (and therefore making research breakthroughs a very much stochastic thing, like you assume). I'd have to give your suggestion more thought, but I can see it integrating very well with that idea - a global "tech rate" modifier can be factored into the rates of all active clocks within the process and operate in the manner you propose.

ThymineC wrote:Nice thoughts, Mr Taken. I propose my own thoughts on research in Semi-Markov Research Trees, where I model researching as a generalised semi-Markov process (and therefore making research breakthroughs a very much stochastic thing, like you assume). I'd have to give your suggestion more thought, but I can see it integrating very well with that idea - a global "tech rate" modifier can be factored into the rates of all active clocks within the process and operate in the manner you propose.

Yeah... okay... so that's good right? Back to Google.

The most recent and detailed view of research was in Development Update 10.

The universal thing is one way of implementing the sort of scaling mechanic I've been talking about. Easy to catch up, hard to get ahead. I'd also suggest scaling based on a combination of local tech (that in the system) and universal tech (that known everywhere). However, that presumes some form of vertical progression rather than a constant redistribution of trade-offs around a fixed utility. For encouraging people to diverge along very different research paths, you wouldn't want to penalise them for that digression.

Think about the ways monopolies can form - horizontal integration and vertical integration.

Vertical is the kind we're most likely to see in the early-mid game: Companies/factions owning everything from the mines to the ships, fully self contained and in command of their own supply chains, but still producing a limited and specialised product line.

Specialisation can be a partial solution to this because even if people are running mines and factories they'll still have to buy the requisite drones from the companies making them. Requiring blueprints and research for material refining can also help to stave off vertical integration because if you're specialising in building lasers you can't also be specialising in refining alloys*, so it introduces a potential discontinuity in your chain that outsourcing to another firm will fix.

Horizontal is the kind we'll see cropping up in the mid-late game as factions merge and technological cross-pollination happens. Companies moving from being laser manufacturers or railgun manufacturers to being munitions manufacturers, and then integrating ship building, then drone building, until you produce factions which can effectively span the entirety of the space-construction sector (which is the only one the game is modelling in any detail).

* Or... can you? A faction is multiple NPCs after all, and one NPC doesn't have to research the same thing as his faction-mates. Hmmm. That kind of spread will definitely end up producing more integrated companies very early on, since there's no point us all researching fighter design if we're all going to step along the same path anyway. You'll really need to figure out some way of pooling research and rewarding agents for doing so if you're to avoid that problem.

One of the failure modes we want to avoid is too much vertical and horizontal integration at any given time T, even if we accept that a certain amount will happen. A scaled continuous advancement system would keep people focused on their specialisms because no single arm of research could become "full". However it doesn't fix the "endless treadmill" problem.

---

Another thing to think about is the granularity of research. Rate of advancement down a path is a function of step size + time to next step: reversing that, if you set a system that slows down or speeds up the universal "vertical", are you slowing down the time to next step or are you reducing the magnitude of the step size? You can certainly presume that some things will need to be lower bounds for gameplay. While I'm all in favour of a diverse economy, even I draw the limit at a new blueprint for every 0.1% increase. Similarly there's an upper limit to how long any step can make before research becomes basically pointless, especially since there's an element of experimentation involved in pushing forward that may not pay off.

I agree that tech should scale locally as well. Another good modifier to add into the mix. Perhaps that modifier does not decay with time, but rather is a function of the total technological disparity divided by the distance between two locations. This would be a % adjustment to the global modifier? Therefore the area with the highest technological advancement would have the full effects of the global modifier, while an area 50 systems away would not experience the global modifier? This would cause the global modifier to be more severe in the areas of highest tech advantage, and would allow for multiple tech centers in a universe.

With the research, I think it should go by Categories and modifiers. At my research facilities, I choose a category and a modifier to go that category and try and research a superior combination of that modifier/modifiers. For instance, I want the same firepower at lower energy cost. Every tick, the research center generates a technology with the combination of traits I specify, and if it is worse than the current tech I have, it discards it.

Or what if research is handled by changing the possible limits to modifiers, and you personally choose the combination you want in your blueprints? Hard one to explain... Assume there are just three variables affecting a particular item. When making a blueprint you have sliders, each with a minimum value and a maximum value for each variable. The min and max values are determined by research. The combination of these three variables effects the stats of the item with definite trade-offs, like an increase in payload causing a decrease in tracking speed.

This way your research increases the degree to which you can affect the variables, but doesn't spit out specific recipe's you have to follow. Horizontal research would affect the degree to which you could manipulate the variables. Vertical research would affect the fundamental equations that determine how the variables relate to each other. Just thinking out loud.

MyNameWuzTaken wrote:I agree that tech should scale locally as well. Another good modifier to add into the mix. Perhaps that modifier does not decay with time, but rather is a function of the total technological disparity divided by the distance between two locations. This would be a % adjustment to the global modifier? Therefore the area with the highest technological advancement would have the full effects of the global modifier, while an area 50 systems away would not experience the global modifier? This would cause the global modifier to be more severe in the areas of highest tech advantage, and would allow for multiple tech centers in a universe.

i think the "locality" of an breakthrough should be centered on the system where it was developed (or systems when multiple research modules participated) and fall off exponentially according to the distance measured in jumps

McDuff wrote:A faction is multiple NPCs after all, and one NPC doesn't have to research the same thing as his faction-mates. Hmmm. That kind of spread will definitely end up producing more integrated companies very early on, since there's no point us all researching fighter design if we're all going to step along the same path anyway. You'll really need to figure out some way of pooling research and rewarding agents for doing so if you're to avoid that problem.

I've been wondering about that, too.

If research is something done by "research modules," which is more likely: that all modules owned or controlled by a character (including the player character) would be forced to be researching exactly the same thing? Or that each individual research module could be programmed to research any potential tech known to the module's owner? If it's the latter, as seems likely to me, then presumably they could all be researching different things at the same time.

Some bonus for allocating more than one research module to a particular potential tech seems like an interesting idea from a gameplay perspective. It creates an interesting choice between specialization for efficiency and a more general research program. Another possibility is that only modules directly linked to each other (for some cost) can receive a bonus for researching the same tech -- I like that idea because it would support the idea of research facilities. Each facility could be researching different things, but each facility would maximize the project its modules are working together on.

Second question: when a research breakthrough is achieved, who gets the result? Say I control several research facilities, each of which is managed and staffed by numerous NPCs. When a tech is fully researched, who can use the result? Just me? Or anyone who worked at those facilities? What about for an individual ship-mounted research module -- do only I get the acquired knowledge as the ship's owner? Or does the NPC commanding that ship also get the new research? Anybody else?

How much do you trust them?

Regarding the question of local vs. game-wide techs, this has long been an interesting area for discussion. I'm going to refer back to my pet "thread that I think it is a good idea but never gets any attention" , because the idea has a direct impact on how technology propagates from its source.

How would this, if implemented in any fashion, influence the vertical progression problem that appears to concern many (though not me - perhaps I've misunderstood it)? Would it have no influence?

I think protecting technology is better done on a per-item basis. Building in some form of delay (at cost) to prevent reverse engineering.

I went through it here, albeit as the result of a total misreading of something Cornflakes said.

Just a quick note that, mcsven, you did indeed post a suggestion that directly addresses tech propagation, and very nicely, too. Don't know why I didn't remember that.

On the whole "vertical progression" thing, I'm handwaving my concerns about that away by assuming there'll be two rules for applying modifier techs to base tech objects:

1. Every base tech object can have only a limited number of modifier techs applied to it. Maybe that number is affected by the quality or size of the base object, but it's finite and low -- let's say five or six modifiers at the very most.

2. The owner of these objects can swap out modifier techs for new ones. In other words, applying a modifier to a base tech isn't permanent -- you can change it if something you like better gets researched.

Even with Assumption #2, the implementation of Assumption #1 immediately puts a ceiling on how much any base tech can ever be improved through the addition of modifier techs. A cap of around six modifiers means that if you optimize mods to maximize damage or defense or some other individual characteristic, you're almost certainly going to get some negative effects imposed as well -- you don't have enough modifier slots to add the additional modifiers necessary to negate the drawbacks that come with the positive effects of each modifier tech. Vertical progression thus has an upper limit.

If on the other hand there's no cap imposed on modifiers, and you can tack on as many as you want, then yep, eventually all characters will have doomsday weapons just through modifiers, with all drawbacks negated by yet other modifiers.

That latter kind of game might be interesting, but I'm not sure how much fun it would be to play.

I'm not sure I follow this idea all that well, Flatfingers? How does it apply given the research tree mechanic we've already seen? Could you clarify a bit, please?

McDuff wrote:I'm not sure I follow this idea all that well, Flatfingers? How does it apply given the research tree mechanic we've already seen? Could you clarify a bit, please?

From what i read out of flat's post:
you dont research the laser of doom with +15% damage +25% energy need as one piece but have a "base" technology, in this case the laser, and an "upgrade" tech (the +15% +25% modifier)

So instead of "laser of doom +15% damage +25% energy need" you have
"Laser of doom (with some stats)" + "damage upgrade +15% damage +25% energy need"

That's exactly how I'm understanding research, cornflakes -- well said.

Is there something I've missed on the LT IRC channel that has changed how (I thought) research was described in previous video updates? My recollection is that you first discover a base technology, e.g., Laser. Then you can research modifiers (visually clustered around the base tech on the node display for research) that provide at least one benefit such as +20% Armor Damage and at least one drawback such as -15% Rate of Fire.

If that's changed, then the suggestion above may be moot. But if that's still mostly the plan for research, then capping the number of modifier techs that can be applied at any one time to a base tech (not an implausible thing, IMO) seems like a viable solution to the vertical progression issue.

Note that this is not a cap on the number of modifier techs that can be randomly generated for a base tech. You can keep researching those as long as there are different combinations and values for modification characteristics. (Bonus mechanic suggestion: let users discover modifiers that change the same characteristics, and let them choose whether to apply the new discovery to the existing modifier tech to make the good characteristic even better and the bad modifier even worse.)

This would mean that, for each kind of base tech, there'd be a practical upper limit on improvements to what it does. You could make a base tech quite powerful in this way compared to similar techs. But if you really wanted to advance in power through your own research, you'd eventually want to research the next more powerful base technology.

Let's say that the base damage of blasters is greater than that of lasers. You might be able to amp up your lasers with a full set of mods so that they're a little better than a vanilla blaster. But for meaningful vertical progression, you'd need to research blasters. Then you can research the mods for them until you're ready to research whatever's better than blasters.

Note that this does imply an end to research, which comes after you've discovered the best base tech of every kind of object, and have maximized and applied the most desirable modifier techs for each of those base techs.

Does that make the idea of capping the number of applicable modifiers more attractive? Or less?

Or did the design for research change to a find-the-pixel adventure game and I missed that announcement?