Technology System Version 5.1

[Richard Bruns]

The E equation:

There was no E=2 problem on paper and in my calculator, but as I transferred the equation to the computer I mistyped it. In the old system, E was 1+(RTL/10). This was often typed as (10+RTL)/10. I had fixed the equation so that the 1 was replaced by the exponential term. All of the numbers and analyses you saw were based on the equation:

E= RTL/10 + (10+L)/10 ^ (RTL/10) when RTL is positive or zero

When I typed up the equation, I somehow moved the "10+" from the L term to the RTL term.

Sorry for the confusion. I should have proofread that better.

Helper techs:

The CD, like the RTL, is a good idea. I don't understand the name and would prefer something like Relative Helper Level (RHL), but that is a minor issue.

However, I do see a problem in the way that the CD is turned into H. Tech levels are on a logarithmic scale and so is the CD, but the equation H = (10+ CD)/10 changes H in a linear fashion based on CD.

I would prefer something like the following equation:

H = 2^(RHL/W)

W would probably be ten, but there might be a need to increase or decrease it.

With this equation, the helper tech effect H changes at a rate similar to the change in the knowledge level of the civ. Also, there would be no need for an artificial cutoff. H would never be zero; it would simply get smaller and smaller as RHL decreased.

I think this would be a more natural change in H, but it might have a problem; I haven't analyzed it a lot. Any thoughts?

[Richard Bruns]

I disagree about the W factor being so low. First, why should research suffer so much for a Relative Helper Level of -10? That would represent about half of the background knowledge that you would normally have, so why is it unreasonable for research to 'only' be twice as difficult?

Don't forget that H affects RP's before the tech loss term is considered. This means that if H is 0.5 then you have cut research in half before subtracting the considerable tech loss factor. This would almost always be the difference between gaining and losing technology levels.

Also, a small W value will give you insane RP gains for a relatively small advance in helper technologies. If W was three and the RHL was nine, then the number of RP's that you got would be multiplied by eight, even though the helper techs were less than twice as good as the tech being affected!

When you consider O values, the problem gets worse. If the O value of the helper techs was six and the techs were the same level, then RP's would be quadrupled if W were three!

I would actually recommend a W value between 15 and 20. This would greatly reduce these drastic and destabilizing effects.

Even with a W value this high, a RHL of negative ten would probably make you start losing tech levels.

[Mark_Everson]

Good call, you've sold me on it. I'm still on more like the W = 10 side for this reason:
Having Half the needed knowledge should frequently result in Much Less than half the innovation IMO. I'm not so worried about the plus side because one should generally not be able to get far above the RHL because 'free' RPs will lead to advancement automatically.

[Richard Bruns]

We can't make the helper techs too strong. If there are a lot of tech connections, then it would be almost impossible to have a technology that is ahead of the others. There would be no way to invest in a specific technology area. All the money you put in would put in would be wasted as the RHL goes negative and you lose a lot of RP. If W is too low, then you might lose more tech than your investment earned you.

As the system currently works, about 80% of your RP's go to the upkeep of the technology, with the excess going to improve the tech. Anything that decreases RP effect by some percentage will be devastating. Similarly, anything that increases the effect of RP's by some percentage would have a very high effect on tech growth. A doubling of RP effect would multiply tech growth by five, and a halving of RP effect would cause a lot of tech loss.

I think that we should fix the system so that H can never cause RP generation to fall below the level needed to maintain the technology. I don't think that it is reasonable to lose technology because the helper techs are too low. Once you discover some technology, increasing that tech to a level above the helper tech should not make the technology fall.

Perhaps H could only have an effect on RP's in excess of the number required to maintain the tech level. This would prevent the problems with large tech fluctuations due to changing RHL, and allow someone to make emergency investments without severe penalties.

So the formula would be:

Let Re be ((m*RP)-ck)

dk=Re*H if Re is positive or zero
dk=Re if Re is zero

A possible problem with that is that a high RHL would not make tech upkeep easier. You could assume that that education and use will not be affected much ny helpers. I think this is reasonable, especially because I don't believe that a low RHL should make tech upkeep harder.

We could use some ideas and input regarding these matters.

[Mark_Everson]

On 80% of your RPs going to upkeep... I think it's just too much. That will be forever causing unstable situations of the sort you describe. This would be much more general than just being related to the helper tech level. For instance, if pirate activity caused half the merchant fleet to be destroyed, I don't think the people would all of a sudden start to forget how to make ships or sail them! We would be forcing the player to micromanage a lot of things just too insure that tech levels didn't go down... I think that tech maintenance part has come to loom much larger in the model then at least I had originally in mind.

Serious tech loss should probably only come into play a few times during the whole game. The times it would really happen would be during catastrophic civ-wide events like the fall of the Roman Empire. By having a very large fraction of total RPs going to upkeep you're going to lock what happens in the tech model into a very narrow range, that I think is undesirable. I think something more like 20% or 30% of RPs required for maintenance is more appropriate. That way if a huge disaster strikes and nobody builds aqueducts anymore, you can understand people forgetting how to do it. But in general is should not be difficult at all to maintain a level. Of course if you accept this suggestion, you'll need to change the parameters so that it takes significantly more to make progress in a tech area, balancing out the job that the tech maintenance used to have. But unless you can see any big flaws in what I'm saying, that seems like a better way to go.

Anyway, this approach would mean that helper levels could have somewhat more effect without really screwing up the system

I agree with your statements that a fall in H generally shouldn't cause the tech to fall. Only through Serious RP starvation should a tech actually fall.

[Lord God Jinnai]

I think may have a good point on very low helper techs causing tech loss, on the other hand i also think you have a good point. I think we should have some sort of level of importance, from 4-6 levels. This would make it so if it wasn't ness at all it would be level 1, a helper tech that can help but not hinder level 2, level 3 is one that some is needed, not much, and more helps, so on till say level 6, where is would be considered a required tech for certain level and above that helps. Being below drastically hurts.

[alms66]

I'm just wondering if you've even considered the effects of disease, disasters, and even war upon the tech model? From what I've seen so far you haven't. I know this is more refining than actual modelling, but I still think you are missing a vital part. I mean, your equations seem to work well with normal situations, but disease, disaster, and war are not normal. I mean that they are more random than anything else in the game, and you should account for such random things in your equations.

At one time or another, a civ will be faced with one of these problems, so the RP generation, and tech model in general should account for such things.

I guess what I'm leading to is that tech loss on 80% is way to high, even Mark's may be to high (20-30%) to historically account for such things as plague, war, and other disasters. Unless, I somewhere missed the point of even having such things, in which case I don't know what I'm doing working on them!

I think LGJ has the right ideas on levels of importance.

[Mark_Everson]

Richard:

On the E = 2 problem that isn't... That's good to know. Looks good, at least good enough to take a first shot at it in demo 5 using that formula.

Don't worry about the name CD... it was just a placeholder. The idea was what I really wanted to get across. I really like your formula for H. However I think that the coefficient W needs to be something like 3 or even less. My justification is that a deficiency of ten points should result in a huge penalty toward achievement in the area, and a penalty of 1/2 just doesn't cut it. If we used a coefficient W = 3 then a deficiency of ten points of tech would reduce the research rate by about a factor of ten which I think is about right.

With this modification I think the research rate model will work well in the original way we conceived it, without the square root stuff for non-vital helper techs. I will wait a little bit and let you try and analyze it first, and see if you come up with any glitches in this way to use it. In parallel I'll try and return to the use of the economic model technologies as the first test of the system as soon as I can. That way I can see if there's any problems with the new approach from the "actually using it" end.

It's looking Really good!

[Richard Bruns]

I'll see what I can do about making tech loss less important. I am pretty sure that it will simply be a matter of decreasing c and m.

By the way, losing RP's would not have caused your tech to wither away completely. Losing half of your RP's would have resulted in the tech stabilizing at about eight levels below what it was before.

Your people wouldn't forget how to sail; they simply wouldn't be as good becaude they would be out of practice. These kinds of minor fluctuations in skill are fairly common.

[Richard Bruns]

I think I have something that will work, but I need more information. How many turns will the game take to get to the year 2000? I need to know the 'normal' amount of tech gain per turn.

[Richard Bruns]

I have already proposed that helper techs only multiply the RP leftover after upkeep was paid. I assumed that had been agreed on.

Also, we had discussed two societal factors, I and E. I means innovation or influence, and is the effect of societal factors on tech growth. It would affect m. E referrs to education or the difficulty of maintaining technology, and would affect c.

We are ignoring these for Demo 5, but I had planned on asking Garth to put a place for them so we can work on those equations when the social model has been finished.

There would probably be some kind of inverse relationship between I and E. Rigid societies would have little problem maintaining technology, but would find innovation difficult. Societies with more freedom would innovate faster, but unused or unsupported techs would have a much greater chance of falling.

[Mark_Everson]

Richard:

Sounds reasonable. I will leave you in peace so you can actually write the framework in sufficient detail to give to Garth. Changing specific equations and constants should be very little work anyway if there are glitches we need to iron out.

Just to be sure can you provide one sample tech of each level with all the constants that affect them when you put up the final equations. That will make sure I don't blow it when I redo the econ techs!

[Mark_Everson]

Turns will probably be composed of a varying number of years or months like in Civ2. Probably the safest thing for you to do is figure a yearly rate, and then we will have to adjust for the scaling factor between 1 year and a turn. So in terms of years we will probably have something like 7000 years over the basic game.

One thing that still bothers me with the current setup (tech growth being driven by a difference between total RPs and those required for maint) is the huge difference we historically had in mind between societies at different ends of the "innovative" scale. Perhaps the modifiers to RP generation for innovative societies should be applied on only the surplus after maintenance is paid. This has the right feel to me, in than non-innovative cultures don't backslide all that frequently, they just don't make as much progress. We could make maint RPs be paid before mods for helper techs also as you suggested above, merging my suggestion and your previous point.

Opinions?

[axi]

Richard Bruns: I wrote this before the v.5.3 came out, but I will post it here, since it makes sense only here. You asked about the duration of the game in order to compile the normal amount of tech growth. You can look back in the "Clash scales" thread to get an idea of how these things might be. We have to be assured that one year's worth of RPs will give the same increase/decrease in T regardless of the number of turns involved.

Your formula is Tn+1=10/ln(2)*ln((2^((Tn-Ts)/10))+(m*RP-(c*(2^((Tn-Ts)/10))))+Ts
Btw, why are you using a natural logarithm instead of a base 2 one?

Among these, both RP and c are affected by the actual duration of the Civil Turn, but if you adjust them proportionately, the final outcome for a given time period will be different. If we refer T to a given time period (the smallest), we do not have to change RP and c, but the final outcome will still be different. T(to+n*dt)/T(to) is a factor converging to some limit (not e=2,71 this time, since the formula is more complex) as the number of turns involved (n) tends to infinity. The modification of T during a single turn will have to be adjusted according to this factor.

------------------
"In a time of universal deceit, telling the truth is a revolutionary act."
George Orwell

[Richard Bruns]

axi:
I was certain that the logarithms would cancel out and it wouldn't matter what I used, but after looking at it a bit more I realize that you are right to question the formula.

I used the natural logarithm because it was easier to work with. The current system works just fine according to all of my tests, but there might be a possibility of this fouling something up. Unless there is some programming reason not to, we should probably switch to base two logatithms.

The varying turn length could cause problems. I will have to think about that some more.