Technology System Version 5.2

[Beör]

Richard

But you did say:

quote: Tech diffusion occurs due to contact, not movement

...and this had me confused for a while there .

Now I see that you intend for migrating EGs to pass rather large amounts of RPs to their new civ - in effect negating 1.2.2. The new civ will not get all the migrating tech level at once, but - and here comes my point - if migration continues, the tech level will even out asymptotically at a fast rate. Remember, the migrating EG has the tech level of it's origin civ (if modelled thÃs way), since it doesn't have one of it's own. So IMHO my objection to 1.2.1 still stands.

For conquest:
It seems that you have to keep track of the tech level for the EGs of the conquered squares, otherwise you couldn't talk about

quote: You get them based on the amount of land you own at the beginning of the turn and the difference between that tech level and yours

So even if you model tech at the civ level you have to model tech level of every conquered square - at least untill the tech level of the civ and the tech level of the conquered squares are the same. And what if migration occurs out of the conquered squares (yikes - this could get complicated )

I hate to say this, but: The modelling of tech at the EG level seams so much smoother, and the overhead compared to modelling at the conquered square level is not that big.
[This message has been edited by Beör (edited September 19, 2000).]

[Richard Bruns]

Beör

The amount of tech transferred in the migration depends on the number of people. So a steady, small volume stream would provide about as many RP's as a trade route. You only get big jumps if a huge amount of people are moving.

Modeling at the EG level would give exactly the same tech diffusion result as modeling at the civ level. If part of an EG moves, you get the same result as if part of the civ moved. What is the difference between having the tech level of the originating civ and the tech level of the originating EG? The mechanics would be exactly the same.

quote: You get them based on the amount of land you own at the beginning of the turn and the difference between that tech level and yours

Sorry I wasn't clear. I meant the actual tech level (A, or possibly something else) of the province. This will be modeled anyway. The ideal tech level of the conqueror is compared to the actual tech level of the conquered square. The actual tech level could decay, but that does not require special modeling. For the purposes of this transfer, the ideal level has no impact. If they know how to do lots of things but you can only take a primitive, undeveloped province it won't help you at all.

I don't know about migration out of conquered territories. That would depend on how the social model treats those people.

By the way, thanks for analyzing things like this. It does help us quite a bit, even if it seems like you are not changing anything.

[Richard Bruns]

Beör:

I just saw your social model post. I see now what the problem was. When I say EG, I mean all of the people in the entire civ that are of that group. So the tech would be calculated once for every member of that EG in the civ.

We cannot reasonably calculate technology at any level lower than EG/province. Each round of tech calculation will take a LONG time, possibly as much as 3-5 seconds. Every turn, the computer has to make and sum the RP's, distrubute them to the techs based on tech tags, run equations for over 100 techs, calcualte as many application levels, and then recalculate tech tag levels and apply that to other models. So if every EG in every province calculated tech individually, it could take several minutes. Calculating it by square would require a supercomputer!

I really think it is best to assume that every person in the civ had the same ideal tech level. Calculating tech once for every civ will already take enough time.

[F_Smith]

Lordy:

The 'tech' represented by factories, govt universities, etc, should be 'infrastructure', tied to a specific location, shouldn't it?

The seperation of the two should be along 'functional' lines -- 'eg tech' is a knowledge object. Govt/business tech is an 'infrastructure' object.

The 'default' system of RPs seems to confuse 'infrastructure' and 'knowledge', when they should (in my opinion, of course) be seperate.

[F_Smith]

Richard:

I've said this before, I'll say it again -- there's no way ya'lls assumptions/estimates of processor usage are correct.

There's no way it'll take 3-5 seconds to calculate the tech info.

The newest machines run 2+ billion calculations per second. And the 2 ghz chips are coming out soon! If it takes 3-5 seconds to do the tech calculations, the programmers have screwed up big-time.

[Lord God Jinnai]

Okay, F_Smith first off on your comment about processesor speeds, i don't doubt that, but we're not making this game ness. for next generation technology which is mainly 3D graphics and FMV oriented as far as the reason for pushing the envelope.

Second, infrastructure isn't quite as clear cut as you have it out to be. Infrastructure can be entirely based also on what the people themselves build seperate of government and businesses interferance therefore it is still ness. for atleast a seperate tech listing with businesses esp in post-industrail nations or where that was the parent nation of the business because EG techs as you propose just won't cut it.

[Richard Bruns]

Sure, the bleeding edge chips can theoretically do 2 billion flops. But:

1) Not everyone has the best machine.
2) The OS and JVM take up a lot of overhead.
3) Data is being shuffled around as well.
4) The system bus is rarely faster than 200 Mhz, so the transfer of information will be slower than what the processor can do.
5) Each equation in the tech model will probably take several hundred or more machine level calculations. Powers and logarithmic calculations take up a lot of flops if I recall correctly.

I've seen how long it takes to run mathematical equations in C++. It can take a lomg time to draw fractals, even on the best machines with good code. It takes several minutes to run millions of calculations like the ones in the tech model. Each round of tech calculations will take thousands of these equations.

[F_Smith]

Lordy:

What is ya'lls definition of 'infrastructure'? I don't understand. I would say that 'infrastructure' is any man-made buildings/institutions/structures, regardless of who paid to have them built/maintained.

Which, as I said, seems quite different from 'knowledge', or 'technology'. A factory isn't a technology. A university isn't a technology. They are products of tech, not techs themselves. They can store technology info, but only in 2 ways -- people's brains (ethnic groups, again) or some form of written knowledge transfer.

So the 'maximum tech knowlege' of a civ seems to be determined by their 'infrastructure' (factories, universities, govt agencies, good roads, telephone lines, networks, etc).

But they are not the same thing.

* * *

Ya'll:

RE: processing requirements/usage

3d graphics are an *entirely* different level of pushing data. You really don't realize what you're saying.

Mainly because the volume of data is at a level you almost can't even comprehend -- dozens or hundreds of polygons, each with highly complex 'transforms' (rotations) being done between 16 to 64 times per second -- every second -- all overlaid with large 'texture maps'. Millions of calculations *per frame*. And I saw a new race game with 64 *frames per second*.

A data-engine game like Clash, even if it simulated an entire world person by person, wouldn't push as much d data as a good, modern flight sim.

Heck, a single JPEG can hold more data than the entire game of Clash.

But that's another matter. I'd really rather that just get settled during testing, rather than try to convince ya'll . . .

[Richard Bruns]

F_Smith:

Tn+1 = GV/ln(MV)*ln((MV^((Tn-Ts)/GV))+(H*I*(RP*m - E*c*(MV^((Tn-Ts)/GV)))^DR))+Ts

It's been my experience that processors start to choke a little when you feed them a few hundred of these things, especially when almost every variable in there has to be brought in from somehwere else or recalculated.

But you might know some speed shortcuts that my friends and I don't.

quote: So the 'maximum tech knowlege' of a civ seems to be determined by their 'infrastructure'

Actually, the tech level determines the maximum infrastructure level. Your ability to build is limited by what you know. The amount you can invest in certain infraclasses without diminishing returns is based on tech level.

FE if you put loads of money into production infrastructure before discovering the concept of a factory, you'd be wasting your money. You wouldn't be getting anything useful; the tech has put a cap on how much you can invest. Once you know about factories, that same amount of money can be invested to give a huge increase in production.

But we are still working out the details of that. The new infra model does some things differently than the old model. . .
[This message has been edited by Richard Bruns (edited September 19, 2000).]

[F_Smith]

Richard:

I must not have been clear -- I'm just questioning your defintion of 'technology' for the game's model. Is it correct that your definition of 'technology' is 'the maximum tech knowledge of an entire civ'?

If so, I was just pointing out that sounds more like an 'infrastructure' object.

And assume your simple equation required a thousand calculations to do. At a very pedestrian 1 billion calculations per second, we can do one million full runs of that equation in one second. And yes, I also plan to use preprocessing tricks and some fancy OOD to speed things up tremendously.

[Richard Bruns]

F_Smith:

quote: Is it correct that your definition of 'technology' is 'the maximum tech knowledge of an entire civ'? If so, I was just pointing out that sounds more like an 'infrastructure' object.

How does that sound like infrastructure? The tech is the best thing you could do, and the infrastructure is what you have done. Please explain.

On the speed issue, I'll just say that one of my classmates made a fractals program that was an order of magnitude faster than the commercial program we were using, and he never claimed to be able to get the kind of speed you say is possible.

[Richard Bruns]

We've almost hit the 150 post limit, so discussion will have to be redirected to the new tech thread.

[Richard Bruns]

Yes, I know there some unrealistic effects when you assume that people all over the civ know the same thing. It was a simplifying assumption we thought we had to make.

There are four options for the detail at which we model technology:

Once per civ
Once per province
Once per EG
Once per EG per province

I think that once per civ is good, as the provinces all have their actual tech infrastructure attributes. But if you insist on more accuracy, you can go to the more detailed levels.

[Beör]

Sorry posted in the wrong thread
[This message has been edited by Beör (edited September 20, 2000).]