Nomads

[shade]

Runs of Nomads, Gold, Techs etc ... if you look at a sequence of computer generated random numbers - these runs of similar values are quite endemic -- I am no statistician, but I suspect that this is also a feature of true randomness - whatever that may be ...

I'm not sure what you mean but ,have you ever seen a list of 'random numbers' produced by a randomfunction of a computer?You'll note that there are most of the time 2 possibilities:1) the randomfunction always gives the same list of randomnumbers when fed with the same seed.
2)if the calls to the randomfunction are very close to oneanother you get list with the same number repeated af few times (like 3 3 3 4 4 4 4 4 4 7 7 7 7 2 2 2 2)
the reason?
=> most randomfunction are just something that should simulate something randomlike but that can be repaeted for programtestingreasons
=>they just suck(only my opinion so don't try to shoot me for it)
=>a randomfunction that gives really 'randomnumbers ' each time uses a timer and are most of the time a lot slower then the above(in programming=> SLOW=BAD && FAST=GOOD)

btw randomness is hard for us ppl just ask someone to give a list of randomnumbers and check if they look random to you.
(had this discussion before on an other thread ,there the question was how to create an real random number function,etc)
shade

[rah]

You're using crappy randomizers then.

The one I use is based on the starting seed number (can be set or time based depending on whether you want to replicate the results in the future.) the first number is based off the orginal seed and then the second number is based on a seed generated from the first random number selected. Ther results are truely random. Which they had better be since we use it for research. But in small sections of the whole some patterns do appear but they are RANDOM. It happens.

RAH

Gee, I hope that didn't confuse anyone

[shade]

You're using crappy randomizers then.

those i discribed were standard randomfunctions in C or C++ i thought called rand() and srand().
and i agree they are crap(they also work like you discribed)(randomness<>predictability => look at what you wrote )
as far as i know they only simulate randomness .
(in that way for statistical use they might be usefull)

shade
(you need more than that to confuse me)

[Scouse Gits]

I only have a couple of degrees in computational mathematics, together with a healthy dislike of statistics - hence the remark about 'whatever truly random might mean' - but modern random number algorithms satisfy the most stringent tests of 'randomness' and can and are used with complete validity for research purposes.

However, no-one forces cheap PC compilers to implement modern algorithms - so your comments, shade, are not without merit.

An unanswered question is just what algorithms of what quality were used by the developers of our favourite game?

[samson]

An unanswered question is just what algorithms of what quality were used by the developers of our favourite game?

Well, the four-year revolution cycle had a lot of people thinking it was "random" for a long time.
So I don't think the bar is set too high.

[Edward]

Originally posted by samson


Well, the four-year revolution cycle had a lot of people thinking it was "random" for a long time.
So I don't think the bar is set too high.

Ha ha ha! LOL

[Blaupanzer]

Each result our brilliant analysts have produced identifying seemingly random events as having patterns has revealed sequencing rather than an actual randomizer function as the base of the patterns. I strongly suspect that the Civ programmers had no more respect for PC randomizers than SG has indicated. (Rightly so!) So, I don't expect to see later results, e.g., hut patterns, be any more elegant once someone figures them out. Such sequences usually use a lot less RAM to resolve, so may be a good choice under the constraints imposed in broad public distribution. (No one ever has enuff RAM for the latest, greatest game on any computer more than a year old. Refreshing that the Civ folks have not followed that trend.)

[Bird]

If you can be assured of a reasonably high number of nomads once the original second settler is out of the way, it could have a strategic impact on OCC games. Let's say you have a good starting spot and have found a hut that gives you an exploring unit. Use the second settler to immediately increase the city size and bank on finding a nomad to return to the city before you really need the terrain improvements. This will be particularly useful with city locations that have a couple of whales, since you don't need to improve them to maximize their benefits (prior to building a harbor, of course). This may help get to monarchy earlier, plus everything that follows from that.....