Home » Stars! 2.6/7 » The Academy » Hab correlation
| | |
Re: Hab correlation |
Sat, 20 January 2007 20:16 |
|
Iconian | | Officer Cadet 2nd Year | Messages: 233
Registered: January 2006 Location: Nevada, USA | |
|
Hab correlation to Location?
A year or two ago I made a race with a hab somewhere around 1/6 or 1/8 and played against the AI with it a bit. The first time I started out in the northwest corner. After scouting about 30 planets, I found that only one or two were green. I figured it wasn't too early into the game, so I just started a new one. Again I appeared in the northwest corner, and again, after several years of scouting, I found that there were hardly any planets available, like a green and a couple yellows. I was a bit peeved at that point, knowing my race couldn't really contend very well, so I started yet again. This time I appeared in the southeast corner of the universe and found that there were a lot of greens. Out of about 30, it seems like there were 10 (or maybe 5 greens five yellows, I don't remember entirely).
Since then I've always wondered what the deal was. Perhaps there is some correlation between grav, temp, and rad. Based on my own experiences with those games though, I'd rather tend to think location has more to do with it. I can't help but wonder if the northern part of the galaxy is always hotter and the southern part cooler, or less radiated, or has lower gravity. Or maybe there are east to west gradients. Or possibly even diagonal gradients.
If someone still has Alan Kolaga's raw planet info, and if it includes the x y coordinates for the planets, maybe someone could sort them to see if there are any correlations.
Those are my thoughts on this somewhat old topic.
EDIT: If I would have to guess, the gradient might be for radiation. The Jeffs seemed to like radiation for some reason--the radram, extra minerals for rad over 90, and the fact that moving the hab to the extremes of rad doesn't actually affect planets available as much as you might think. OTOH, maybe they decided to use temp or grav to balance out rad in this way . . .
[Updated on: Sat, 20 January 2007 21:07]
Yeah, bread too.
Don't Let the Stars! Fade AwayReport message to a moderator
|
|
|
Re: Hab correlation |
Sat, 20 January 2007 21:28 |
|
mlaub | | Lieutenant | Messages: 744
Registered: November 2003 Location: MN, USA | |
|
Iconian wrote on Sat, 20 January 2007 19:16 | Hab correlation to Location?
If someone still has Alan Kolaga's raw planet info, and if it includes the x y coordinates for the planets, maybe someone could sort them to see if there are any correlations.
Those are my thoughts on this somewhat old topic.
|
I've generated many universes in attempts to make very even starting positions in team games. In the process, I have noticed no such trend overall. This is supported by the Kolaga data. Since this link is still active, he may still have, and be willing to send you the data, as it states on the web page.
I have noticed that each universe sometimes has trends in min cons and hab that run in specific areas of the universe. I am not a statistician though, and it is probably within the parameters of "random".
-Matt
Global Warming - A climatic change eagerly awaited by most Minnesotans.
Report message to a moderator
|
|
| |
Re: Hab correlation |
Sun, 21 January 2007 16:08 |
|
mlaub | | Lieutenant | Messages: 744
Registered: November 2003 Location: MN, USA | |
|
Kotk wrote on Sun, 21 January 2007 11:33 | Randomness works funny ... everything that may happen will happen. Statistics just say how often it happens.
|
Yea, supposedly it works that way. Last time I tried understanding it in another game, I failed miserably.
In that game, when you recruited a particular mage, you would get a random magic field associated with him. The number of possible magic fields was 6. During several games were I recruited anywhere from 8-12 mages in a row, I would get really screwy results (IMO). One game I would get 8 mages with the same magic field, in a row! then the next few in another field. Or, I would get 3 fields out of the 6. It seemed that the lowest probabbility was getting 1 represntative of each magic path from 12 plus mages, as it never happened. This went on for about a year of play time.
I decided to gen a long serious of mages in a test game. I genned hundreds. It sure looked like the randomizer would get stuck on a particular field. However when I presented the data, to the game forum, I was told that that's the way random works.
All I know is that if I pull a fullhouse, or roll 7 that many times in a row at a casino, I'd probably get taken out back and beaten to a pulp. So, I don't put to much faith in programers view of random.
-Matt
Global Warming - A climatic change eagerly awaited by most Minnesotans.
Report message to a moderator
|
|
|
Re: Hab correlation |
Sun, 21 January 2007 16:27 |
|
|
mlaub wrote on Sun, 21 January 2007 21:08 |
I don't put to much faith in programers view of random.
|
Here's a quick test with the Java random number generator. I generated a million random numbers between 1 and 6, and counted the number of "runs" of different lengths, where a run is a set of consecutive identical values. So, "3, 2, 4" would be 3 runs of 1, whereas "2, 2, 2" would be a run of 3.
Run: Count
1: 693631
2: 115989
3: 19430
4: 3240
5: 485
6: 92
7: 20
8: 3
As you can see, on 3 occasions we had the same number appear eight times in a row.
If that happened to you in a game, what would you think ?
Report message to a moderator
|
|
| |
Re: Hab correlation |
Sun, 21 January 2007 23:56 |
|
Iconian | | Officer Cadet 2nd Year | Messages: 233
Registered: January 2006 Location: Nevada, USA | |
|
Ah, yes, random numbers. You can see this in testbeds as well sometimes. If you have 100 BB's with 20 missiles each and an accuracy of 36% after computers, you can expect that about 1/3 of the missiles will hit if the enemy has no jammers. BUT, that's only approximate, and with smaller numbers you can see big fluctuations. Bring a single such BB, and you might end up hitting 50% of the time, or 10%--or possibly even 100%, or 0%.
But, let's not get too much into randoms and stuff. I know it's certainly possible that the placement of greens and such is random, does anyone know this with certainty. Kelzar was writing on one post about how he was going to test to see if temp, grav, and rad might be associated with each other. How about location then? Is it possible that, statistically, planets in the northern part of the galaxy have higher radiation than those in the southern part? Or that those on the east have higher temp than those on the west?
It ultimately will be random. As we've seen, the mineral concentrations on all planets are random--but, those with rad over 90 have higher average concentrations. Isn't it possible then that planets with an x or y coordinate over 800, or 1200, or something, have temp, rad, or grav higher on average?
Yeah, bread too.
Don't Let the Stars! Fade AwayReport message to a moderator
|
|
| |
Re: Hab correlation |
Tue, 23 January 2007 17:33 |
|
|
Just to quickly dip into this thread...
I did hab correlations on the Kolaga data (well, I actually generated several fresh sets of data each with 10x as many planets using his tools.) I can state that there is NO correlation between any of the habs. Each is determined independently of the other hab values.
The only correlation I found at all in planetary values, is that those very high rad values (those over 90mR IIRC) have higher average mineral concentrations.
Report message to a moderator
|
|
| |
Re: Hab correlation |
Tue, 23 January 2007 18:07 |
|
|
Iconian wrote on Wed, 24 January 2007 09:58 |
Quote: | I can state that there is NO correlation between any of the habs. Each is determined independently of the other hab values.
|
OK, that's certainly nice to know.
Did you happen to check if there was any correlation between the hab values and planet location?
|
I haven't actually tested that, but I'm extraordinarily confident there wouldn't be any such correlation. There's no reason that I can imagine for the designers to want to do such a thing (and I can imagine good and obvious reasons not to, like balance issues.)
Mathematically speaking, if any position on the map had any correlation with *more than one* hab value, then that would've revealed itself in a correlations between the habs. So if there were any correlation between hab and position, then it can only be in one field. And I very seriously doubt even that much would be the case (but haven't generated data to verify, and don't intend to.)
[Updated on: Tue, 23 January 2007 18:08] Report message to a moderator
|
|
|
Re: Hab correlation |
Tue, 23 January 2007 20:14 |
|
|
I can state from experience that the planet assignments are random (as random as the random number generator in use is anyway). In all the years I have played Stars!, I have certainly played well over 100 games. I've also generated at least 6 times that many universes for various race design testing or testbeds. If planets were at all related to each other by location, I would have noticed it.
Ptolemy
[Updated on: Tue, 23 January 2007 20:14]
Though we often ask how and why, we must also do to get the answers to the questions.Report message to a moderator
|
|
| | |
Re: Hab correlation |
Wed, 24 January 2007 11:04 |
|
LEit | | Lt. Commander | Messages: 879
Registered: April 2003 Location: CT | |
|
Some things to keep in mind:
The 1 in 8 hab is not too accurate, it's mostly useful to compare to other races, not necessarily what you'll find in the game. The method the game uses to find that 1 in 8 is fairly complex, and I don't understand it all, but I don't think it's completely accurate, however, it is a good benchmark: a 1 in 4 race will find more planets then a 1 in 8 (not necessarily twice as many though).
The other thing is that often random number generators are biased or misused. Biased in that they aren't as random as one would hope, patterns and favoring some numbers is fairly common, especially with the default random number generators. Misused as in they normally generate a number between 1 and some max, and then you take the remainder after dividing (modulus) by the number you want, if the number you want is not too far from the max (a common max is 32767) then you will introduce a bias. For example, if your target is 1000, and your max is 32767, the numbers up to 767 will occur a bit more then then larger numbers. For 1 in 6 this can probably be ignored however.
- LEitReport message to a moderator
|
|
|
Re: Hab correlation |
Wed, 24 January 2007 21:43 |
|
Iconian | | Officer Cadet 2nd Year | Messages: 233
Registered: January 2006 Location: Nevada, USA | |
|
Quote: | Some things to keep in mind:
The 1 in 8 hab is not too accurate, it's mostly useful to compare to other races, not necessarily what you'll find in the game. The method the game uses to find that 1 in 8 is fairly complex, and I don't understand it all, but I don't think it's completely accurate, however, it is a good benchmark: a 1 in 4 race will find more planets then a 1 in 8 (not necessarily twice as many though).
The other thing is that often random number generators are biased or misused. Biased in that they aren't as random as one would hope, patterns and favoring some numbers is fairly common, especially with the default random number generators. Misused as in they normally generate a number between 1 and some max, and then you take the remainder after dividing (modulus) by the number you want, if the number you want is not too far from the max (a common max is 32767) then you will introduce a bias. For example, if your target is 1000, and your max is 32767, the numbers up to 767 will occur a bit more then then larger numbers. For 1 in 6 this can probably be ignored however.
|
While I don't quite understand all of that, I gather that Stars! random number generators may not be quite as random as I thought . . . hmmm.
Thank you for the info.
Yeah, bread too.
Don't Let the Stars! Fade AwayReport message to a moderator
|
|
|
Goto Forum:
Current Time: Sat May 11 11:05:41 EDT 2024
|