Matrix Games Forums

ORIGINAL: Ron Saueracker

I wannit, I wannit, I wannit!!!!!!!!!

Better check with Termmy. He has the "cant have it" list [:D]

ORIGINAL: witpqs

ORIGINAL: pad152

Random events have a funny habit of not being all that random in the long term.

Could you explain that to those of us who are not superstitious?

Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

In WITP/AE something random may look or work great for 2-day turns but see more repeats on 1-day turns, WITP/AE is a long long game, something that looks random the first 100 turns may repeat more then you like in 600 or 900 turns. Imagine having two or three of your carriers blow up sitting in port doing nothing in a single game. That could be a real game killer, why risk it?

I can lose my own ships, I don't need a computer random number generator sinking them for me!

ORIGINAL: Ron Saueracker


Cool, so if a fully loaded AE like USS Pyro is hit by a torp, there is a greater chance of a grave explosion than say an empty Liberty ship. Bang on guys![:)]

Always has been that way, back to stock WITP 1.0. Ammo ships and tankers with volatile cargo can suffer cargo explosions.

In AE, you can have an accident while refueling/rearming from an ammo ship or tanker that can damage both the ammo ship/tanker and the ship being refueled/rearmed.

Incidents during loading and unloading can damage both the ship involved and the port facilities.

Incidents during repair can damage both the ship involved and the repair facility (port, tender, repair ship/floating drydock).

ORIGINAL: Yamato hugger

ORIGINAL: Ron Saueracker

I wannit, I wannit, I wannit!!!!!!!!!

Better check with Termmy. He has the "cant have it" list [:D]

Termmy has a list, now that is scary, someone better call Homeland Security!

ORIGINAL: pad152
Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

Getting repeats of numbers should happen when things are truely random (not sure exactly what you mean in the first sentence, sorry if I've misinterpeted you). The second is false, I don't know of any programming language that represents time using AM/PM.

ORIGINAL: pad152

Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

In WITP/AE something random may look or work great for 2-day turns but see more repeats on 1-day turns, WITP/AE is a long long game, something that looks random the first 100 turns may repeat more then you like in 600 or 900 turns. Imagine having two or three of your carriers blow up sitting in port doing nothing in a single game. That could be a real game killer, why risk it?

I can lose my own ships, I don't need a computer random number generator sinking them for me!

I figure if this one random routine is bad then all of the hundreds of other random routines in the game must be bad too.

Given what recently happened to the USS Port Royal I think this is a great and realistic improvement.!!! Keep at it guys, I can hardly wait.
[:'(]

ORIGINAL: witpqs

ORIGINAL: pad152

Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

In WITP/AE something random may look or work great for 2-day turns but see more repeats on 1-day turns, WITP/AE is a long long game, something that looks random the first 100 turns may repeat more then you like in 600 or 900 turns. Imagine having two or three of your carriers blow up sitting in port doing nothing in a single game. That could be a real game killer, why risk it?

I can lose my own ships, I don't need a computer random number generator sinking them for me!

I figure if this one random routine is bad then all of the hundreds of other random routines in the game must be bad too.

Not much to back that up (unless you're being sarcastic)...

ORIGINAL: pad152

ORIGINAL: witpqs

ORIGINAL: pad152

Random events have a funny habit of not being all that random in the long term.

Could you explain that to those of us who are not superstitious?

Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

In WITP/AE something random may look or work great for 2-day turns but see more repeats on 1-day turns, WITP/AE is a long long game, something that looks random the first 100 turns may repeat more then you like in 600 or 900 turns. Imagine having two or three of your carriers blow up sitting in port doing nothing in a single game. That could be a real game killer, why risk it?

I can lose my own ships, I don't need a computer random number generator sinking them for me!

What I remember is that most Pseudo Random Number Generators use 32 bit integer arithmetic and ignoring overflow. the starting seed is usually the 32 number of seconds since midnight variable available from most operating systems. I have not programmed for a long time, but I have written a number of PRNGs in many different languages. It is not that hard to write a good one.

ORIGINAL: Terminus

ORIGINAL: witpqs

ORIGINAL: pad152

Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

In WITP/AE something random may look or work great for 2-day turns but see more repeats on 1-day turns, WITP/AE is a long long game, something that looks random the first 100 turns may repeat more then you like in 600 or 900 turns. Imagine having two or three of your carriers blow up sitting in port doing nothing in a single game. That could be a real game killer, why risk it?

I can lose my own ships, I don't need a computer random number generator sinking them for me!

I figure if this one random routine is bad then all of the hundreds of other random routines in the game must be bad too.

Not much to back that up (unless you're being sarcastic)...

Not sarcasm exactly. Just pointing out that if you really have suspicions about that random function then there are all the same reasons to be suspicious of them all. Which I am not.

ORIGINAL: tombom

ORIGINAL: pad152
Well it's hard to produce a good random number generator, even then you get repeats, like the number 8 repeating in a lotto game for 3 or 4 straight games. Most random number functions in programing languages use date/time as a key, in which 10AM and 10PM on the same day can produce the same random numbers.

Getting repeats of numbers should happen when things are truely random (not sure exactly what you mean in the first sentence, sorry if I've misinterpeted you). The second is false, I don't know of any programming language that represents time using AM/PM.

Getting repeats of numbers should happen when things are truely random (not sure exactly what you mean in the first sentence, sorry if I've misinterpeted you). The second is false, I don't know of any programming language that represents time using AM/PM.

If you are trying to produce rare events (repeats) may not be a good thing! Nobody said any programming language uses AM or PM time, I said a random number function using time/date as a key can produce the same results at 10AM and 10PM on the same day.

We've all seen funny things happen in WITP, like when badly damaged ship(s) get safely to port, the damage goes down for a few turns, it looks like the ship can be saved but, then it starts going up, and the ship sinks. I have no problem with that, the ship was damaged in combat and couldn't be saved even in port. I've also seen moral for air units jump up and down like a yo-yo for no reason, some random event?

I do have a problem with a ship(s) with no combat damage, randomly blowing up in port! WITP/AE is a long game and something like random ships blowing up can ruin a game if it happens one too many times and nobody knows how many, one to many is, until it happens! I don't want to find out 2 or 3 years from now on turn 914. Anyone who thinks this is a good idea will be the first to scream when it happens to them.

If anyone remember North Atlantic 86, you would get a damaged ship all the way back to the east coast of the US only to have it sink one hex from port (another random event) that wasn't so random (sinking 50% of all damaged ships) or Bombing of the Reich where after conducting a successful bombing run, you would get the endless "Bomber crashed on Landing", another random event that caused a third of all bombers losses.

ORIGINAL: witpqs

ORIGINAL: Terminus

ORIGINAL: witpqs



I figure if this one random routine is bad then all of the hundreds of other random routines in the game must be bad too.

Not much to back that up (unless you're being sarcastic)...

Not sarcasm exactly. Just pointing out that if you really have suspicions about that random function then there are all the same reasons to be suspicious of them all. Which I am not.

Good.[:)]

I even get more suspicious if Terminus starts to agree with me![;)]

ORIGINAL: Ron Saueracker

I wannit, I wannit, I wannit!!!!!!!!!

You do realize that your copy of AE will be delayed an extra week or two. After all, they have to integrate the "make sure Ron gets it in the rear non-random number generator" from WITP into your version.

[X(]That (collided at sea)must have been very impressive Terminus. I can't recall seeing any of your verbiage go beyond 1 sentence before. Usually just a syllable and/or a smiley!

[:D]

ORIGINAL: rhohltjr

[X(]That (collided at sea)must have been very impressive Terminus. I can't recall seeing any of your verbiage go beyond 1 sentence before. Usually just a syllable and/or a smiley!

[:D]

Or a grunt [;)]

(I THINK it was a grunt - may have been further south [:D])

worrying about the non-random nature of pseudo random number generator or series will only be relevant if the implementation of the random events is very odd. for example, if the random event were set up like this:

Generate random number between 1 and 1000000

result: 1= CV Lexington has magazine explosion
2= CL Lex has fuel explosion
3 = CL lex has some other sort of damage
4 CV Enterprise has magazine explosion

etc, in which case you would be very vulnerable if a specific number (in this case say '1' or '4') comes up too often.

In reality I suspect it is a series of random numbers are used, in which case it would be something like:

Generate a random number between 1 and 10000. If 1 is rolled, then a random event occurs.
generate a second random number between 1 and number of eligable ships to select the specific ship
generate a third random number between 1 and x to classify the event...

This is still vulnerable to non-random behaviour - the most predictable being the chance of a random event occurring in the first place being higher (or lower) than the pure expectation (in my example 1 in 10000 per 'test'). However the chance of bias towards the same events repeating is vanishingly small. The key to masking these effects completely is use the best possible random number generator possble (and avoid obvious repeat seeds), Also, try to avoid the generating too low odds rolls - the 1 in a million random necessary in my first example is far more vulnerable to patterns repeating specific events being obvious to the user than a sequence that rolls 1 in 10000 many times - especially if other random numbers need to be generated for unrelated needs 'during' the sequence (or in my case, the number of elegable ships changes)

As long as they use a good RN generator (not the old MS Excel one which was anything but random!) all will be fine.

Of course, some people reload the game whenever a combat result comes out against them...

non-random pseudo-random-number generators are a serious problem. (I had to address this in my PhD work, and I ended up working in this area for the Boost libraries.) Go here.

ORIGINAL: pad152

In WITP/AE something random may look or work great for 2-day turns but see more repeats on 1-day turns, WITP/AE is a long long game, something that looks random the first 100 turns may repeat more then you like in 600 or 900 turns. Imagine having two or three of your carriers blow up sitting in port doing nothing in a single game. That could be a real game killer, why risk it?

I can lose my own ships, I don't need a computer random number generator sinking them for me!

I agree that a bad random number genrator could wreak havoc, but would suggest you consider two things:

1) AE has been playtested extensively. We outsiders don't know the full number of turns for human versus human, human vs. computer, and computer vs. computer, but I'd guess it's in the 1000's. And what we hear - anecdotally - is that random ship accidents are very rare. If the random number generator was causing non-historic levels of accidents, the team would have noticed and fixed it already.

2) A significant percentage of the mathematical calculations in AE involve randomness. Search results, FOW, CAP interception chances, Naval interceptions, combat results, etc, etc. - the list could go on for ten pages and still be incomplete. The point being that accidents are far-and-away one of the least game altering uses of randomization. If the random number generator was hosed, it would show it's head in a thousand places, not just this one.

I'm willing to bet the team has a good handle on this.

Yes, there are a lot of random things in WITP but, randomly losing ships to non-combat events isn't one of them.

If the developers want to put in a real weather system with ships sailing into them getting sunk and combat ops get canceled that's fine but, I don't want to lose ships to a non-existent weather system, sunk by a random number!

Unless someone can answer the questions, how many times can this happen during a campaign? or how many times is one too many? This item shouldn't make the cut.