Random engine with or without memory ?

[Jennson]

Well in different games random mechanisms can be implemented with memory or without. The most classic example for random-machines without memory is roulette. Every single turn is a single event with the same chance to occur regardless the prevoius results.

In opposite a Cardstack is a random system with memory, because after you draw 10 red cards the chance to draw another red decreases because less red cards are left in the stack.

Both systems can be implemented easily within the loot engine without much processing workload.

What are arguments within entropia for(or against ) both different engines ?
What data would need to be collected to determine the nature of the random machine ?

 

[MegaDeth]

Well in different games random mechanisms can be implemented with memory or without. The most classic example for random-machines without memory is roulette. Every single turn is a single event with the same chance to occur regardless the prevoius results.

In opposite a Cardstack is a random system with memory, because after you draw 10 red cards the chance to draw another red decreases because less red cards are left in the stack.

Both systems can be implemented easily within the loot engine without much processing workload.

What are arguments within entropia for(or against ) both different engines ?
What data would need to be collected to determine the nature of the random machine ?

~90% tt return can not be achieved without a memory of what has been spent and looted.

i don't consider this a theory; i can see multiple ways in how a non memory system would crumble here

the problem is the volume of spending, the more you spend in a short amount of time ... the lower ones short term tt returns can get and the higher ones hofs can get, this can NOT be achieved without memory of prior dice rolls

that's why peoples tt returns over arbitrary periods of time don't match

 

[trance]

a random system only, give more luck as we have in eu (just try out random generators in your code) .... in eu we call this dynamic and it is far away of random!

 

[ermik]

its not random , jesus how much proof do people need?

its always easier to blame the system than to change our own behaviour.

system is just fine, but hard to profit from just like its supposed to be.

 

[valentin]

its not random , jesus how much proof do people need?

its always easier to blame the system than to change our own behaviour.

system is just fine, but hard to profit from just like its supposed to be.

If it not random who can huge active ppl who depo a lot lose mega and ppl who has just started here get big ATH.

I belive in this text.


I think about it as such, I burn ammo to hunt, I burn bombs / probes there will all players get 90% back if they play on the right level for theres avatar. If you playing over your level or under your level you will get less then 90% and this is no lottery and have all to do with yours skills and gear. It also includes the normal items in loot, and if you want to gain over the 90% its about the markup of your loot.


Then we have normal globals and hofs, there are here to create addictive existence in the game and to prevent the game becoming boring and monotonous.


Now we coming to the Big or "Uber" ATH:s and "uber" items. This phase has nothing to do with your skills,gear or your dedication or activity in the game and as Starfinder who has the Entropia Tracker proved more noobs then ubers gets it.

So how do we get this ATH:s and ubergear--> the answer is simple ITS pure lottery or if you can say MA puts it there on purpose to some avatars who many in this game belive.

And who pays for this ATH:s and ubergears... We all do of the part who is under the 90%


P.M. And you think MA gave a lot fore free whit tis new hangars--no they did not the loot will go down a bit and we will all pay for this new hangars.

 

[Mikeemooo]

I think: individual mob based memory, random based on a variables for that mob (and the memory of the kill).

I don't think there's any memory longer than the length of a kill.

Or to explain in dodgy make-believe code:

mob = new Mob();
mob->addDamage(player1, 30, damageCost);
mob->addDamage(player2, 30, damageCost);
mob->addDamage(player1, 30, damageCost);
mob->onDeath((){
  mob->payoutToHighestDamagePlayer();
});

I don't believe there's a global system that says that Mr X has lost 100000 PED over the years, lets give him an ATH. I don't believe there's a system that says that Mr Y has a higher loot factor than Mr Z on any particular mob.

But hey, I may be wrong eh

 

[aridash]

~90% tt return can not be achieved without a memory of what has been spent and looted.

this simply isnt true. flip a coin, even odds, it might be 40% heads one 100 run, 55% heads over 200, but take a long enough period you will end up at near 50%. does your coin have a memory? go find a 10 sided dice, roll it 1000 times and tell me what % of rolls were not the number 6. i'd wager a ped its ~90%.

its not random , jesus how much proof do people need?

what is not random? such a statement needs qualification. individual loot events? per 100 mobs? hour, day, week, year periods? per 1k, 10, 100K ped turnover? the incidents of noob hofs and ATH kinda suggest that individual loot events at least are random. OTH loot contents clearly isnt random, there are certain items from certain mobs. loot amounts seem tied to HP, but values vary wildly, is it random within specified bounds?

 

[trance]

it is more like this:
variables =[
Player_skills
player_gear
player_ammo_used
player_biorythm
player_total_spend
lootpool_total
per_mob_used_ammo
Items_in_drop_list
some_more_variables]

And then find a formula with some random number and use this variables to calculate the loot

just my test with some lootsystem i use for my own game.

If you only use the random number, the avatar will have too much luck. It needs a lot other variables, to balance a lootsystem.

 

[Jennson]

~90% tt return can not be achieved without a memory of what has been spent and looted.

Well basically a system can. http://de.wikipedia.org/w/index.php...arge-numbers.png&filetimestamp=20090901233851

The question is which system is in place. My personal belief is there is some kind of memorizing, but i don't have any proof. E.g Ace 10% TT return curve looks more similar like the one i linked, so it could be also a non-memorizing system in place.

 

[Jennson]

a random system only, give more luck as we have in eu (just try out random generators in your code) .... in eu we call this dynamic and it is far away of random!

Actually i coded simplified simulation of those two systems and both explain patterns that can be seen in EU, but just because i can verify similarities in pattern doesn't give any proof, so i'm looking for arguments for both to determine a method to prove or disprove one of the systems

 

[Jennson]

its not random , jesus how much proof do people need?

its always easier to blame the system than to change our own behaviour.

system is just fine, but hard to profit from just like its supposed to be.

Calm down, the question isn't about good or bad returns. And nnone will deny there is a random component. Both can be balanced for steady returns. And i'm not complaining abbout good or bad returns nor i'm saying it's good or bad, i just try to determine the exact nature of the random component of the system

 

[Jennson]

it is more like this:
variables =[
Player_skills
player_gear
player_ammo_used
player_biorythm
player_total_spend
lootpool_total
per_mob_used_ammo
Items_in_drop_list
some_more_variables]

And then find a formula with some random number and use this variables to calculate the loot

just my test with some lootsystem i use for my own game.

If you only use the random number, the avatar will have too much luck. It needs a lot other variables, to balance a lootsystem.

A start could be to ignore items and market values and first boil it down to cost in -> cost out and the question wether player based vars ( skills /biorythm / whatever coefficient) create different results. When we see logs from players on high skill levels we see it seems to make no difference in the sample size ( beside using maxed gear, Defense cost etc, which can be explained by using the right gear ).

One thing i can see is good and bad phases for avatars come in waves. For me it seems to be an argument for the memorized system.

Maybe our math Gurus can create models which help us to determine which system is more probable with the results we see.

 

[Xavier Target]

I would think you would have some kind of memory for item drops. Many have shown they can achieve a fairly reliable % return TT. That is pure random, since that is much like the coin flip. Flip it enough times you all get the same % result. The type of items in that loot seems to vary more. MA could adjust rate at will or have a memory mechanism to control Item drops without affecting their take. Seems to me that is the purpose of a balancing manager.

 

[The Goat]

I personally believe that the system has a memory. In my many many many many hunts cycling un-godly amounts of ammo I have found cycles. These cycles suggest memory. When you think of it from the cycling point of veiw you are almost forced to believe there is memory in the system. I also believe that said memory is based on the HP/lvl of the mob being hunted.

Will you notice the cycles on daikibas? probably not...
Will you notice the cycles on Leviathans? Yes (and wow is it a long cyce)

That is pure random, since that is much like the coin flip. Flip it enough times you all get the same % result.

Not with an American Quarter - because heads weighs more

 

[Leona]

I think: individual mob based memory, random based on a variables for that mob (and the memory of the kill).

I don't think there's any memory longer than the length of a kill.

What makes you think that loot is "in the mob"* in the first place?

*Let me add "still / anymore" for the old timers.

 

[Mikeemooo]

What makes you think that loot is "in the mob"* in the first place?

*Let me add "anymore" for the old timers.

I don't.

 

[Leona]

Good...

 

[remontoire]

One thing i can see is good and bad phases for avatars come in waves. For me it seems to be an argument for the memorized system.

Maybe our math Gurus can create models which help us to determine which system is more probable with the results we see.

There are very few things that require memory. Having a function based on the time (since some epoch) and the avatar ID be a term used in the loot calculation takes care of the waviness. The same takes care of any biorythm like behaviour. You do not need memory for this when you have time available.

 

[Dahlifyr]

It`s trigged and it`s rigged, nothing more to talk about here realy....The whole game itself is a punch in the face to serious gamedevelopers!

 

[Mikeemooo]

http://jsfiddle.net/k3y6N/15/

 

[Jennson]

http://jsfiddle.net/k3y6N/15/

Great simulation, now add some sample mobs, let the players use different equipment and create an ATH list and checkout how the list develops when 100 player compete against each other a few hours My PC is already puking running 50 players

 

[Mikeemooo]

Great simulation, now add some sample mobs, let the players use different equipment and create an ATH list and checkout how the list develops when 100 player compete against each other a few hours My PC is already puking running 50 players

Hehe, then some 3d graphics, and a depositing method? ;D

 

[Jennson]

Hehe, then some 3d graphics, and a depositing method? ;D

Only problem is, simulated players may have very limited funds

 

[Jennson]

--- ATH list --

The main reason was because my belief is the trend to higher ATHs isn'T directly managed manually but also part of the Algorithm to allow the ATH list to "grow" (valuewise ) , otherwise the chance to get a "dry" list which isn't updated anymore and thus lower player motivation due to lower ATH chances. That might be the reason why ATHs in the beginning were really small and raised to 6 digits during last years

 

[Dr.D.C.]

it s not random. es.

you buy one or more lands for many thousand dollars
the day after you go to mine and you get the biggest ATH in this profession

do you want name this a random system???

 

[JohnCapital]

Great simulation, now add some sample mobs, let the players use different equipment and create an ATH list and checkout how the list develops when 100 player compete against each other a few hours My PC is already puking running 50 players

Great simulation, now add some sample mobs, let the players use different equipment and create an ATH list and checkout how the list develops when 100 player compete against each other a few hours My PC is already puking running 50 players

Actually, if you want to get a better view:

• Add multiple mobs of different health
• Add different weapons w/ different costs
• Add folks using multiple weapons on the same mob
• Add team hunting
• Make multiple players spend different amounts (some 2 hours/week, other 10 hours/day)

Now apply this same "theory" to mining and crafting. (don't forget varying degrees of amps, and qty/condition clicking) For example:

I think: individual mob based memory, random based on a variables for that mob (and the memory of the kill).

Do you also believe there's an individual BP memory? (Look back at Starman's ATH) What about (L)(L) BPs? Does mining hold "location" memory, based on each spawn area?

I've said it before, but:

ANY loot theory has to be able to cover at least all three main activities.


Not saying anyone's right or wrong. Simply giving the variables that most folks forget. Most loot theories are usually "designed" only for covering the person's playing style and forgets the massive variety of different playing styles.

 

[DoubleWolf]

There's a test I've always wanted to do (or at least see the results of) that in theory could answer this question.

Take two miners. One miner (miner A) will be the control, the other (miner B) the guinea pig, if you will. Both miners will check for the exact same type of resources, using the exact same equipment for a minimum of 1000 drops.

Have miner A lead the way and drop where ever they wish. After the drop and the results of the drop are determined, have miner B come in and drop on the exact same spot. As anyone who has tried mining in an area that has recently been covered can probably tell you, you don't get a lot of finds combing through that area. Your returns suffer. It's not until you get out of that area that you start to get claims again.

In theory, miner A should have a much better return because they found the resource first, if there is one. Miner B will inevitably find a few resources just from what would otherwise be a double bomb, but for the most part should have poor returns.

There are two things we are looking for from this test. The first is while doing the test. Are Miner B's finds that he does get larger than average to make up for the losses? The second is what happens to Miner B's finds AFTER doing this test and setting off again on his own.

For the results of this to work out in favor of the "with memory" theory, then unsuccessful finds dropped in areas just covered by another person will eventually count toward a larger find down the road. It would seem rather unfair to code the game differently as it's possible that a player could always just happen to mine right after someone else and always have poor returns. However if the system is 100% random, then that scenario is entirely possible.

 

[Jennson]

it s not random. es.

you buy one or more lands for many thousand dollars
the day after you go to mine and you get the biggest ATH in this profession

do you want name this a random system???

No not the system, but the looting action. This is a HUGE difference. But Squall also turned over Gazillions of Peds in his mining Career which could also indicate a memory based system. And still even in such a system the single looting action is still random.

 

[Mikeemooo]

Actually, if you want to get a better view:

• Add multiple mobs of different health
• Add different weapons w/ different costs
• Add folks using multiple weapons on the same mob

http://jsfiddle.net/k3y6N/18/

Multiple weapon costs with multiple damage ranges and multiple weapons per person. The system handles that by design.

• Add team hunting

I can't be bothered to simulate this up because it'd need a bit of a rewrite to the structure, but the algorithm would be just the same, the only difference being that it'd store the percentage of damage done by each player then split the loot based on whatever team rule.

• Make multiple players spend different amounts (some 2 hours/week, other 10 hours/day)

Makes no difference, it'll still even out to the same amount no matter how many cycles you run for each person..

Now apply this same "theory" to mining and crafting. (don't forget varying degrees of amps, and qty/condition clicking) For example:

Do you also believe there's an individual BP memory? (Look back at Starman's ATH) What about (L)(L) BPs? Does mining hold "location" memory, based on each spawn area?

I've said it before, but:

ANY loot theory has to be able to cover at least all three main activities.


Not saying anyone's right or wrong. Simply giving the variables that most folks forget. Most loot theories are usually "designed" only for covering the person's playing style and forgets the massive variety of different playing styles.

I haven't looked into the crafting or mining systems, so wouldn't like to attempt something like that, but I don't really see why they need to run from the same algorithms. They're entirely different systems. So long as they have an average payout of a % over time... Just like a casino has many games, but they all have different logic and a set payout rate.

I'm certainly not saying that my system is HOW IT WORKS. Just trying to show why I don't think there needs to be any kind of 'memory' in the system when it can all be done just with basic maths (and in this case: shitty maths), even in shitty code in a shitty browser language.

 

[Jennson]

http://jsfiddle.net/k3y6N/18/

Multiple weapon costs with multiple damage ranges and multiple weapons per person. The system handles that by design.



I can't be bothered to simulate this up because it'd need a bit of a rewrite to the structure, but the algorithm would be just the same, the only difference being that it'd store the percentage of damage done by each player then split the loot based on whatever team rule.


Makes no difference, it'll still even out to the same amount no matter how many cycles you run for each person..



I haven't looked into the crafting or mining systems, so wouldn't like to attempt something like that, but I don't really see why they need to run from the same algorithms. They're entirely different systems. So long as they have an average payout of a % over time... Just like a casino has many games, but they all have different logic and a set payout rate.

I'm certainly not saying that my system is HOW IT WORKS. Just trying to show why I don't think there needs to be any kind of 'memory' in the system when it can all be done just with basic maths (and in this case: shitty maths), even in shitty code in a shitty browser language.

did you put math.random() on the cost/shot parameter ?