RNG

• I have done a some research into the subject of RNG (Random Number Generators) and have discovered that there are several types.
There are differences in the bias that can effect outcomes based on probability of randomness. Numbeer custers of the 0 vs 4 for example. There are few references of good generators I have found that apply to games, most are labled as 'good enough for who it's for' in the reaserch articals.
I am curious, as I read on these issues, if the bias can be relative to a player themself, as some see times where their "luck" based on the RNG outcome differs from different games they enter where there is significantly different rankings of the players.
Can the players themselves create an annomoly that causes bias to a particullar ranked set within the game?
A TRNG is expensive to operarte and maintain, and as the newest are relaint upon quantum technology, are ever adapting as studies into their function continues.
I work with the assumption therefore, working form the information I have read, that here the RNG used here is subject to bias and the outcome in never a true random event.
There is some degree of commfort offered in a simple talent based game where the subjectivity of the outcome is based on simple math and less on the randomness of a potentially biased generator.
I use the example of chess. The pieces, moves options and placement restictions of the game iteslf create an evironment where suprise based on random interjection of a variable outside the players control is negated. You do not base the winner on a dice roll. It is based on ability.
Here, the variability is outside the control of the player. The dice rolls at every interaction between the pieces placed on the board.
There is, therefore, no true reliance upon a players ability to perform in a tactical application, as the randomness and potential for bias within any engagement can yeild unanticipated results, rendering the statistical information relative to HP, morale and stack unit composition totally outside the realm of any certainty.
The actual ability of the player is degraded to an serious extent as engagements occur, as RNG, not TRNG, can trend to a repeating bias based on the application of the player as used. Clusters for data.
To illistrate, if I am constantly fighting, the RNG in use far more involved regularly in the ourcome, and a disrtribution results in a bell shaped curve, but over time. Within that time, it can cause wild actions that are seriously influential to the outcome of the result. And it is based on the roll of a dice. It is there fore INFLUENCED by player ability, but not CONTROLLED by it.
Intuatively, this seems to be a reverse of what the expectations are for a normal game experience.
I want results to be in line with what I can do based on unit compostiion, use, power and tactical application.
Not on the roll of a dice.

A half truth is merely a half lie.

Edited 3 times, last by South Paw: Typo ().

• My observations based on collecting hundreds of data points from particular battles is that, at least for the variance applied to damage, the distributions do not significantly differ from the expected distributions of a normal (bell curve) distribution. That's not to say that there aren't bugs in certain situations, but in general, the randomness seems to be as expected. It's hard to see this without recording many data points.

• This lends itself to the question: Does the curve developement rely specificly to a particullat point in time and begin data aquasition (A battle) in an attempt to estable the curve, or is it accumulative for a game as it advances, or part of a series of continuous calculations from multiple data contributors?
In laymans terms, do they leave the engine running for additional trips when you get out of the car, or do you shut it off and restart every time.
If specific to a time and specific unit, the RNG never developes the curve. It is not balance in it's application therefore and can be biased as a result. If clustering is coumpounding this as it applys to every battle due to it not having enough data points to establish a curve during the first calculation (Repete number simulation habitual clusters). Bias again can be encountered, if it uses these data points and continues the data collection, using each contributor within the population in order to establish a Universe. Establishment of the universe is nessessary to claim the 6 sigma rule is followed. Otherwise it is not a true reflection of the development. It would be inconsistant size sampling of subgroups. This again can artificially bias data points if clusters are present.
The very reason that 7 does not pay, or pay well in craps. It has a higher probability of appearance.
What we see is habitual. We assign variation in the output as "luck", when in actuallity, it may be a bias that a player encounters. Depending on how the curve is developed, the distribution end result then can skew, adding credability to the appearance of a series of battles, or a string of games we attribute to this "bad luck".

A half truth is merely a half lie.