FFXIV - Hiroshi Minagawa and algorithms

Who am I
Lluís Enric Mayans
Content warning:
Article evaluation:

Are you wondering about the game's algorithms? Here is a message from Hiroshi Minagawa which explains what the main programmer chose for FFXIV.

Perfect Lodestone source


This has been discussed on the Interface forum, and while I'm not supposed to deal with it, I still want to comment on it.

I consulted our main programmer "K" and it appears that in FFXIV we are using 3 famous random algorithms, depending on the application. Of course, everything takes place at the server level, and the random setting differs for each process.

So, to answer the question of whether there is a bug in the program that generates the same random number by repeating an action, it turns out that the system has been implemented so that this type of problem cannot occur. produce.

In some older games it was possible to determine the archetype and take advantage of it to complete the game, but there are no games that work that way anymore. Probably not...

However, the random processes of older games, and even up to now (not just FFXIV), have always been questioned by gamers.

From my point of view, this is just a matter of chance, but it might not be enough to convince you, so I will try to explain this to you.

This discussion is not specific to FFXIV and more generally relates to why random numbers are often questioned in programming.

(Of course, it has happened that real anomalies exist in the code ...)

To begin with, I will introduce you to the essential points.

A. The basic random numbers used in FFXIV, and other recent games, generate convenient “real random numbers”.
B. In a range of given numbers that can be understood by an individual, given that a real random number exists, this can influence us.
C. Our brains are very prone to detecting this as an anomaly, or a singularity, and that is why real random numbers are perceived as an irregularity.

I will detail each point individually.

I will first start with the program used to generate random numbers.

FFXIV, as well as other modern games, use a random number algorithm (known as pseudorandom number generator), invented by highly skilled mathematicians and computer scientists, tested in various fields of application and recognized as a benchmark algorithm.

For example, I'll tell you a little bit about the periodicity of the random number algorithm known as the Mersenne twister.

“This algorithm offers a period of 2 ^ {19937} -1; it is uniformly distributed over a large number of dimensions (623 for 32-bit numbers) ”.

You understood ? No ? No worries.

Random number algorithms are part of a dimension based on pure mathematics. As long as we, the users, understand that the numbers generated are correct and practical, that is enough.

★ For those of you who are familiar with random number algorithms, and examples of problems that have occurred in various games, you might be wondering why I haven't covered a specific point. I will add an explanation at the end of this post.

Let us now approach the case of “real random numbers”.

When it comes to random numbers, there are two distinct types, each with a different distribution: uniform random numbers and normal random numbers. By increasing the number of tests, we begin to approach ideal conditions.

Conversely, if the number of tests (in number samples) is too low, irregularities occur in the random numbers.

As humans faced with the number of tries in games etc. that we are able to assimilate, it is normal to perceive irregularities in random numbers.

In other words, the assumption that it is strange to perceive an irregularity in the random numbers is wrong.

Of course, over hundreds and thousands, if not millions of tests, we approach a uniformly random distribution, and the irregularities tend to disappear. This is unfortunately impossible for a human to observe.

It is for this reason that whenever one observes an irregularity on a limited number of tests, one cannot help but think that there is a problem.

Finally, with regard to cognitive misconceptions.

Humans have the ability to instantly determine various things by intuition. It is often said that our brains are particularly good at this, where computers struggle. On the other hand, some aspect of this ability is weak, however, and when harnessed, it opens the door to pitfalls and mistakes.

The pattern of deviation in judgment and illogical interpretation are collectively referred to as "cognitive irregularities".

A major example of cognitive irregularities is found in gamer illusion.

This theory states that if a person observes an event multiple times over a period of time, they expect future results to be based on what happened before, and will therefore be convinced that there is has a cause and effect relationship, or there is something wrong.

If you flip a coin five times in a row, the probability of tipping a coin on the sixth toss is 50%. However, we are firmly convinced that "this time it's the right one!" »And we modify the expected result. When the coin falls back on the pile, we say to ourselves "Impossible!" "While if she falls face down, one thinks" That's what I thought! ". It's cognitive irregularity.

As I mentioned in point B above, when we perceive an anomaly even if there is none, it is because our brain is trying to create a cause and effect relationship that we anticipate, while the probability works in a totally different way.

(I tell you all of this, however, I am only highlighting what I read in a book.)

Cognitive irregularity can be likened to a bug in the cognitive abilities of human beings, but it seems to be a system acquired for our survival in order to be able to make decisions and act effectively in an emergency situation.

This subject is really fascinating, and if you are interested in it, I recommend you this book.

This explains why we often create our own odds in games.

Bonus Info: An aspect that I haven't touched on before regarding random number algorithms.

Among random number algorithms, one of the methods often adopted and which is very efficient in terms of probability is the linear congruent method. On the other hand, programmers know well that problems arise with large-scale irregularities.

In fact, one game inappropriately used this feature, and as the sample size increased, so did the irregularities. This means that while this method is well known, it does not mean that it is necessarily foolproof.

In addition, even if a sequence of numbers is generated correctly, depending on how the application is used, strange series may emerge.

That being said, this post is not intended to demonstrate proof of the accuracy of the various probabilities in FFXIV. As I mentioned at the start, the second part of this long post is just gossip, but to answer the question "Is there a problem with the probabilities in XIV?" The answer is no: there is no problem.

What do you think ?

Add a comment from FFXIV - Hiroshi Minagawa and algorithms
Comment sent successfully! We will review it in the next few hours.