In game testing interviews, the question of how to measure probabilities sometimes arises. For example, how to measure the lottery probability of an activity? For example, after a boss kill, there is a 1% probability that the material treasure chest will drop. How to measure it? Material upgrade has a 20% probability of failure to strengthen, how to measure? Improved equipment has a 5% chance of crit. How to measure it? …

How do you test probabilities in a game?

The probability question is always more or less the same, is asking the probability test method after all. First of all, I do not have or know the correct answer, I just leave a reference, how to understand or how to organize the language, it is up to you.

In our actual work, for this kind of probability problem, in fact, do not pay much attention to, 1% probability and 2% probability what is the difference? If you have a dark face, 99% of the time you will have to draw 100 times, so I think the important thing to ask this question is thinking.

How about in practice? For example, a turntable has a 1% chance of winning a big prize. How do we test that?

First of all, to see whether the function is normal, that is to say, can draw this prize normally. You try to draw and you find out that after n draws, you do win the lottery, and then you worry about the probability.

As for probability, it’s a little metaphysical, but as long as it stays within a certain probability, it’s OK, it doesn’t need to be so precise. For example, if you draw a lottery 10 times, draw a lottery 10 times, draw a lottery 10 times, draw a lottery 10 times, then the 1% probability is definitely wrong.

After all that smoking and not being able to intuit the 1% chance?

The easiest way to do this is to look at the plan configuration sheet. Because in order to facilitate the probability function, the program will not do stupid things like writing the probability of death, and the planner will not agree, he will require the probability controllable, that is to say, the program to achieve the probability function, the planner can automatically control the probability by matching the numbers in the table.

The advantage of plotting probability is that it can adjust the drop, or output, at any time, which is what numerical plotting is all about. Low probability means high value and high benefit.

Back to the configuration table. Since it is a configuration table, it must be the number in the field that can be filled in, that is, the number between 0 and 1. (In practice, the common denominator is 10,000)

Now that we’re designing numbers, then for our test we can test by the boundary value method, one is 0, you can never draw the big prize; One is 100, so any draw is a prize. The next one is 50. The probability of winning and missing is equal. I believe that by changing the configuration probability, you can know that the probability program of this function is correct.

Now that the program is correct, all we have to do is check the schematics. You said the probability was 1%, but I saw that it was 2% clearly written in your configuration table. Although the difference of 1% May not be obvious, the evidence left by your configuration shows that the probability does not meet the demand, that is to say, the plan is fooling me.

Well, here’s my answer.