# Pros and cons of randomised item stats?

I'm working on an old-style isometric RPG like the early Diablo or Fallout games. It has a central story with branching paths, and a lot of player freedom to explore the open world. The combat system is pretty integral to the game as well, considering the number of spells, attacks and enemies I'm hoping to add.

Currently I'm wondering whether I should have fixed item stats, or have them randomised within a given range (say, +-0-10% from a given value). But I can only see disadvantages with the randomised stats system. If the range is too large it'll become grindy, and if it's too small it'll become irrelevant.

Could someone give me a list of the pros and cons for each approach?

The advantage of randomizing item stats is that it is a cheap way to get more content and more variety. But even if you spend a lot of time into fine-tuning your generation algorithms, procedurally generated content will never reach the quality of hand-designed content. So you are trading quality for quantity here.

When you have a large number of items with similar stats, you invite the player to compare their numbers and meta-game. The player will have to decide whether the weapon with +16 base damage, +10 accuracy or +5% attack speed is the better choice. In order to make that call, the player needs to gain a deeper understanding of your game mechanics.

The question is do you want this? There are players who like to treat games like mathematical puzzles. I can understand the appeal in this. It appeals to me, too. But there are also players who do not want to play number games. They just want to pick up the weapon which looks and feels cool, have a good time with it and get immersed in the game's narrative. Which demographic you want to focus on is your decision. It is an important decision to make, because it greatly affects your UI design.

Another aspect of randomized items is that the game experience becomes less planned. This is a double-edged sword. On the one hand, this increases replayability. The player can play the game multiple times and will have a different game experience each time because the different items they got during their run will mean that they will have to approach problems in a different manner.

But on the other hand, this uncertainty makes it harder for you as a designer to fine-tune the game experience. You can not rely on the player having or not having specific equipment at a specific stage of the game. That makes it harder for you to anticipate how difficult a combat encounters will be and how the player will approach it. The player might be lucky, find a ridiculously overpowered item and the game becomes far too easy and far too monotonous. Or the player might be unlucky, not find the stuff their character build relies on and get stuck. Both result in bad game experiences you can avoid by deciding yourself when the player receives which item. You also have problems with balancing rewards. When the player accomplishes something awesome, you will want to reward them with some loot. But when your loot is randomly-generated, then the reward for a major accomplisment might turn out to be useless junk while the reward for a minor random encounter turns out to be far too good.

You can certainly do randomised stats, as long as you make sure the weapons are still balanced.

Depending on how the rest of the game goes, think of it this way: whenever an item appears, randomise its stats. For each point of damage it loses, it gains one point in speed. This creates weapons which are random, but still balanced, and also makes the player pick a weapon based on their strategy (some like attacking speed, some damage etc).

You can also make items which are stronger (like uncommon or rare in some games) that have better stats, like for each point of damage it loses, it gains 2 points in speed.

• It takes time to create a nice balancing algorithm to apply to the weapons. In my example I assumed that 1 damage == 1 speed but that is not true for all games, it varies greatly with gameplay and other factors.