My sword can beat up your sword

Items are fun. Grinding for incremental improvements is a great hook, but ultimately shallow and frankly somewhat misanthropic. Today I'm thinking about a non-achievement oriented item system, the heart of which is that items grow with you and are better for particular things rather than being better in an absolute, statistical sense. Is this complexity that adds depth or just tedium and learning curve?

Let's think about weapons. We start with a weapon with no stats on it, just a type. Okay, well clearly different types of weapons should be better at different things - ranged, piercing, blunt, reach, weight, magic-channeling, etc. So the underlying model allows only certain attacks and defense, and gives bonuses or penalties to some as well. Through use, perhaps these modifiers change to fit your style. And because feedback is important, we add flashy light effects to indicate when a weapon is getting "good" at a certain attack in a certain context. This also opens up a new realm of gameplay - discerning the nature of a weapon. But how easy or hard should it be? We want the casual player to be able to enjoy the game, and their weapon, without having to consult spreadsheets to see if it is the best weapon possible.

Alright then, we introduce tiers of weapons. Something like common, exceptional, magical and legendary. The higher tier weapons are not inherently more powerful, they have more potential to be more powerful, and with that they also require more effort and knowledge to use properly. By having a hidden user/item specific underlying model, we try to avoid min/maxing based on templates and force actual gameplay to determine success. Perhaps the bottom two tiers give the casual player room to grow without getting overly involved, while the top tiers allow the obsessive to squeeze out that extra little advantage. So the common sword you've been using for months is in fact superior for you to the shiny new magical one that just became available, but if you're willing to make the investment, that shiny one can become better.

Then we parameterize the weapons. We already have type, let's add in material, crafting school, runes, gems, magical embuing...All these things can serve to customize the weapon in a visible way, while still allowing the underlying model to control the specific numbers. So you can make your weapon more suited to defense, but not in terms of specific numbers.

Because two weapons can't be directly compared except by actual use, and because the value of an item is tied to the player using it, this system is as much about exploring as it is about achievement. Is that good? I think so, but at the very least it is different.

This also addresses a huge problem with crafting - can you craft better weapons than you can find? Can you fail to make the best weapon and make a mediocre one instead? These questions are less problematic if you don't live in a shallow world where items can be ranked pretty absolutely in terms of value.