I realise this is debatable and so I'm not going to push it hard, but I do quite strongly feel that the modification of the PERF metric to use the square root of weight instead of just the weight in the denominator only serves to introduce an extremely subjective 'balancing' of the pros and cons of armours. If that was all that it amounted to I'd be fine with that, but in my opinion it also greatly detracts from the usefulness of this page to get rid of the original, linear PERF, which was actually extremely useful in building armour sets in a fairly concrete and objective way.
Say I have a target of getting 150 AC on a particular region (this is a plausible objective for a person to have: AC is directly proportional to damage reduction under the condition that Damage < AC, and many NPCs do less than 150 damage base rounds).
What is the most 'efficient' armour set I can use to meet this objective? Well it's the armour set that gives me a minimum of that AC, but weighs the least of any other armour combination that gives likewise.
How can I calculate that? I can add together all the ACs of an armour set, and the weight, and swap in and out armours that meet reqs of >150AC, with lower and lower weights. This is robust, but fairly tedious.
Under the old system however, there was a really easy shortcut:
1. Pick the region you need covered
2. Pick the item with the highest linear PERF in each layer that covers that region
3. Check: Does the AC of this collection meet your AC requirement?
3A. Yes: You have the most 'efficient' armour set that meets your AC requirement for that region
3B. No: Swap out the item with the LOWEST linear PERF in your collection, for the item with the NEXT LOWEST Linear PERF in that item's layer. Return to step 3.
the new, modified PERF makes this method impossible.
I do realise that the modified PERF *might* be more useful for people who can't be bothered with going through that process, and that could be legit a good reason to use it. However I have no idea how often its used in that way, and if it *is*, how much utility it adds in terms of decision making. For me certainly, it's much less use.
As I say, not too interested in the debate, just wanted to put forward this opinion and leave it out.
Nevertheless happy to clear anything up if I've explained badly.
- Oh yeah there could totally be another perf or if there's something even more appropriate to what you/other people can use. I'm not quite seeing how linear perf (so just divided by weight?) does all that.
- What I want to ask is why not just use the actual AC? AC average? By multiplying it by the chance to hit don't you also muck things up? Maybe AC average / weight would be better ?
- But yeah if the Best Thing Ever (tm) is linear perf I'll just add that like Lperf before or after perf. I think without the to hit it might work better for what you describe?
- How about this: Lperf = average of 3 vurderes / weight
- Also I'll re-order the coverage zones, it keeps confusing me even after all these years. Short version for zones was also not thought out so well.
- I'm thinking it could be in that order(abbreviation): Hair(Hr) Head(Hd) Eyes(Ey) Face(Fa) Neck(Nk) Hands(Ha) Arms(Am) Back(Bk) Chest(Ch) Abdomen(Ab) Legs(Le) Feet(Ft)
- Trying to keep together zones that are in the same item but still keep a top-down order.
- --Frazyl (talk) 12:19, 6 September 2020 (EDT)
- Fiddling with it some more I can see how not multiplying it by to hit doesn't get you the most effective item if you sort by zone rather than by layer (for npcs the same size that don't focus on anything, otherwise to hit % is wrong).
- So thanks, now people can experiment with both and use what they like.
- The values without to hit were interesting for the narrow case of not caring about coverage, but not sure that it's worth adding it like the average of vurderes. If you want to go by vurdere then you'll probably pick the one you need.
- --Frazyl (talk) 14:07, 6 September 2020 (EDT)
- Ahh PERFect :p the current lambda perf is the ideal thing for what I was using before, and having them alongside with a description of what they are makes sense to me. Perfectly fine to have them side by side because I say the sqrt perf could indeed be useful as a more 'finger in the air' measurement for people who aren't quite as obsessively nerdy as me.