sk starting points

[DeathsSilkyMist]

Originally Posted by bcbrown [You must be logged in to view images. Log in or Register.]

The game content is too easy; that's why we all agree that starting point allocations have minimal impact. There's things like solo artist challenges or low-man raids, but that's only relevant within those subsets of the larger game content.

This is correct. That is why it is best to put your starting stats into the statistically best option, because it isn't going to matter much anyway. If you agree that we can statistically discover which champions are better or worse in LoL, you will agree the same can be done in EQ. There is nothing special about EQ's code and equations that make it impossible to apply statistics to things like starting stats.

I think the problem is people have an inflated concept of what "best" means. I think they assume it means "so much better you will regret everything else". I am not saying that. INT is simply the best choice in a set of options that are ultimately not going to impact your character in a significant way. This is backed up by statistical evidence.

There is nothing wrong with saying:

1. INT is the best choice statistically.
2. You can pick any other stat, and you will not have noticeable issues with your character.
3. Other stats have benefits, which are as follows...

The best solution is to give people the facts, and trust them to make the right decisions. I see no benefit in basically saying "everything is subjective, therefore it doesn't matter". Not only is this a false statement, but it hurts the players who enjoy min/maxing. Players who do not enjoy min/maxing will not be hurt by the facts. It is a win/win to supply the best information possible.

[bcbrown]

Originally Posted by DeathsSilkyMist [You must be logged in to view images. Log in or Register.]

There is nothing special about EQ's code and equations that make it impossible to apply statistics to things like starting stats.

What's your cost function?

[DeathsSilkyMist]

Originally Posted by bcbrown [You must be logged in to view images. Log in or Register.]

What's your cost function?

I have mentioned it many times.

If we agree that starting stats will not affect your character in a meaningful way, then the benefits of each individual stat are deemed to be outside the scope of our consideration for this conversation. It doesn't really matter if you get +1 DPS, +100 HP, or +200 MP.

Therefore, the only real consideration left is the statistical probability that you will cap the stat. This is because a capped stat provides zero value.

If you take a look at the distribution of which stats appear on useful SK equipment, you will see STA and STR appear with higher frequency. INT appears less often. In addition to this, the amount of STR that can be applied to a character via buffs is also much greater than INT.

Therefore, INT is the statistically best choice.

[Troxx]

Originally Posted by bcbrown [You must be logged in to view images. Log in or Register.]

I'm aware you have some sort of professional experience in the games industry, and I'm sure you have far more experience than I do with things like adjusting game mechanics to guide the meta towards whatever you're attempting to accomplish.

But this concept you're talking about is well within my bailiwick. I've built epsilon-greedy multi-armed bandits for online commerce. I've worked on systems that attempt to correct for the "banana problem" in collaborative recommendation systems. I've worked through the mathematics of recursive descent, and understand mechanics of and motivation for using something like simulated annealing to avoid local optima. So yes, I'm explicitly arguing from authority here.

The difference between what you're talking about with LoL and what we're talking about with EQ SK attribute starting points is that the cost function is well defined in your LoL example, but there is no well defined cost function here, which is exactly what I mean when I say "best is subjective".

Min-maxing or formal optimization is about using a cost function across some domain of parameters, where you try to choose parameters that minimize the cost function. With LoL, that cost function would be loss percentage, modified by something like an Elo rating system to account for opponent player strength. The parameters would be something like team champion choices, opponent champion choices, map; not really sure because I've never played LoL.

What's the cost function in EQ? I genuinely can't think of any formal cost functions that could apply. The game content is too easy; that's why we all agree that starting point allocations have minimal impact. There's things like solo artist challenges or low-man raids, but that's only relevant within those subsets of the larger game content.

So when I say that the best is subjective, what I mean is that there's no universally applicable cost function. I mostly enjoy leveling minimally-twinked characters through pickup groups, so my cost function is going to overweight attributes like undergeared DPS, carrying capacity, and so on. You seem to mostly enjoy end-game soloing and maybe raiding, so you're going to underweight those attributes.[You must be logged in to view images. Log in or Register.]

If we don't share a cost function, we won't agree on "best".

[You must be logged in to view images. Log in or Register.]


I’m glad you said all that. I’m just sad I had such a hard time understanding it. Lots of big words and unfamiliar terminology I had to google.

#headache

[You must be logged in to view images. Log in or Register.]

[bcbrown]

Originally Posted by Troxx [You must be logged in to view images. Log in or Register.]

I’m glad you said all that. I’m just sad I had such a hard time understanding it. Lots of big words and unfamiliar terminology I had to google.

Optimization needs:

1. Some outcome you want to minimize (or maximize), and a way to measure it
2. A set of parameters to fiddle with

Multi-armed bandits: Imagine you're at a casino with a row of slot machines, which are sometimes called "one-armed bandits". Each slot machine has a different unknown payout rate. How should you choose which machine to play (exploit), and when should you choose to stick with the current machine versus move to another machine (explore)? Epsilon-greedy is one strategy for how to choose when to explore and when to exploit.

Recursive gradient descent: Imagine the space of possible outcomes is a valley, and you're somewhere on the side of that valley. Play an iterated game: from where you are currently, figure out which direction is "most downhill"; take a step in that direction, then repeat.

Simulated annealing: imagine there's a small pond or low spot or "false valley" on the hillside of the valley. To avoid getting stuck there and never finding the actual valley, change the size of your step, and each time add a small random jump. That way, you have a chance of getting out of the false valley and eventually finding the real valley. The metaphor is with metallurgy, where the final crystal structure is a function of the rate of cooling of the hot metal; you add a "temperature" parameter that automatically lowers over the course of the iterations, so that you take smaller and smaller steps with less random jumps. Instead of metallurgy you can also think about tempering chocolate when cooking.

Banana problem: Say you want to recommend some food item to a shopper on your grocery site. You might start by saying, based on the current item the shopper is looking at, what is the item that has been historically bought most often by people who bought that item? This is the "people who bought this also bought that" set of recommendations on Amazon. The problem is, bananas are so popular that no matter what item you're looking at, it's likely that the item most commonly paired with that item is bananas. So you need to adjust the ranking to account for the overall popularity of each item. So the new question is: given this item, what items are bought along with it at a rate above the base rate.

Collaborative recommendation systems: We will recommend something for you based on the past behavior of other shoppers.

[DeathsSilkyMist]

Originally Posted by Troxx [You must be logged in to view images. Log in or Register.]

If that will help you sleep better tonight, you go along thinking that sugar-bear.

I am not sure why you think you have provided anything of substance in this thread. So far you have given us a lot of gifs, bad troll attempts, and a parse that you deemed to be invalid. You have contributed nothing, but expect victory to be handed to you because other people agree with you. If you want to let the thread die, just let it die. But you haven't done anything here to support your position.

Originally Posted by bcbrown [You must be logged in to view images. Log in or Register.]

Optimization needs:

1. Some outcome you want to minimize (or maximize), and a way to measure it
2. A set of parameters to fiddle with

Multi-armed bandits: Imagine you're at a casino with a row of slot machines, which are sometimes called "one-armed bandits". Each slot machine has a different unknown payout rate. How should you choose which machine to play (exploit), and when should you choose to stick with the current machine versus move to another machine (explore)? Epsilon-greedy is one strategy for how to choose when to explore and when to exploit.

Recursive gradient descent: Imagine the space of possible outcomes is a valley, and you're somewhere on the side of that valley. Play an iterated game: from where you are currently, figure out which direction is "most downhill"; take a step in that direction, then repeat.

Simulated annealing: imagine there's a small pond or low spot or "false valley" on the hillside of the valley. To avoid getting stuck there and never finding the actual valley, change the size of your step, and each time add a small random jump. That way, you have a chance of getting out of the false valley and eventually finding the real valley. The metaphor is with metallurgy, where the final crystal structure is a function of the rate of cooling of the hot metal; you add a "temperature" parameter that automatically lowers over the course of the iterations, so that you take smaller and smaller steps with less random jumps. Instead of metallurgy you can also think about tempering chocolate when cooking.

Banana problem: Say you want to recommend some food item to a shopper on your grocery site. You might start by saying, based on the current item the shopper is looking at, what is the item that has been historically bought most often by people who bought that item? This is the "people who bought this also bought that" set of recommendations on Amazon. The problem is, bananas are so popular that no matter what item you're looking at, it's likely that the item most commonly paired with that item is bananas. So you need to adjust the ranking to account for the overall popularity of each item. So the new question is: given this item, what items are bought along with it at a rate above the base rate.

Collaborative recommendation systems: We will recommend something for you based on the past behavior of other shoppers.

Thanks for taking the time to post this!

[Troxx]

Yeah no, you are not the subject matter expert as what constitutes looking bad. If you need me to explain … it was a jab at the line of reasoning that simply because a stat is less prevalent on gear it means it is the most right choice. It’s not about which stat is the hardest to cap; it’s about which stat is going to actually do for you over the course of your character’s life. That’s the part of this (the discussion part) that you never got around to having. You’re stuck in your head very narrowly focusing on BiS arguments.

Ok so are we allowed to let this thread die now? You think I’m wrong. Everyone else thinks you are wrong. You think you’re somehow winning. Everyone else thinks your captaining the fail boat.

We are at an impasse.

Time to move along

[DeathsSilkyMist]

Originally Posted by Troxx [You must be logged in to view images. Log in or Register.]

Yeah no, you are not the subject matter expert as what constitutes looking bad.

Ok so are we allowed to let this thread die now? You think I’m wrong. Everyone else thinks you are wrong. You think you’re somehow winning. Everyone else thinks your captaining the fail boat.

We are at an impasse.

Time to move along

We can let it die.

You simply need to stop proclaiming you are victorious due to an argumentum ad populum fallacy. Nothing you said in this post correct, other than you have a strong desire to reach an individual impasse with me. You are not speaking for anybody else in this thread.

I am currently winning the debate, as I have evidence to back up my claims, evidence you have yet to rebut.

Here is my evidence: https://www.project1999.com/forums/s...&postcount=300

There is nothing wrong with admitting defeat, coming back with better evidence later, or letting the thread die with the possibility of people thinking you may have been wrong.

[Troxx]

Originally Posted by DeathsSilkyMist [You must be logged in to view images. Log in or Register.]

I am currently winning the debate

[You must be logged in to view images. Log in or Register.]

If that will help you sleep better tonight, you go along thinking that sugar-bear.

[DeathsSilkyMist]

Originally Posted by Snaggles [You must be logged in to view images. Log in or Register.]

If this is the thing keeping you awake at night either you are a simpleton who faces no real-life difficulties OR are a saint and you have nothing else in life to regret than EQ starting point mistakes. Since saints don't play EQ...well...

The simpleton is the person who posts insults when they are unable to articulate their points with logic and evidence.