Death to the RNG!

[Gaell Stormracer]

EQ's RNG is indeed random - you seem to be assuming that streakiness is somehow non-random, when it is an absolute characteristic of it -- which suggests a fundemental misunderstanding.

[Gaell Stormracer]

If you want to remove some of the randomness, it needs to go BOTH ways.

Why? If the end goal is to keep the overall number of attempts required to skill up the same, the same result can be achieved by reducing the skill-up chance by a microscopically small amount. Essentially, you're cutting the end off the distribution and re-distributing the stuff over the tail into the main distribution. Same end result, but without the soul-destroying droughts.

[Itek]

Well the math is on my side.

Tanker agrees that the math and game design are on my side.

Ngreth agrees that it's not a great idea.

Everyone seems to understand that IF it happened it would happen so rarely as to NOT happen.

And as for "statistics don't effect people outside casinos" ....

Statistical analysis of human behavioral psychology in everyday life 101... have you walked into a grocery store in the last year. Then you have both been a subject of research, experiment, and it's applied results.

"no one own's the mouse's paddle in EQ" ... Wrong. SOE owns the paddle, and they want you in the cage as long as they can keep you there. The problem of SOE is they need to bait the paddle, as the cage door is open and the mouse can leave any time it wants to.

[Gaell Stormracer]

Well the math is on my side.

Uh? No. Your last post proved conclusively you didn't understand what was proposed.

[Itek]

*sigh*

Had to go back and read all of olhoss' post. Since I have him on /ignore I wanted to be sure I'd read enough of his proposal to have the gist of it.

I guess his "proposal" was even more... /eyes mods...

What was proposed was more simplistic than I had thought in the first place. It makes it LESS complicated and MORE of a bad idea. See Ngreth / Tanker posts for why. (They are random for a reason, and as hard as they are for a reason, and convincing the money men otherwise is hard for a boat-load of reasons.)

"when people get these crappy runs of 300+ combines without a skill up they quit."

Really? Do you have evidence of such bad runs?

"yeah, it happened to me in 2 skills, and nearly as bad in 4 more"

So... it happened to you... and you didn't quit... but did MORE... and that's evidence that when it happens people quit?

"yes"

Um... no. It's evidence that random reward schedules are addictive, which is precisely what we want. Thanks.

Coding a "failsafe" to hard-limit a number of failures wouldn't be hard. (The hard part would be scaling it, if you went to that much effort, from 200 skill to 300 skill say.) It's just not a good idea. Has nothing to do with my personal feelings for olhoss, it's just that multi-billion-dollar industries rely on random payout for a very good reason. It works.

Oh, and the math still works out, in that you can calculate the precise likelihood of getting 100 successive failures, and thereby tripping the "failsafe" limit on failures. Even if it's 98% likely to not get a skill up the odds of getting 100 failures is about 1 in 8.

Math always works, that's why it's math.

And on that note, the rest of the discussion (where to put a limit, if a limit is useful, if limits should be two-sided, yada yada yada) boils down to opinion / philosophy. And I'll refrain from giving an opinion as ... it just doesn't seem likely to matter.

[olhoss]

Here's why the evening out issue doesn't hold water:

Getting 10 skillups in a row from 20-30 in NO WAY "evens out" taking 700 combines to go from 295-300 because of the difficulty in acquiring components (be it rarity or expense).

I am not ignoring the math with this, I am COUNTING on it. The pity skillup SHOULDN'T come up very often, and I do not believe it would, because I have not proposed making it easier to skill up. I have proposed a simple method to prevent long dry runs without skill ups.

This wouldn't significantly change the number of combines it takes overall to max skill, EXCEPT in cases where someone would otherwise have suffered from an anomolous run of bad luck, where my proposal would at BEST hold them to twice the expected number of combines per skill up.

Ngreth said it would be a big hassle to code, and I am not adept enough at coding or familiar enough with the inner workings of the EQ code to argue that.

My math, however, holds up.

[Ngreth Thergn]

Still just having a conversation here with no plans on making a change...

I would modify my "anti-pity" call to only happen if you get a pity skill-up. Preventing the double hit, but allowing double hits to continue normally on a normal skill-up... so there would be some sort of "last skill-up was a pity skill-up" flag.

[olhoss]

based on that necessity...perhaps then a plus/minus system where a non-trivial combine adds one, 100 gets you a skill up, and getting a skill up subtracts 100. Thus, you might still get 2 in a row, but it means that you can't get another pity skill up for 200. This makes the math so that you can't increase your skill up rate but you CAN keep it from going more than double the expected. Examples:

Skill 290..100 combines, Ding! 291, counter to 0. 1 combine Ding! Random skill up, counter to -99. If this person continues random luck, the counter will have only affected one run of combines, and gained them one skill up at a point where they have done about twice the number of expected combines.

Skill 290, 78 combines, Ding! 291, counter to -22. 122 combines Ding! 292, counter to 0. 55 combines Ding! 293, counter to -45. 3 combines Ding! 294, counter to -142. 62 combines Ding! 295, counter to -200.

In other words, the more luck you have in general, the less likely you are to ever benefit from the pity counter. Someone who goes from 290 to 295 in 5 combines would not get a pity counter for 296 until 495 combines later. and even THEN they would have gone from 290 to 296 in 500 combines.

Of course, this would hose you if you are lucky at lower levels. I would suggest either not using it at all until a given number (250, 280, 290) or perhaps resetting to 0 every 10 to prevent the counter from getting so low from random skill ups at lower trivials that it never comes into play at the higher ones.

[Rinikku]

NOTE!!!!! this is just for conversation sake and in no way anything planned, or even really being considered at this time.

But what did people think of the "reverse" part of the pity counter Tanker mentioned.

Honestly, I think I had more bad luck than good luck with skillups. I would definetly take the antipity for the pity skill up. I think on my last run to get to 300 in alchemy (most on the last 3 points) I used several THOUSAND nodding blue lilys. This was a horrible bit of bad luck, and I would greatfully take away the possible three skill-ups in a row so that I would never have that much bad luck again. I've always believed in taking the good with the bad.... the RNG does not believe in that saying at the moment. LOL.

[Abazagaroth]

Tradeskills in everquest is not a casino. People do not play to "win" skillups. Skill caps that are eminently achievable make this a complete logical fallacy even if it wasn't limited by the practical fact that the vast majority of tradeskillers do so in order to be able to make something for themselves, for their guild, for resale, to "max" all skills, or for quests/items (artisan charm, aid gimel).

A casino is a casino. Its about "winning", about gaining more than losing. The everquest casino is not a casino (you can't gain winnings, you simply pump in plat over and over until you finally get the item you want or quit, its a random item generator with a fee per turn, not a casino). These analogies are off the mark and simply irrelevant to tradeskills or this discussion.

The everquest RNG is not an iterative function. If the devs want to put a hard limit on the number of unsuccessful attempts per skillup, that still does not make it an iterative function.

Even if the limit was placed to something like 100 (or hell, 60-80 even) it would have almost no impact on the rate at which components are used or farmed for skillup purposes. I doubt there is more than 2-5 skillups per tradeskill run to 300 in even the harder skills that go longer than 100 (and I personally had a 500+ and 300+ run without one, so I know they happen, but they are rare), and as Tanker said, it just ends up very minimally reducing a very minimal subset of combine durations prior to skillup.

That begs the question of why is it worth the time wasted on changing it, just so that 1 in X tradeskillers doesn't have a pure hell run once in a blue moon?

People aren't pulling a one-armed bandit hoping for triple 7s and hoping to hit the jackpot, people are simply applying resources with various levels of luck in order to progress along a predefined path and reach a limit. They are two very different mechanisms.

There should be no concern about making tradeskills less interesting as a money sink, as long as there are things worth making people will do them, that has always been the #1 driver of getting most people to do tradeskills.

With that said, it is a huge mistake to implement some "anti-pity" code even if a "pity code" was enabled. For the following reasons:

1) If there is not a chance to get a skillup on any non-trivial combine, it completely contradicts the basis of the system in the first place

2) It is, more importantly, not equitable, as the number of dry spells (100+ runs) is likely outnumbered by short runs, and, more importantly, it would mean that on every single skillup there is X combines after that you can't skillup on. So let's say you would have had a 119 and 123 bad spell from 1 to 300 in some skill. That's 42 combines you "save", but in return, (lets say 5 anti pity combines) you lost 1495 combines that there was zero chance you'd get a skillup on.

Basically, since you get 299 functional skillups (300 doesn't matter since you don't skillup after) where the 5 combine anti-pity code would go into effect, in order to "break even" you would have needed 1495 combines OVER the pity code limit (e.g. 100 or whatever) during your 1 to 300 run. I doubt more than a handful of people *ever* in eq have had that many runs that were that bad.

In order to balance something like that out, you'd need to set the pity code extremely low. And that is easily calculatable as the average combines per skill up per tradeskill minus the anti-pity code number of combines.

Doesn't sound much like everquest tradeskills.

If any change should happen, it should be to modify the everquest RNG with an iterative algorithm, which increases, slighly, the probability of a skillup with subsequent failed attempts. Extreme bad runs are still possible, they are just so extremely unlikely as to almost never happen, and such a function could be balanced to make no discernable impact other than to reduce the longest bad runs.

[Abazagaroth]

Still just having a conversation here with no plans on making a change...

I would modify my "anti-pity" call to only happen if you get a pity skill-up. Preventing the double hit, but allowing double hits to continue normally on a normal skill-up... so there would be some sort of "last skill-up was a pity skill-up" flag.

That makes zero sense. All that does is effectively add X combines to the "pity" limit (i.e. if you go 100 fails then get the pity skill up, then can't skill up for 5 more combines, it is no different than making the pity cap 105).

[Wyvernwill]

True randomness has no guarantee of "evening out" in such a way...

You're 100% correct. Afterall, my researcher is 2 skill gains from 300 and his trophy is at 46%. He's been and has stayed insanely far ahead of the curve the entire time. I'm more than willing to place bets that 300 comes before he breaks 60% exp.

In the same respect, in tailoring, my trophy dinged with me still at 290. It's definitely random, sometimes you get good, sometimes you get bad. Like in life... some people don't realize that without the bad, you never realize what "good" is.

[gggrant]

Ooooooo,
We could have a pity counter and a PITA counter.




Edit: Sorry, couldn't resist.

[Abazagaroth]

You're 100% correct. Afterall, my researcher is 2 skill gains from 300 and his trophy is at 46%. He's been and has stayed insanely far ahead of the curve the entire time. I'm more than willing to place bets that 300 comes before he breaks 60% exp.

In the same respect, in tailoring, my trophy dinged with me still at 290. It's definitely random, sometimes you get good, sometimes you get bad. Like in life... some people don't realize that without the bad, you never realize what "good" is.

That's because if a combine is 100 trivial above your skill, you don't get credit on your trophy, and most of the research combines that people do over 220 or so are more than 100 trivial above, since those combines go up to 400+.

The 100 trivial thing needs to be changed for any combine over 250 considering how many research/tinkering (and quite likely smithing and tailoring next expansion) combines that are over 100 more than skill that people actually do the combines at.

[olhoss]

Some of the alleged numbers crunching above is really way off base and has nothing to do with either of my suggestions.

My first suggestion was for a simple code addition that would prevent outrageous runs of bad luck which are a major source of frustration and cause considerable and unreasonable loss of time and resources. It would add very few additional skill ups, and would in no way unbalance the game.

My second, based on Ngreth's saying that there would need to be something to prevent this from resulting in a little too much help, is to make it more difficult to benefit from such a system, which I believe unnecessary but a workable compromise.