RNG and streaks

[Splup]

Heya,

I remember that Draconi (or some other devteam member) said that RNG is bit "broken" and has a habbit of creating streaks. I just can't find it anywhere.

Can someone verify if I remember correctly? And if I do, were they supposed to be working on it (To make it "less streaky" and more random)?

 

[grandpa otis]

RNG streaky??? I find it RARELY is anywhere near right in its posted percentages,,,I appear to average about 2 to 3 times worse than it reports over the long hall, ...Want to see streaks, just try imbueing...yesterday while trying to imbue a 6 pc suit of armor, on the 5th mod on 2 pcs I got the following result,, imbueing phys resist on gorget had posted on menu a 17.4% chance of success(that means it should avg out to be successful about 1 in every 5 3/4 tries), it took me 41 attempts to get it,,, gloves on same suit had 16.7 % chance stated in menu,,, tht pc took me 37 attempts,,, sound streaky??? and even worse in my opinion is trying to imbue a mod tht took 4 relic frags that had a 91.2 % chance of success and got 3 failures before getting it,,

 

[Splup]

RNG streaky??? I find it RARELY is anywhere near right in its posted percentages,,,I appear to average about 2 to 3 times worse than it reports over the long hall, ...Want to see streaks, just try imbueing...yesterday while trying to imbue a 6 pc suit of armor, on the 5th mod on 2 pcs I got the following result,, imbueing phys resist on gorget had posted on menu a 17.4% chance of success(that means it should avg out to be successful about 1 in every 5 3/4 tries), it took me 41 attempts to get it,,, gloves on same suit had 16.7 % chance stated in menu,,, tht pc took me 37 attempts,,, sound streaky??? and even worse in my opinion is trying to imbue a mod tht took 4 relic frags that had a 91.2 % chance of success and got 3 failures before getting it,,

I have noticed the same thing with imbuing. Thou I have gotten some nice succeed streaks with low chances too. I feel like getting breaking and succeeding streaks in enhancing too.

 

[Alezi]

The streakiness is determined by the alignment of moon, other planets and the formation flying of birds outside.

 

[Chap]

when it comes to Random Number Generation it can never be random.

I guess by broken he meant not as random as one would think.

Original UO was made 11 years ago, the definition of random since then have changed.
What was a good random generator at that time, is these days a bad random generator.

the question is how much effort to put into coding of a random generator.

 

[o Kraken o]

The streakiness is determined by the alignment of moon, other planets and the formation flying of birds outside.

Cool! Could you post some moon/planet/bird formation charts, so I could make better predictions of the RNG when I key for peerless ;-)

 

[NuSair]

the question is how much effort to put into coding of a random generator.

It was bad then and it is bad now.

In programming there is a function call srand, and typically looks like this:
srand ( time(NULL) );

The thing is, is that srand really isn't random and normall people use the time on the CPUs clock to modify the result to be a 'random' number.

Why isn't srand random? Well, it's used for testing. If you use srand without some modifier then you will get the same numbers back everytime. So, it is helpful to put varibles in to get out results, and if there is an error, it is easier to track and see what caused that error.

There are many ways and many differing opinions on how to program an effective random number generator. Just google or bing how to program a random number generator and you will get alot of results. Some of those papers are good....other, not so.

 

[WarUltima]

I got 4 crimson cinctures dropped to my backpack in 4 Lady Mel within 2 hours. We 2-men'ed her. Is that streaky?

 

[Lil' Lucien]

WarUltima
Re: RNG and streaks
I got 4 crimson cinctures dropped to my backpack in 4 Lady Mel within 2 hours. We 2-men'ed her. Is that streaky?

Since when do they appear in the backpack?

 

[WarUltima]

Since when do they appear in the backpack?

I am pretty sure they do drop in backpack. I have had quite a few dropped in this fashion. And a few of them appear on peerless boss corpse.

 

[Professor Moriarty]

I got 4 crimson cinctures dropped to my backpack in 4 Lady Mel within 2 hours. We 2-men'ed her. Is that streaky?

Yeah that probably means the next 40 dropped 0 Crimsons working with the current RNG then you will get a run of drops again.

When I smelt ore I get 3 fails in a row with val ore then 1-2 without failing, 3 fails 1-2 without..etc..etc I see it when I enhance bows 3 –4 success in a row then 1-3 fails.

Imbuing, as its been said above. You get a percentage in one corner telling you’re your chance, but when you have an 80% chance and fail 6 times its not working, as it should.

Its broken it’s been broken for ages and me thinks its going to stay broken for a lot longer.

 

[Tina Small]

Try taking a peek at this thread from last June: http://vboards.stratics.com/showthread.php?t=144665.

 

[Cailleach]

The RNG hates my guts. There's no other explanation for the amount of stupid fire seeds I have when I really need black, white and magenta

 

[Professor Moriarty]

Try taking a peek at this thread from last June: http://vboards.stratics.com/showthread.php?t=144665.

So the long and short of that thread was they know there is a problem with the RNG, but the RNG is tied into the core of the game engine. Then it looks like Daconi was working on a fix while working on SA….then SA goes live and Draconi gets thanks for a great job but were no longer requiring your services.

RNG is still broken, and no Draconi to push a fix forward.

Splup, this might be a good one for ask the devs question

 

[Lorax_Pacific]

I'm trying to make a suit for some of my characters and one item I thought would be good was the Bramble coat, which my carpenter doesn't have the recipe for. So I go do the heartwood quests and chunk through the agony of that quest. I spent a couple days lots of time and logs without success. One thing that was weird is last night I decided to eat and fill my belly before doing the quest. After a few times I got a heartwood runic.

What was a real agony is that after a day with no searchuo result there was one recipe available for 5k and it was 15 minutes ago. I run over there as fast as I could and it was bought. The search tool keeps displaying that same vendor laughing at me while I'm in the heartwood grinding.

I've gotton like 3 ironwood crowns though.

Regarding color streaks: I have noticed that backpack colors during the heartwood quest reward come in streaks. I should keep track of the pack color when I get a saw or rare recipe...

-Lorax

 

[Beastmaster]

The thing is, is that srand really isn't random and normall people use the time on the CPUs clock to modify the result to be a 'random' number.

The cpu clock is typically the core of all RNG routines. That's why streakiness is observed. If you do something at close to the same speed all the time and it happens that your action cycle is in alignment with the RNG cycle you will get the same result over and over until the cycles diverge.

 

[aarons6]

The cpu clock is typically the core of all RNG routines. That's why streakiness is observed. If you do something at close to the same speed all the time and it happens that your action cycle is in alignment with the RNG cycle you will get the same result over and over until the cycles diverge.

so your saying then to try to either max out your cpu cycles or reduce them prior to doing any rng type task?

 

[Jhym]

Yep, Draconi had said some of the code seemed to be broken but wasn't specific about it.

The problem is we don't have enough info about how they coded it to determine how bad the problem is.

I still suspect that the supposed streakiness is actually from rounding errors in the determination of the dice "rolls". If you scale a large number down to a percentage roll, it is likely that you would get supposed streaks, even though the RNG is actually generating wide ranges of numbers.

In any case, we can't come up with any further info until the devs decide to tell us more or ask us to test a new implementation.

 

[Basara]

There was also an expansion on the mentioned RNG thread on Draconi's Facebook page that was associated with UO; that facebook account disappeared when he left the UO team, the essay going away with it (He's since started his own personal FB page).

 

[JL from Europa]

Does the RNG also apply to paragons? Because I've been hunting balrons for a few weeks now and I notice how I get 4-5 paragon balrons in a row, then a long time normal ones spawn, then I get a few paragons again and so on.

 

[Beastmaster]

so your saying then to try to either max out your cpu cycles or reduce them prior to doing any rng type task?

No, the cpu that controls the RNG would be the server cpu, not yours.

 

[Thunderz]

I know the feeling about the RNG but it goes both ways take imbuing for instance first, iv imbued a property that had 5.2% [chance so i should get about 1 in 20] i got it first time! But its gone the other way aswell its take me 65 tries to imbue an item at 8.7%!!

Other example is DOOM! I had 2 years of not getting a single arti from about 10 hours a week there [way back] then i xfered to napa and on my first doom run bagged me an orni

thunderz

 

[MalagAste]

RNG just plain hates me. I'm sure of that.

 

[Megilhir]

My 2 copper on "Randomness".

Envision a Roulette table in your mind.

There are 36 whole numbers and a 0 (international) or 0/00 (American Standard) at the top of the table of numbers.

Therefore it would seem logical to state that a number had a 1/38 chance of coming up.

Yet due to "randomness" the chance of any number coming up is exactly equal to that number being repeated infinately or not coming up at all. Improbable yes, but entirely possible.

So there you have it.