Please look at the loot code & RNG

So a random algorithm will return some number between a min and max. This number then needs to be redistributed to the correct range. This is very complex problem, and the naive solution (which most people do) is extremely inadequate (modulus)

Also how you seed will also depend on the algorithm as well as how long it's going to be needed to be run.

You also basically said RNG is easy if you RNG.

One would not think it had anything to do with it BUT (big but) experience has taught us the less consistent the coding the more likely to have exactly that occur. I have never seen EQs code, but well founded rumor had it as a bit of a hot mess in places. Bard song code comes to mind immediately, gotta give them credit on that bear given it worked as well as it did. Frankly I thought it amazing they pulled it off like so much in EQ at that time.

Not saying at all this is the issue, how would I know, but many times we would pull some logically random data to seed the RNG or other function only to find some later change completely unrelated would influence the RNG. Programmers do some odd things in particular late at night or in a rush thinking it an elegant solution only to find long after that solution somewhat flawed. Just the cost of the learning curve and try not to pay for the same lesson twice.

You don't know the math behind it, or the code behind it. Stop pretending to. You are still missing the point, Astley and AgentOfChange. Something is wrong and it isn't just one person noticing it. If I get time I'll go back and do some in era number crunching. It isn't just a "feeling", it is a noticeable trend in our loot, all the way back through NToV (whend I started with the guild) all the way through the end of PoP.

Right... because the items in the loot table are NOT equally probable to drop.

Like, there are 15 loot items, and 4 BP patterns that can drop off Quarm.

I'm not sure, but I think he can only drop 1 chest pattern, if he does, so that may be restricted a bit. Or they could just be very rare from my experience. We've gotten 1 maybe 30-50% of kills at most

As for the Silver Hoop, I think that was designed to be one of the ultra-rares off Quarm, due to it having over-haste.

Go ahead... bring some numbers. I'm saying multiple pulls of loot have been in this game since 1999. It's not uncommon, it's intentional except on fights they specifically made it work otherwise. It hasn't changed for more or less than at any other time, except for a few items that were removed from the loot tables of various mobs over the years.

The cord in question is a common drop. It is on multiple loot tables in VT, meaning they meant for more to go out in the world.

I'm in the biggest zerg guild of neckbeards and cheeto-crusted basement dwellers on the server. None of us have jobs and we do nothing but train and grief and kill EQ raid mobs all day and all night. We've killed thousands upon thousands of raid targets in both Open World and Instances since the server started, which I would argue is a pretty decent sample size. While there are certainly times where the loot gods are not favorable, and even times you might question the drop rate of certain items, I've seen nothing to suggest it is anything other than bad luck or RNG. Your sample size is just not large enough to make a definitive claim that "something is wrong".

I agree that I do not know the exact coding or the exact math behind the drop rates for each mob, and I am fine if you want to say that you think they should change drop rates or loot tables or whatever based on your opinion, but I am highly confident that there is nothing different or out of the ordinary with "RNG and the loot code" as suggested in this thread.

I got you Malaki

The seed was never generated from the RNG function except when it really did not matter. It is fast to code and appears utterly random and that was good enough in almost all cases. Pulling in a true random seed entailed pulling it from real world interaction(ever seen the I am not a robot pick?) but one had to be consistent and code control imperative. One instance I remember was my partner in crime changed a counter of time in zone where I was using the last two digits, seconds and tenths as the seed, my mistake and I never used that again. Instead I went to the system clock and generated it there from then on. Primitive sure but this was a long time ago and it was not the Space Shuttle we were dealing with there. Sometimes one wished for a real life dice roll.

Like I said and hold to the RNG was easy compared to getting the loot tables to reflect logic because everyone has an opinion.
In the end it was pathing that convinced us to get a new hobbies, was not paying the bills just to run the server any way. Pathing was a nightmare that we just never were able to get right and frankly neither did EQ so I dont feel bad given they cheated a bit and it is still wonky. Heh have to wonder if any one is left that remembers where they got that solution from this far down the road.

Sorry to derail, but I am curious why mod is such a bad choice for that?

Because it does not translate a uniform distribution to another. So let me steal an example from stackoverflow since I don't feel like doing math.

Let's assume our rand function returns [0, 32767] and we want a number between [0, 2]. So how do we use modulus to get this in range? We do % 3. The issue comes from the fact that in our given rand range mod 3 will give us 10923 0s, 10923 1s, and 10922 2s. So if we have to roll a 2, that's a lot less common than 0 or 1! Which is not what we wanted at all.

Never get coders talking about RNG, just dont, it aint fun.

As you pointed out later in the thread (or someone else), what sample size we have is going to be too small. Even if I did compile stats, the response is going to be 'well thats RNG for you' so frankly that doesn't seem like a valuable investment of time.
On the other hand, If I was an EQ dev, I could fairly easily get some stats compiled on this by mass killing and respawning a zone like time and then analyzing two things- first the proximity of actual drop rate % compared with the intended probability, and second of all the incidence rate of dupes, trips, and quad drops.

Magelo has a sample size of 139 kills for Saryrn and regarding Times Antithesis it lists it at 28%. We've still not seen one, but if we got one on the next kill, that would put its drop rate at under 10%. I get that probability isn't going to match up with actual drop rates because random, but when they are so distinctly different, and that same mob has a streak of dropping 2x rings of evasion - which by the way has a smaller drop rate % in that same 139 kill sample , if I had the loot table data I could really start to make some meaningful comparisons and atleast identify if there is a problem. (Meanwhile we've still never seen the item with the highest probability of dropping - aka Symbol of the Planesmasters at 31%)

Could just set all loot tables to act like the Twilight Sea quest armor drops: you get a massive pile of drops off each named, arguably the entire loot table each time.

That would result in some rotting...

The drops from each Thall Va Xakra boss are roughly equal chance. Of the six items, four have a 17% chance and the remaining two have a 16% chance (totaling 100%).

I logged over 1500 drops from these NPCs and the chance for any given item to drop didn't seem far from what I saw in game. I suspect they'd even out further with a larger sample size.

NUM DROPS ACTUAL % DB % ITEM_ID NAME
160 17.0% 17 28887 Cloak of Icy Shadows
153 16.3% 17 28888 Shadow Footpads
160 17.0% 17 28889 Living Symbol
156 16.6% 17 28890 Cord of Midnight
164 17.4% 16 28891 Bracer of the Silent Star
147 15.6% 16 28892 Cap of the Silent Star

With regards to seeing dupes or trips or quads, when rolling 4 six-sided dice (slight simplification because some items are a little less common than others), there is a 72.2% chance that there will be a duplicated item.

There are 6 * 6 * 6 * 6 = 1296 ways to roll four dice, and 6 * 5 * 4 * 3 = ways to roll non-duplicates, so there are 1296 - 360 = 936 ways to roll at least one duplicate and that comes out to a 72.2% chance of seeing a duplicate. (Please double-check my math)

In my tests, there was at least one duplicated item 72.3% of the time, which is the expected outcome.

However, like you mentioned, I did notice some odd patterns.
The first and fourth items were the same item 30% of the time.
The second and fourth items were the same item 28% of the time.
The first and second items were the same item 0.3% of the time.

That's... strange. Some further research would be required by someone smarter than me to determine what could cause this. Thank you for pointing it out.

"Hi I don't play on this server with different rules, but I feel very strongly about other servers pixels."

Prathun, thank you for looking into that. It truly is interesting to see.

What that tells me is that the RNG does even out over time (which does make sense- even with instancing our sample sizes were nowhere near the level of what you could simulate) but that the weird patterns you saw in duplication and streaks of items is what is setting our "gut tests" off about something being wrong.

Take the sample size of what players can actually accomplish through instancing. If Time gets raided 30 times during PoP era. Does the loot drops match its designed drop rate %??? Should a guild raid a zone every week, and only see a specific loot drop in era 1-2 times with a 10-20% drop rate?

Thank you very much for spending some time to look at this Prathun. Sincerely!

The selfish part of me would ask you to run some tests on in PoTime to see if the class specific items are evading their intended droprates... (TA, Serpent of Vindication, Shinai etc)