Gradual Reduction in MF Effectiveness

[gekko]

However, if we are trying to find out if "good" items drop less over time, we must not count high durability items as "good" items.

What's a "good" item? As has already been mentioned, a 450% ED rare collossus blade with 30% IAS and normal durability is a heckuvalot better than a hellplague from a user standpoint, but from a purely mathematical MF standpoint, the hellplauge is "better" quality.

We could also suggest that blizzard employs a bunch of teenagers they have locked underground who decide what item drops for every single monster killed in every game on every server. In theory, it *could* be done. It doesn't make any logical sense, but it could be done.

Your theory is much the same. While it could indeed happen (and frankly, it's much more likely than my supposed "chained up teenagers"), it doesn't make any sense. Why would blizzard implement a realm only bit of code designed in such a roundabout way? If they truly wanted to render boss runs less effective, there are many very simple ways to do this -- the MF multiplier that has been mentioned being the most obvious.

The idea that it is not intentional but rather a simple error in the RNG is even more confusing -- why would blizzard use a different RNG in the multiplayer version of the game than in the single player?

gekko

[Jeger]

What's a "good" item?

:lol: Why are you asking that of me? You'll have to ask the people who think the number of "good" drops decline with time.

Seriously, if there is a real phenomena, then "good" items are probably uniques and LoD sets. Those are the ones that you get excited about when you see them drop.

If they truly wanted to render boss runs less effective, there are many very simple ways to do this -- the MF multiplier that has been mentioned being the most obvious.

Actually, the way I mentioned is the simplest. If I wanted to implement a scheme that reduced the effectiveness of boss drops for botters

1. I would only put it on the closed server, so it could not be disassembled,
2. I would not modify the treasure loot routine, as that seems to work already,
3. I would put a compile-time flag in so I could compile to different server versions using the same source code,
4. I would track boss kills daily for a toon, and
5. I would generate a failure chance based on the tracked boss kill. If a random roll indicates failure, then I would reroll the loot drop when an set or unique dropped and the item was exceptional or elite.

It's very simple, and probably the best way to do it if Blizzard decided it needed to be done. If you disagree, it merely means you program differently from me. :)

The idea that it is not intentional but rather a simple error in the RNG is even more confusing -- why would blizzard use a different RNG in the multiplayer version of the game than in the single player?

As I said, I will be glad to discuss the RNG theory in PMs. :)

[Jeger]

There is a hypothesis that you get less "good" objects later in repeated MF runs.

Upon reflection, it seems a lot of people are basically assumng that hypothesis is false while discussing how to test it. While I agree, that's a dangerous stance to take while designing the test. Let's take a minute, and assume that the hypothesis is true. That could be a result of a deliberate action by Blizzard, or an "accident."

If the action was deliberate, then Blizzard appears to be hiding it. It's not in the open server that Jarulf can find, and it has not been announced. With that in mind, there was probably a little effort spent in making it difficult to see. If I were in charge of the project, I would be more sophisticated than the simple scheme I proposed above. (That scheme could probably be coded in an afternoon.) However, it would be basically similar to that scheme, and the bottom line is that combining high durability items with uniques would never detect what was happening.

If the action was not deliberate, then it seems the obvious place to look would be the RNG. On that subject, to answer Jarulf's issue, when I speak of time, I am speaking of repeated calls to the RNG - not use of a time variable. In other words, the ratio of HD items to uniques could rise as the RNG was called more often (which would then be later in time.) This would be an artifact of the particular mathematical function used for the RNG, and how the loot routines are coded. Since it wa an accident, it could have worked the other way. (The loot could get better the longer the runs go.)

Frankly, I am interested, but not enough to to the actual tests myself. :lol: If someone is interested enough to actually do the tests, though, they should not assume that everything is working as they think.

[JustAGuy]

If someone is interested enough to actually do the tests, though, they should not assume that everything is working as they think.

I am interested enough to do the tests, and lots of them.

I've already listed how I'm going to do it, but the problem now is time. Real Life™ has decided to make Diablo II play a little less frequent for the moment. With this in mind, I'll do the tests as time presents itself.

The arguments presented in this thread are very convincing. They mainly lean toward the aforementioned MF effect being completely psychological. When I do the runs, I won't be doing them assuming anything; I'll be doing things consistently, and methodically. I'll be cataloguing everything that drops, and I'll leave the data for others to decipher.

So, just as you think this thread is about to die, it'll be back. Muhahah. I started this thread, and I intend to finish it. That doesn't mean I'm biased toward my own opinion of MF decreasing in effectiveness. I am simply trying to uncover the commonly held belief (superstition) that many users hold regarding the quality of drops over time. Be there a mechanism to decrease quality, a problem with the RNG, or a bug we don't know of, or not, I'd like to provide some data that will help put this issue to rest once and for all.

[bernard shakey]

someone bragging about starting a hoax in the lurker loung? I am deeply sadened. This goes against everything the loung stands for. maybe you should consider posting on some other forum?

[HunkyDory]

I have been doing some testing and as far as I can tell there is no gradual reduction.

What I did notice were two things :

Psychological it is hard to not call a drop bad when Isenhart's Case, Soulflay and Naj's Circlet drop.

Not items you can call uber but still a good quality drop with one Unique and 2 Set items.


The game has a tendency to make the first drop of the day a good quality drop, usually one Unique and two or more Set items.

First drop of the day means after you have been offline for some time.

It is not all the time but that first drop seems to be favoured but once again could be psychology and coincidence.


Regards, Hunky

[JustAGuy]

I have completed 52 Mephisto runs, using the "Bishbosh bug". I catalogued all the drops, every single item that fell and their durabilities. Lots of failed sets and uniques have been dropping.

After 52 runs, there is no change is the quality of drops. All can be considered good. Those that look poor consist of mainly failed sets. I've collected a ton of data, but I'm not sure if I should just post it. It's 345 rows of data in a spreadsheet. I don't have any webhosting either...

Of note, Mephisto was kind enough to drop me a Rainbow Facet, although it's a low end poison one. Still cool to get one, though. Some other drops of note: two Griswold Armours, a Corspemourn, two Duriel's Shells, an Occulus, Lidless Wall, Nord's Tenderizer, Gavel of Pain, two Flamebellows (I have 6 now, all from Meph! He just loves dropping them), two IK gloves, 3 Cathan's Seals, and Trang Oul's scales. I've given lots of this stuff away.

I'm going to re-format the data, a summary of the drop per line, and post it here. Maybe someone could cross reference the durabilities with d2data.net and maybe see a pattern of failed sets and uniques?

After seeing the quality items that I've been getting, consistently, I am now a firm believer that there is no reduction of MF on the first Mephisto drop, and that the effect is all psychological in that situation. My small but detailed sample data backs that up. When using the "Bishbosh Bug", the quality of the drop is incredible, all the time. I've gotten all blue drops before, but most of the items had very high durabilities.

When not using the bug, a "quality" drop only comes up once in a while, which leads people to believe that things are getting worse when they're not getting that lucky drop. The reduction effect may still exist, though, when not using the bug. There could still be something causing things to get worse and worse, and since my testing was done on the first-kill only, I can't tell for sure what is happening in that non-first-kill situation. I'm going to bring my level 87 hammerdin out of retirement to do more runs without the using the bug.

[Ruvanal]

You are using a 'bug' in the way the quest drops are generated, not the 'Bishibosh Bug'. The 'Bishibosh Bug' was present in v1.07 to v1.09 and caused Super Uniques that were activated and then moved away from to end up reverting one of their internal pointers to use a zero (bishibosh's index in superuniques.txt) instead of the correct index. This resulted in the SU using Bishibosh's TC for generating the drop instead of teh correct TC. This bug was fixed in the v1.10 patch.

[JustAGuy]

My mistake. Not sure what this particular bug is called, if it's a bug at all. I had a feeling I was using the wrong name, since Bishbosh can't be killed this way, that is, he has no "1st drop" as far as I know.

I was, though, doing runs and getting the 1st kill quest drop every time, and the loot was very nice. I haven't started without the 1st-kill drop yet, since my chances of getting a Herald of Zakarum or a nice Phase Blade seems to be better this way...

[GreyReaper]

You know I think you may be right. But i think it is many very small factores that cause the loss of rair items such as, if you have 30 Pgems on your person or equiped to wepons then you start to get less gems. But, honestly that is just a hypothesis I could be wrong. I hav'nt actually tested that hypothesis, it is just an observation I have made. Just like I have been noticing that certain character class' tend to find certain character class only items thats other character class find it harder to find. For example I find that one of my Barbs finds sorc globes more than my other characters. Now this is purly just observations it is not actualy tested and proven. BUT you may want to try and find a link in your characters so youwill know what character class you may want to buddy up with in B-net. Well now back to the main subject, I think that things like having sertain rare items for to long creates a decline in find such rare items. Now if you switch out items on yourperson frequently then you mayget more rare items. But there are other theries as well such as, I have a new Barb thats skills are completly masterys. Well without mf he tends to find more rare items then my other new characters. Now this bard is only at lvl 18 but since I have been playing him he has found 4 times more rare items then all my other characters at lvl 18. I think it may have to do with you get more rare items when youuse no magic to kill monsters and only just the regular attack. But like I sayed it could be a number of very small factores that are going into it. Such as not only it is the normal attack that gets more rare items but is maybe that the barb is attacking normaly. Maybe There should be a mf barb with +20 to the skill find item with lots of mf.

[adeyke]

There is no truth to the idea that having gems or using two weapons makes you find less gems.

There is no truth to the idea that you'll find less of the class-specific items that your class uses.

There is no truth to the idea that using normal attack will give you better drops.

There is no truth to the idea that having an item already will reduce the chances of it dropping again (the only restriction vaguely related to that is that each unique will only spawn at most once per game).

Really, these assertions are all in blatant contradiction with all actual knowledge about the drop algorithms. If you have a ridiculous claim like that, prove it. Don't just toss around rumours without any basis in truth. If we want false rumours, we can make plenty of our own ("If your character name starts with A, and your experience is a prime number, and at least one of your items is a rare with "Chaos" in the name, and you're between 20 and 30 years old, and you've squash all the critters in act 2 in exactly 1 hour, and you have your entire inventory full of stacks of 2 keys each, and you defeat a hell quill rat using a Windforce socketed with an El on the midnight of a full moon, you'll be 0.13% more likely to get a good drop!!!"), but such rumours are completely worthless. We want knowledge, not rumours.

[Warleader]

Ejem, i put a El in a Grandfather for mistake in 1.09, that counts ? :(

[dkass]

Does the initial entry of a character into a game with a unique item count as spawning it? I ask because the game actually has to create the item "from scratch" at that point.

If so, could this be part of the problem for some players? They're using their runner/game creator as a mule of the "loot" they find so as the character finds items and keeps them, they're reducing the pool of unique items they can find (and increasing the pool of rare items).

[adeyke]

No. Items brought into the game are completely irrelevant. Only those uniques that actually drop or that appear in a gambling screen (whether you know it or not) are no longer spawnable afterwards.

[dkass]

Not thinking today

[dkass]

Not thinking today

Hmm, it appeared twice, sorry.

[Daemonlaud]

If I may ask - doesn't mephisto have no interaction with MF in 1.10?

I heard that MF on Meph works at 0% always...

[adeyke]

MF does help for act bosses, both for the quest drop and for the non-quest drop.

[pmpch]

All of this is "Conspiracy Theory" in my opinion.


Also according to the very first post, Pindelskin has worse drop chances in 1.10. This isn't true, AFAIK. If TCs (treasureclasses) are linked, the highest Unique/Set/Rare/Magic values of all 'linked' rows will be the ones that are used, so for Pindelskin we would have:

X = Act 5 (H) Super C, Unique = 512, Set = 654, Rare = 972, Mag = 1024
Y = Act 5 (H) Uitem C, Unique = 800, Set = 800, Rare = 800, Mag = 1024

MAX(X, Y) = Unique = 800, Set = 800, Rare = 972, Mag = 1024 <= like any other SuperUnique

The funny thing here: the "Super A/B/C" TCs are effectifly useless :P

[Thrugg]

Yeah, I think we only realised that after this thread started. It's been going a while :)