Gradual Reduction in MF Effectiveness

[MadScientist]

You have to take failed sets and uniuqes into account, too. In the case of items without duarbility no difference can be seen (a set hydra bow will become a blue one that is indistinguishable from a normal blue hydra bow). When people complain about blue items in the quest drop they forget that these are failed set items.

The Mad Scientist

[whereagles]

The reason for people having the impression their mf is wearing out is psychological.

Because it was a novelty, people tend to remember the first times goodies dropped for them. After 100 mf runs, the goodies become a common occurrance and you forget them quickly. Therefore you have the impression meph is dropping jack. That's just it. It's the same sort of phenomena like when you were a child and the years seemed to go by soooo slow, whereas now they seem to go by at a very fast pace.

So... the "diminuishing mf effects" are just a very natural psychological reaction, which is normally called "superstition" :D

[Vash]

Whereeagles...we've heard enough people writing off our theories like children's superstitions. Cut us some slack all right?

And Meat...I thought they fixed that "first boss kill" cheese...am I to understand that it's still around on the realms...??

[vor_lord]

Looking at the game play part, there really is no differences. Looking at the code that drop items, there is no such check for single player, multiplayer or realm play.

What about ladder-only items? There has to be some kind of check what kind of game is being played in the item drop code.

[whereagles]

Vash:

So sorry, but I won't believe a word of the whole thing until I see some solid statistic data. Like in someone doing 1000+ meph runs, plotting the number of greens/yellows/golds per run, and putting up a graphic on that, showing some curve clearly sloping towards zero.

Until someone shows me that, I'll dismiss any "mf fading out" talk as pure lore.

[Jarulf]

What about ladder-only items?  There has to be some kind of check what kind of game is being played in the item drop code.

Ehh, I was not aware of specific item types only existing in ladder play. Or do you refer to cube recipies? There are some supposedly on ladder only. That is nothing strange since the game supports and uses two different cube data files, one for realms and one for non realms. The full code supporting both exists in the common d2common.dll file. The onlyy difference is that Blizzard doesn't ship that data file with the downloadable patch. However, all code and data for it is in there for full view and tha is the only data file were that is the case. SUpporting, by the way, that there is not a different set of data files for realms as in the case they DO have it (for cube), they handle it by different data files, both fully supported in the code, not just swapping files with identical names and such, creating possible messes, confusion and errors.

EDIT: Ohh, I noted the unique items in ladder only. Have not checked the unique picking code in 1.10 though. However, the uniqueitems.txt data file that list unique items, has a column specifying if the unique is a ladder only item and when picking uniques, it would compare that entry of the data file with the flag indicating if the current game is a ladder one or not. Nothing strange here, no special files for ladder only. Remember that as a server, the game runs ALL types of games at the same time. A server can run both hard code and non hard core games. It can run ladder and non ladder games and so on.

[Raziel]

So, where's the soj sale handling and dclone spawn code in the client?

[adeyke]

There's a ladder column in both cubemain.txt and uniqueitems.txt . Any recipe/unique with a 1 in this column can be used/dropped in ladder, SP, TCP/IP and open games, but not non-ladder realms games.

Does this mean that column isn't actually used but that there's instead a special cubemain.txt and uniqueitems.txt file for non-ladder realms games that has all the ladder-only lines removed?

[Jarulf]

So, where's the soj sale handling and dclone spawn code in the client?

The client doesn't handle selling, the server do. The client would simply tell the server that a player want to sell an item.

[Jarulf]

Does this mean that column isn't actually used but that there's instead a special cubemain.txt and uniqueitems.txt file for non-ladder realms games that has all the ladder-only lines removed?

No, only one cube and unique data file is used in all cases (well, on realms a realm cube data file is used, but for both ladder and non ladder). The game simply chack game type and if the unique (or recipie) is of ladder type. For unique selection, this is the test:

:6FC4CDC0 8B4C2414 mov ecx, dword ptr [esp+14]
:6FC4CDC4 80796A00 cmp byte ptr [ecx+6A], 00
:6FC4CDC8 750B jne 6FC4CDD5
:6FC4CDCA 83797400 cmp dword ptr [ecx+74], 00000000
:6FC4CDCE 7505 jne 6FC4CDD5
:6FC4CDD0 85420C test dword ptr [edx+0C], eax
:6FC4CDD3 7552 jne 6FC4CE27

First test is the game data records info on ladder type. Next the game check for LoD and finally if nessecary to see if the unique is dropable in ladder, otherwise it is skiped.

[HunkyDory]

Well very interesting all and some nice theories to think about.


Question was is MF fading?

Answer is no the game does not remember what you have found or how many times you have found it.

The code to do this would be too cumbersome and more important would require loads of memory and hard drive storage.

You can trust Jarulf on this one.


However...... (there is always a however not?)

There is a condition which effects Magic Find and that has to do (I suspect) with the way the games are initialized.


If you do many games consecutively with a short period of time between them you will notice that MF seems to fade away.


Two possibilities here :

Intentional to discourage the botters.

Unintentional as in something does not get initialized properly.


This has been around since I don't know what version.


So if youre into that Magic Find business take more time between games and the fading effect should go away.


Anything else you notice about Magic Find can be explained by reading a good book on statistics.



Regards, Hunky

[Taem]

And Meat...I thought they fixed that "first boss kill" cheese...am I to understand that it's still around on the realms...??

I thought so too, however I've been killing Baal in Nightmare with ZERO MF and, so long as I haven't kill him yet, I almost* always get a unique and a set the first 10 times I kill Baal.

* almost because there have been occassions where I do get all blue, but I can't be sure if it was because I killed Baal, or my friend did when we MF togeather.

[JustAGuy]

Ah, Hunky, good call. I have indeed played games in which uniques seem to fall from the sky, and others where nothing but blue and yellow come up. I agree that it would probably not be in Blizzard's best interest to have extra programs running that eat up for valuable server CPU time or disk space.

Sounds like games are seeded with some kind of number that is used as a base for MF calculations. If Blizzard added this sort of thing, and made it relative to the time between when you last made or joined a game, this would effectively reduce drop quality for those who keep at it. It could possibly be a time-stamp subtraction thing, very quick, very easy, very effective.

Since Jarulf can apparently read unlabelled assembler code (I can barely WRITE it!), perhaps he knows if there is any seeding happening at the start of a game?

As for reading a book on statistics... Good lord, no! I'll stick with my paranoid delusions and conspiracy theories, thanks. :)

EDIT: That wasn't the submit post button....

[Samka]

It wouldn't actually require much stored information to represent the fact that a character was either killing the same creature (type) all the time, or running the same level(s) a great deal.

Hunky's theory could be correct, as I've often found that the more efficient I get at running, the less I gain from it. But personally I feel that it has a lot to do with the areas that one runs in, rather than the time taken.

In any case, I'm pretty sure that something's happening, as people tend to notice patterns quite quickly. Take this post at the Blizzard forums as an example (and I just went there and clicked through about 4 pages - god knows how many similar queries have been made) :

Ive got 554 mf and ive done exactly 54 meph runes today u mighta seen me in Us West Renee is Hott1-54 yap tat was me and ive found $!@% well 2 nagel rings a cerebus but tats it where should i mf do i have enough mf and can i have to much cuz i heard i can plz help

[Occhidiangela]

What exist is a client component of the game . . . Then there is a server part, which runs the actual game.   Now, when you play on realms, the realms obviously runs the server, and you run a client on your computer (each player do).

Of course, nothing would prevent Blizzard from actually DO having different server programs, but that makes little sense, why in effect programming and maintaining several different “games” when they could all be the same or at least extremely similar. 

Jarulf, while I completely agree with you that Blizzard would have to do more work to write and run a modified version of the server code that we have on our PC's on all Realms Servers (they would all have to run the same "mod" so to speak) Blizzard could for their own reasons add a level of difficulty to their configuration control of Diablo II-Server if they chose to. It would boil down to being a sort of Mod, put on all of their Realms Servers.

I have played a few mods on private realms, and as a player, you just play it and some of the things are different.

Without adding the subroutines and code changes that some people think are causing a "decaying MF value," Blizzard could assign different values to any part of the game code(at the risk of goofing things up) change flags, change numerical values, or do any number of other things to make the Realm Game slightly different, and the Realms players might never notice it. For example, the Realm version could have a MF = MF/3 operation inside how it decides the final MF value added and we would never know it since it would not be readable on our Diablo II "server part" that we can read from our copy.

Blizzard would in that case be running a Mod on all of the Realms Servers (that means that they would have to load all of their servers with a slightly different "patch" of the 1.10 that we can call 1.10R) and it would probably be transparent to us. Players would not be able to tell the difference unless a striking difference (such as all Flayers are all of a sudden Physical Resistant at 100%) was noticed and someone checked playing it in Single Player and found a different value at the same difficulty.

Unless someone reverse engineered the game code that is running on the servers, downloading all the files, and then matching them file by file against what was on one of the copies of the game from the CD + the download of the 1.10 patch from the Blizzard FTP, we could not confirm a difference.

They already duplicate some code between the client and server component, and that has in the past (and probably still) lead to several bugs where the code have not been similar enough and the information displayed on the client have not matched the actual server data. On top of that, it would be VERY confusing for players, playing a game that would differ depending on what type of game you player, even more so if the differences are small, subtle and unknown. That would be horrible.

Horrible? If all Realm Servers were running the same Patch, the 1.10R I suggest up there, then all Realm players would be playing the same game, but a slightly different one from Open or Single Player 1.10. Since you point out that they often patch the servers without sending the players a patch, it is at least possible that in the process of that patch, a subtle change to the server part of the game can be changed to "1.10R." As a Realms player, you could be playing the 1.10R Mod right now and not know it.

Why they would want to do that with all of the bugs they have not yet fixed would be beyond me, and I agree with you that it makes little sense.

So no, there are no such run “counters”, nor any special MF dimishing effects in the game, not in single player, not on realms.

I wish I could agree that your point is irrefutable: your assertion can be proven for the game we all have on our PC's. You and others have looked at it. But your observation regarding "not on realms" cannot be proven (unless one of your friends at Blizzard has confirmed that to you) even though I think you are right. ;)

[Raziel]

The reason why I asked about the SoJ selling and DClone spawn is that it is an example of a difference between client and realm server code. I don't think the code for what happens when you sell a SoJ is anywhere on the client. The message and the Diablo Walks The Earth thing is there, though. But the guts of it isn't.. right?

My point is.. blizzard is already running with slight differences server-side. Why can we not assume there is more than the obvious?

But, like Occhidiangela said, if there's not some difference we can measure and compare against client files, we'll never really know short of hacking in and downloading what's on the machine itself.

[Jarulf]

Jarulf, while I completely agree with you that Blizzard would have to do more work to write and run a modified version of the server code that we have on our PC's on all Realms Servers (they would all have to run the same "mod" so to speak) Blizzard could for their own reasons add a level of difficulty to their configuration control of Diablo II-Server if they chose to.  It would boil down to being a sort of Mod, put on all of their Realms Servers.

I skiped the rest of your message since I think this part is enough for my reply. Anyway, yes, the can of course have a completely different system running on the realms. I know for a fact that it has not been the case before, mainly from asking the programmers (among other things) and if they DID want to have different code for realm and non realm (easy enough with compile switches and so on), why did they not remove for example ALL code and even complete functions (some exported dll functions) that are only ever used on the realms? They have not. And I also tend to believe the programmers. I can think of some added extra stuff to fight duplication of items and some other small cheat preventions that might have been added realm side only in late patches. I have no idea of 1.10 since I have barely looked at it and have not talked with anyone at Blizzard about it though.


To answer the thread in general:
I still stand by the fact that there is absolutley no sense, nor possibility that Blizzard would add something as "stupid" (yes, it is a stupid idea in my opinion, because it gives no enhanced game play effect nor add anything to the game) as a counter to slightly reduce the ammount of unique or set items in some specific, in part complicated to see, situations like killing the same monster many times in a row. There is simply no reason to do it at all, andif the effect is so small that people hardly notice it and even debate it and is unusre of it and want to do tons of runs to see a small pattern, then iteven have no effect.

And no, there is no such seeding at the start (there are tons of other seedings of course, and I have not looke that much at game initialization code to know it all, but I have, and any others, looked at the item generation code (both in the server part and the client part, which at least in old versions was a duplicate and thus not changable on realms only really) and there is NO such thing at all. NONE.

[JustAGuy]

Wow, somebody's gettin' all riled up. Ok, now some words regarding the "stupid" issue at hand.

The realms version vs. the single player version: definitely not identical, probably mostly minor differences, i.e. bug fixes. We know for a fact that they are different, because Blizzard has stated that they made some realms-only patches for a couple of things. To what extent, we don't know, but according to you that's immaterial.

The effectiveness of MF does not reduce after multiple runs of the same monster.

What was Blizzard's aim with Diablo 2 v1.10? To make things harder. Things got a lot harder, if only temporarily. It's actually hardest when you have crappy items. Blizzard has made bosses drop well the first time, and poorly afterward (but still good, in most cases). Blizzard has made Pindleskin (arguably the easiest to boss to reach) drop good things less frequently. Blizzard made the Durance level 2 massive in order to slow Mephisto runners (I assume, there isn't any other reason for it). Blizzard made the tower bigger, presumably to slow Countess runners.

Now, what breaks this nice little model Blizzard has created? The items make the game fun, the items make the game unbalanced, the items make the game easier; having easy access to the all the great items with very little effort and almost no risk shatters it. Do a few runs, and you are guaranteed to get some really great, if not the best, items in the game. What about all those other schmoes that bust their humps with only a sharp short sword of the leech and a sturdy leather armour to take Hell on with?

Is it so unbelieveable that Blizzard would add one more thing to the already long list of changes to make the game that much harder? Less good items = more difficulty. Items are there for incentive, and eventually you will find them. In the mean time, you have to suffer? No -- people run for it. But running isn't what Blizzard intended, otherwise they would not have decreased the drop after 1st kill, otherwise they would not have made the durance and some other levels bigger, otherwise they would not have MADE THE GAME GIVE YOU CRAP IF YOU KEEP RUNNING. Your MF doesn't give you crap, the game does by way of a mechanism we don't know about.

So yes, some of what we as runners are experiencing is psychological. I concede that one. But that's not to say that it isn't happening. Technically, there is no degredation of MF, but I highly doubt that these results are simply statistical. When drops have the perceived curve that we are experiencing, you know it isn't just the law of averages working itself out. There is very possibly a hidden mechanism causing this behaviour.

You mention that people hardly notice this problem. That's probably because the people who do notice it are the ones doing it the old fashioned way, and not using a "bot". People who use "bots" to run wouldn't notice anything, nor would they care, since they just let the bot go over night and, by average (even with diminished possibility of a good drop), they would have the loot they sought.

While doing a bunch of runs won't prove anything, I'm still going to do them. Why? Well, it gives me something to do, and I just might find those sweet items. I'm going to write down the results for sanity's sake, and if a clear pattern emerges, you'll be the first to know. ;)

You know, the Blizzard developers are quite talented. Diablo II is a very complicated game. They could have come up with something very ingenious to achieve this "midly annoying" feature. It's not necessarily someplace you've looked before. After all, you haven't really taken a long hard look at the game code this time around, so it's quite possible you could have missed it.

I see what you mean though; why spend time trying to prove or disprove something that, at this moment, is in your opinion, non-existant. My answer is, why not?

[Ruvanal]

What was Blizzard's aim with Diablo 2 v1.10? To make things harder. Things got a lot harder, if only temporarily. It's actually hardest when you have crappy items.

For dealing with the combat of monsters, yes they moved to making the game harder. But as far as general item generation goes they changed it so that it easier overall to get 'better' (unique, set, rare, magic) items than you were in the earlier patch due to how they changed the values for ALL checks in itemratio.txt. To give a rough idea these changes to the values, it makes it about 25% more likely you will get a unique item from ANY of the monster than you would have in the v1.09 game using the same amount of magic find (if you had say a 2% chance of the item being a unique in v1.09; it would be about a 2.5% chance in v1.10). This is points to their awareness that the item finding is noted as something that the players want and need, not something that they are really trying to discourge.

Blizzard has made bosses drop well the first time, and poorly afterward (but still good, in most cases).

This is not an accurate assessment of the quest to non-quest drops as they are listed in the game files. The non-quest drops in v1.10 are set up exactly the same as they are in the v1.09 version. So they cannot really be considered 'poor' ar all compared to how they were in v1.09. In fact due to the changes that I noted in the itemratio.txt file they will actually be somewhat better due the globally enhanced chances of getting uniques, sets and rares (these drops are already forced to magical quality as a minimum). It is the quest quality drop that has been added at an even better than normal chance of better quality of base items (no gold or 'junk' items will be picked) and the enhancements to the unique, set and rare* chances have been increased. (* Rares are the minimum quality that should be dropped by checks, but you may get some magic quality due to their being failed set checks.)

So it would be more accurate to say that the quest drops (supposedly first time drops) are 'uber' and then they drop down to the 'normal' good drops that used to get from the act end bosses.

Blizzard has made Pindleskin (arguably the easiest to boss to reach) drop good things less frequently.

Actually he drops just as well as any of the other boss monsters that are mlvl=86 using the "Act 5 (H) Super C" TC for their drops. Due to the way the game handles the drop checking, the "Act 5 (H) Super Cx" TC that is given to Pindleskin has exactly the same chance of dropping items (and quality) as the "Act 5 (H) Super C" TC that is given to many other boss monsters through out the act 5 hell area. Only his mlvl is slightly lower than you can find if you travel to some of the level 84 and 85 areas (giving you mlvl 87 and 88 bosses). Considering the ease of getting to him, he is probably still one of the better 'item farming' choices, even if he can't drop a handful of the uniques requiring the highest ilvls to spawn.

Blizzard made the Durance level 2 massive in order to slow Mephisto runners (I assume, there isn't any other reason for it). Blizzard made the tower bigger, presumably to slow Countess runners.

These cases were probably done as much for server stability than as to slow down the 'farming' of two of the "cheesier" monster for item farming (Mephisto for the bulk of unique and set items and the Countess for the runes). Having many very short duration games being created and destroyed on the servers (max of 128 per server) can be a bit of CPU and memory overhead to be allocating and then de-allocated. If enough of this is happening in a short amout of time you might even be able to generate fragmented memory in the RAM similar to what happens on hard drives that decreases their performance. True the modern set of computers are less susceptible to this than the older models that I first saw this happening on (a virus in an earlier type of computer intentionally did this to crash the system), but they are not immune if they do not have an opportunity to cleanup their memory allocation usage (done during lower activity levels typically). Doing things that discourage rapid game creation can solve most of this problem (like the realm down situations due to logging into and out of games too quickly).

So yes, some of what we as runners are experiencing is psychological. I concede that one. But that's not to say that it isn't happening. Technically, there is no degredation of MF, but I highly doubt that these results are simply statistical. When drops have the perceived curve that we are experiencing, you know it isn't just the law of averages working itself out. There is very possibly a hidden mechanism causing this behaviour.

I would say mostof it is psychological, since I have never seen one of these claims ever get supported by some solid statistical evidence. In fact the few cases that I have seen someone really gather solid statistical evidence; it ended up matching the derived values to 3 significant figures or better each time. All I have seen from your statements and other agreeing with you is that is some "percieved" thing that you seem to be noteing, which is a situation that can be very much influnced by the psychological factors that other such as whereagles or Professor Frink noted earlier.

Personaly I find that 99+% of people recollection is rather flawed even over only a few minutes unless they are doing something to accurately record what is going on the whole time (like carefully noting all the particulars of every item that a boss would drop for every run). In the past every time that I cornerd someone about having had impossible items drop from chests, they quickly recanted because in fact they had not really been making sure that what they faound had actually come from a chest in the first place. But they sure felt special when they were claiming something that went 'against the expert knowledge'. This is a decidedly psychological trap the most people actually like to fall into; yet more of the psychology problem of only having ones memory as the only real record keeping of an event. I'll take some hard evidence over ten eywitness any time, just look at what magicians do to the audience in performance to see why.

You know, the Blizzard developers are quite talented. Diablo II is a very complicated game. They could have come up with something very ingenious to achieve this "midly annoying" feature. It's not necessarily someplace you've looked before. After all, you haven't really taken a long hard look at the game code this time around, so it's quite possible you could have missed it.

Jarulf is not the only one that has been looking into the code. There are many over at Phrozen Keep that are doing this type of digging. As of yet I have not heard of single instance of their being any spot in the item generation routines that would allow the insertion of extra parameters that would do the types of effects that would be required to achieve the types of results that you claim are happening. On the other hand, using the Occams Razor principle, the easier explanation is that of a psychological answer. What you are percieving is just that; a perception sitiuation, not a real effect.

[Samka]

Well, the effect *could* be psychological. But if so, then we would have seen the same psychological effect prior to 1.10. In fact it would be more pronounced, because MF was apparently lower then. Yet this is not the case. I and many others have only noticed an effect in 1.10.

This kind of thing is very hard to test, because by the time it is noticed, the MF reduction (if there is one) is already in effect. If I show a string of improbably low drops, then you can simply say that it was just bad luck. It would be exceedingly arduous to manually gather enough data to avoid this rebuttal.