Atlas mechanics / map drops: Is GGG lying to us? Deliberately not telling the truth?

"

Zrevnur wrote:

We do not actually need all those details. All we need is an "expected map drop spectrum for the average mapper" for all tiers and all "drop limit classes" (boss, rare/magic, normal, ...).
Example: In a tier 1 map I get on average 0.1 tier 2 maps and 0.5 tier 1 maps from rares and magics and other stuff with a +1 limit.
That would be sufficient for simulation.

Of course it isn't enough, lol. Finding the average for all T1 maps alone would take ages. Now when you go to T15 maps, where there is tons of other tiers to drop, you'll need far more.

"

Perq wrote:

"

Zrevnur wrote:

We do not actually need all those details. All we need is an "expected map drop spectrum for the average mapper" for all tiers and all "drop limit classes" (boss, rare/magic, normal, ...).
Example: In a tier 1 map I get on average 0.1 tier 2 maps and 0.5 tier 1 maps from rares and magics and other stuff with a +1 limit.
That would be sufficient for simulation.

Of course it isn't enough, lol.

Dont know what you mean with "it" here. Having such information for all tiers and all "drop limit classes" would be sufficient for simulation.

"

Perq wrote:

Finding the average for all T1 maps alone would take ages. Now when you go to T15 maps, where there is tons of other tiers to drop, you'll need far more.

For GGG it would be relatively easy to get such information. They probably have statistics based on simulation anyway.

For players it would be a lot harder.
There is a 'Path of Maps' I have "heard". I do not know the details. But it may actually be a viable approach for getting such data. Maps have an item level which corresponds somewhat with the "drop limit classes". This will probably not work for drops from rare monsters though. So some manual input is required to make it work. At least for drops from bosses and +2 strongboxes.
But the atlas makes this more difficult due to drop boni and drop tier limits - not unlocked tiers cannot drop unless adjacent.

"

Perq wrote:

But lets have fun here, as you seem to want it so bad. I don't know exact rules of drops, but if I were to guess, I'd say it works something like this:

-> A map drops. Check for Tier (including +X level from, say, strongboxes).
-> Check from what kind of monster it drops. If Magic/Rare add +1 to range, +2 for unique (strongboxes can drop whatever +X level they rolled). Roll the dice. Increase resulting tier by 1 (or by 25%, by 2), if map is not yet maximum roll possible.

No one here knows the exact rules for map drops, though we know from statistics that it is not a discrete uniform distribution (like a dice roll). Doesn't really matter, a discrete uniform distribution is good enough to illustrate that a 100% atlas bonus would have a massive impact.

Let us assume the following things:

1) You always run the highest map you own
2) You full clear all maps
3) Vaal orbs, strongboxes, double bosses - all minor details, we ignore them here
4) Map rolling is also ignored (or assumed that all maps are rolled, on average, the same)
5) On average, we assume you get 3.5 maps per map you run: 1 on average from normal mobs, 2 from magic/rare/exiles and so on, 0.5 from the boss
6) The maps are dropped in a discrete uniform distribution (as I said, we know that this is not true)
7) No map up-vendoring is considered


From these, we can easily simulate map drops, cause we know all we need to know. After simulating 100'000 map runs, the result looks like this:



As we can see, it is possible to maintain T8 maps in this scenario, though the tier that is ran the most is actually T9, and we very very occasionally run T15 maps - less than 1% of the time. The average map tier of maps ran is 9.65. And yeah, I know that maps were not that bad pre-2.4, but its just a model to illustrate a point.

Now, let us implement 100% Atlas bonus (i.e. all maps get upgraded by 1 tier if possible):



Because, as I said, map running is a recursive process, the average level of maps goes up by much more than one - in fact, it rises up to 13.22, 3.57 tiers higher on average than without the bonus. Now it is possible to maintain T11s, actually mostly T12s, and a lot of high tier maps are ran.


Now, you can criticize that a lot of that model is incorrect - but what the point is, is rather independent of the actual drop function: The Atlas bonus, with no other modifications to map drops, is MASSIVE. Like, very massive. It helps a shit ton. And yes, it does help, but not as much as I would expect, which is why I am trying to figure out whether other things were changed.

The fact that 100% Atlas bonus would help a lot more than just upgrade the average map by one tier is actually relatively independent of the actual drop function. The only way you could make it less impactful is if map drops were scarce to begin with, and almost always already max level, such that they cannot be upgraded. As you can probably tell from the approximately 287 Mud Geyser T3 maps you dropped pre-Atlas per week, this is not how map drops work, thus the Atlas bonus is in fact very strong.

"

Perq wrote:

And why high tier maps are so rare? Simple: they only drop on high level maps. And high level maps have far more map tiers that can drop. Maintaining at least T1 is easy, since every single map that drops, from any tier will always be at least T1. Maintaining T16 is hard, because on T16 map you can drop everything from T1 to T16.

True, but still map drops are biased towards same-tier maps. Honestly, T1 drops in T12+ maps were pretty rare pre-Atlas, and now they are impossible once you get 100% Atlas bonus.

"

Perq wrote:

This is why I doubt you managed to get a working model. Any wrong assumption that might have worked for non-Atlas situation can shift your results into complete shit... And everything seem they did.

Well, I am confident that my model is good enough to show one thing: The Atlas bonus should have a damn massive impact.

"

6) The maps are dropped in a discrete uniform distribution (as I said, we know that this is not true)

try simulating distribution like "relative weight of tier X drop is 1/X", or even 1/X^2. I'm curious to see the result.

"

silumit wrote:

"

6) The maps are dropped in a discrete uniform distribution (as I said, we know that this is not true)

try simulating distribution like "relative weight of tier X drop is 1/X", or even 1/X^2. I'm curious to see the result.

Can you be more specific?

Relative weight of T10 is 1/10th of T1? That is certainly not the case, rather the opposite is true. But I can simulate. Can you also give average drops from normal/magic/boss mobs so that I have all the required parameters? Happy to do another simulation or two, they are fast and easy to implement (just have to change the drop formula).

I think you are beating a dead horse here.

My guess is that for rolls on “if map should drop”, “which tier of map should drop”, etc. server uses same RNG seed as for rest of non-map drops. Therefore the reality of map drops, which you question, differs from raw DB settings and Atlas bonus, and is affected not only by those values, but also by other events unrelated to map drop mechanics, like having 2 boots dropped before map instead of 1 two-hander and 2 gloves.

So, I think GGG didn't lie about anything, they just don't have exact control over map drop reality. All they can do is collect sample data over some period, do statistics and then adjust some DB values here and there. And then again, to see actual effect, they can only collect more sample data after adjustment and compare the statistics.

"

Char1983 wrote:

Now, let us implement 100% Atlas bonus (i.e. all maps get upgraded by 1 tier if possible):



[...] The Atlas bonus should have a damn massive impact.

And it does. This one here is far more accurate to what I've experienced so far. No we all just run T15 nao kind of thing.
Still think this is not incredibly accurate, but still far more accurate than the previous one.



My pool fluctuates, of course, but I mainly run T12+. So yeah, good job on this one.
The only possible problem is that you assume (still) that this is not how things are atm. :P

map drops are insane

use invasion or beyond mod on non corrupted 30+pack t13+ maps with sacrifice fragments too
corrupt most of the t12+maps, insane rolls and sometimes +1tier

sell good layout maps, buy same tier maps, example, sell 10 gorge and buy 15 t13 for the same currency

be smart on which map to shape, i have shaped arachnid nest mainly because i can use 3 apprentice sextant to add mods to this T12 map, in this way you increase map drops, don't shape maps where you can't add apprentice sextant but maybe just 1 or 2 apprentice which are rarer

buy cheap corrupted maps (at the same price of white ones) people can't do or hate to do (scary mods/phys or ele reflect/temporal chains etc), you save currency to roll them


keep insane rolls and use journeyman and master sextant after the patch to increase the map drop even more.

i can't wait to see how many maps im going to drop with these+sextants

"

Perq wrote:

"

Char1983 wrote:

[...] The Atlas bonus should have a damn massive impact.

And it does. This one here is far more accurate to what I've experienced so far. No we all just run T15 nao kind of thing.
Still think this is not incredibly accurate, but still far more accurate than the previous one.

Nice, however, you are ignoring a few things:

1) The drop function I use is definitely wrong, we know that from map drop statistics (my own and other people's).

2) While the result may match your experience post-Atlas rather well, does it match your experience pre-Atlas? It would have to match both. And pre-Atlas, I could definitely sustain more than just T8 maps.

3) You are still not acknowledging that a 100% Atlas bonus has a much more massive effect than increasing the level of maps you run by 1 tier - or do you agree with that now?

4) This is for 100% Atlas bonus, I am at 121% at the moment.

5) I would still like to understand the underlying mechanics - and they still seem to be inconsistent with what GGG has said and what we experience.


I do start to get better map drops though, and the high tier maps do seem less gated than before. T16s are still pretty rare, but I am slowly getting there (also because I produced 3 Vaal Temple already - that was lucky I guess). Getting to Shaper without trading seems still rather unrealistic, mostly because I would have to drop some Cores first, which then in turn would have to drop the Phoenix map (which is unlikely cause I unlocked the others already).

That said, my map pool feels very solid now (45xT12, 37xT13, 14xT14, 10xT15, 8xT16). This is much better than ever before, and the game slowly starts to give me reasons to update my gear again (Guardians, Lair, and Dark Forest bosses specifically, the rest not so much).

I would still like to understand what is going on, mainly to gauge whether the last bit of Atlas bonus is useful or whether I can un-complete some shitty layouts and still be good.

"

Perq wrote:

My pool fluctuates, of course, but I mainly run T12+. So yeah, good job on this one.
The only possible problem is that you assume (still) that this is not how things are atm. :P

No, the problem is that we know the map drop function used is incorrect, and the results do not match the pre-Atlas situation, which is weird.

Clearly GGG didn't do a good job this time explaining how shit works , but at least some of the new bosses are fun .