I decided to do a little magazine testing

[Pain Weasel]

So I read something about skill points only increasing drop rate of the related magazines by 5% 2% (Thanks Cookie). That didn't really match my experience with A21 so I figured I'd do some testing to see how much skills impact the drop rate over a large number of loot boxes.

There may be better posts about this and I'm sure there are some videos somewhere, but I felt like testing it myself and I figured I may as well share my results.

Forge Ahead was used as the test magazine and Working Stiff crates as the test loot container. Forge Ahead is in the loot pool for that container and it should have a high drop chance by default.

A new game was created (no mods, default difficulty, default map, Pine Forest) and 500 Working Stiff Sealed Shipping Crates were taken from the creative menu. 100 crates were placed, opened and looted, then the magazine loot was saved in a marked box, and a skill point put into Advanced Engineering. The process repeated for each scenario.

4 scenarios were tested: 0, 1, 2 and 3 skill points in Adv Eng. I didn't go up to 4 skill or test more crates because I have things to do today although ideally I would have tested larger batches of crates.

The other magazines looted during the test were roughly similar in number for all 4 tests (Handy Land found for each test: 23,31,30,31), which is what I would expect without any skill points, so I feel like I had a decent sample size. More would have been better but I just didn't feel like it.

Results:

0 Adv Eng- 13 Forge Ahead out of 100 crates.
1 Adv Eng- 29 Forge Ahead out of 100 crates.
2 Adv Eng- 36 Forge Ahead out of 100 crates.
3 Adv Eng- 48 Forge Ahead out of 100 crates.

On paper that really looks worth the skill investment with each point resulting in a significant increase to the base chance (maybe with a bit of diminishing returns) and with a large enough sample size it probably works out to be a great way to increase drop chance. Granted Working Stiff crates were specifically chosen for the high probability of finding a Forge Ahead mag so the drop chance is inflated, but there should be similar increases for each point spent when looting other containers.

3 of the 4 tests had a very even distribution of Forge Ahead magazines looted, however in the scenario with 1 point in Adv Eng the first 30 crates opened had 0 Forge Ahead magazines and all 29 of the magazines were found in the last 70 crates. Over time it averages out but I can absolutely see how people are complaining about not finding any magazines even with points, while others have the opposite experience. RNG can result in pretty bad streak of luck. I personally have never seen that bad streak other than during this test, so I'm sorry for doubting any of you who claimed a really long run of bad luck.


If you see any issues with my methodology let me know, otherwise do with this what you will.

[Nocternaldaze]

Dont worry it easy too think it is bad rng and it can not be overcome by taking steps ,things like, Seed used for game, not putting points into the right options and not looking in the right places. These are all things players need to think about but is often overlooked.

[Pain Weasel]

Originally posted by Crimzn:

curious question why don't you just read the config files?

Honestly, because I don't know how all the various modifiers interact and I don't feel like taking the time to learn. Sure, I can open progression.xml and find the increased probability chance from Adv Eng, but that just tells me how much it increases the base chance, which I am not sure of. I could hunt down the base chances by loot container for the specific item group, and the chances of that item in the group, factoring in having multiple item picks per container and blah blah blah.

Or I could drop 100 boxes in creative, shoot them open with the dev super digger, loot them and see how many magazines end up in my inventory.

Elegant? No.

But it was quick, easy and I didn't have to worry about missing something buried in one xml file or another and making a false assumption based on incomplete information. I mean TFP already made the calculation, I figured it was easier to see the result in the game than to try to recreate it exactly myself.

[thestile]

Oh, wow, that's a really useful test. Thanks for sharing that. The points definitely carry a little more weight than I thought when it comes to receiving the magazine.

Well, I'm gonna go loot some Working Stiffs to try and get my chemistry station. And I might be really lucky or the polar opposite. haha

[Hellebron]

It's not only books. I have three points in Pummel Pete and am overflowing with steel club parts.

[Saigoth_74]

just search A21 Loot groups on YouTube, someone made a video to show their research into the files themselves. It has a link to his sheets as well for reference. Its a really solid listing of all loot-able objects and which mags they can drop etc. Breaks it all down. It works a lot different isn't pure RNG either.

Shorthand of it, is mag's don't drop vs loot stage really. They drop vs the object your looting and a table predefined for drops. If said object has 4 mags for instance, and you have no perks it will first roll to see what items will be in the object from its primary drop list. If mag's are selected on that roll, then it will roll on a RNG for those available to that object. Things like mailboxes for instance can have almost every mag available, making them a source but not a good source for getting specific mags. Where as like working stiffs crates only have a few mags available, making them a far more reliable source to get Forge, repair tool and salvage tool mags.

It weights the second roll vs perks you have learned. Meaning if you have lockpicking and engineering for instance and also salvage operations, you'll at min have % chance higher for those. Believe its up to 10% based on perk points spent. Then once you have all the mags for one group it removes the bonus and leaves only the original base chance. Giving better chances for others.

Only draw back is when you invest in multiple perks that offer mag increase and loot from containers that have more drop types for mags. Cause it seems some mage types have higher base chance to spawn, even if isn't coded that way. But overall, loot across the board beyond mags is way better drop wise with how they changed the loot system in A21. Also makes a little risk vs reward go a long way. Raiding tier 3+ in all biomes is a noticeable increase in loot, specially when you hit the harder ones like snow and wasteland. Even sneaky dip into the edge of wasteland early on can net you a huge boon in drops more so that what I experienced in previous alphas.

[Cookie]

Originally posted by Saigoth_74:

just search A21 Loot groups on YouTube, someone made a video to show their research into the files themselves. It has a link to his sheets as well for reference. Its a really solid listing of all loot-able objects and which mags they can drop etc. Breaks it all down. It works a lot different isn't pure RNG either.

Shorthand of it, is mag's don't drop vs loot stage really. They drop vs the object your looting and a table predefined for drops. If said object has 4 mags for instance, and you have no perks it will first roll to see what items will be in the object from its primary drop list. If mag's are selected on that roll, then it will roll on a RNG for those available to that object. Things like mailboxes for instance can have almost every mag available, making them a source but not a good source for getting specific mags. Where as like working stiffs crates only have a few mags available, making them a far more reliable source to get Forge, repair tool and salvage tool mags.

It weights the second roll vs perks you have learned. Meaning if you have lockpicking and engineering for instance and also salvage operations, you'll at min have % chance higher for those. Believe its up to 10% based on perk points spent. Then once you have all the mags for one group it removes the bonus and leaves only the original base chance. Giving better chances for others.

Only draw back is when you invest in multiple perks that offer mag increase and loot from containers that have more drop types for mags. Cause it seems some mage types have higher base chance to spawn, even if isn't coded that way. But overall, loot across the board beyond mags is way better drop wise with how they changed the loot system in A21. Also makes a little risk vs reward go a long way. Raiding tier 3+ in all biomes is a noticeable increase in loot, specially when you hit the harder ones like snow and wasteland. Even sneaky dip into the edge of wasteland early on can net you a huge boon in drops more so that what I experienced in previous alphas.

I believe that's what he was already referencing in a conversation we'd had in another thread (though some numbers appear to be off).

The most interesting thing I found, when looking into the progression files, is that each skill related to a crafting magazine has a passive effect. It reads something like this:

Operation="perc_added" Level="1,5" Value="2,10"

I'm away from my pc now so it may not be exact, but that's where I'd like to think that Pseudo Posse got the 2% number from for each level. However, when comparing other passive effects that use the perc_added operation, it usually includes numbers like 0.2 to mean 20%. So a flat 2 would mean 200% in regards to the magazine finding bonus for level 1, 400% increase for level 2, etc.

But I haven't experienced that. At least not after the first level into a perk, which usually DOES yield double in whatever magazine I'm aiming for. In my few tests, I've noticed that after putting 1 point into a perk, further points provide diminishing returns like OP observed. I'm not sure what factor I'm missing, but I'm missing something.

[Pain Weasel]

Originally posted by Cookie:

I believe that's what he was already referencing in a conversation we'd had in another thread (though some numbers appear to be off).

The most interesting thing I found, when looking into the progression files, is that each skill related to a crafting magazine has a passive effect. It reads something like this:

Operation="perc_added" Level="1,5" Value="2,10"

I'm away from my pc now so it may not be exact, but that's where I'd like to think that Pseudo Posse got the 2% number from for each level. However, when comparing other passive effects that use the perc_added operation, it usually includes numbers like 0.2 to mean 20%. So a flat 2 would mean 200% in regards to the magazine finding bonus for level 1, 400% increase for level 2, etc.

But I haven't experienced that. At least not after the first level into a perk, which usually DOES yield double in whatever magazine I'm aiming for. In my few tests, I've noticed that after putting 1 point into a perk, further points provide diminishing returns like OP observed. I'm not sure what factor I'm missing, but I'm missing something.

Thank you Cookie! That is what we were talking about. I must have misremembered that 2% as 5% for some reason.

[Dreadstone]

My first playthrough I deliberately put zero points in any skills to see which magazine would complete first. Results?

1) Seeds (Obviously, you only need 20)
2) Electrician
3) Traps

I did only a handful of quests, didn't buy any books from the trader (Good ol Rekt in this case), and looted the F out of everything (Mailboxes, forges, rooftop containers)

[I can't spell]

dear lord someone is a masochist 😳

[Cookie]

I FIGURED IT OUT.

It ISN'T a 2% increase, it's a 200% increase per level. 200% for level 1, 400 for level 2, etc.

I tested on some savage country crates. I adjusted the xml so that each crate would drop one mag each (and no other loot), and each book had the same drop chance (high). There are 5 mags that can be found in that crate, so any given mag has a 1/5 chance of dropping. I tested on 250 crates total and got almost exactly that. Each mag had around 50 drops.

I then tested another 250 with one level into archery. I got exactly 100 bow mags. Instead of a 1/5 chance, it was now essentially a 2/6 chance (2 chances for bow mag, 1 chance for each of the other 4). I got lucky with this run and ended up a little higher than I should have. I really only should have gotten about 85.

So then I tested another 250 with 2 levels of archery, meaning a 400% increase and got 131 mags. This time around my chances were 4/8 (4 chances for bow, 1 chance for each of the other 4), so I ALSO ended up a little lucky with this run.

My last test was another 250 with 3 levels of archery, which should mean a 600% increase in probability. Following the same pattern of progression, I now realized that bow mags had a 6/10 total chance. Out of 250 crates, I SHOULD get 150 bow mags. I ended up right at 153.

So now I can surmise that if I had done another 250 at level 4, I'd have an 8/12 chance. This means out of 250 crates, I would have expected about 167 bow mags.

At level 5 with a 1000% increase, my chance would be 10/14. Out of 250 crates, I would expect about 178 bow mags.

The diminishing returns exist because the chance is still weighted against 4 other loot chances. In a Pop-N-Pill crate, which only includes 2 magazines (medical and blades), putting 5 points into Physician for a 10x chance means that your chances would be 10/11 for a medical mag. But in something like a crack a book crate, which has a chance at EVERY magazine, your odds start at 1/23. If you invest 5 points into a perk for a 10x multiplier, your chance is still only at 10/32.

[Pain Weasel]

Originally posted by Cookie:

The diminishing returns exist because the chance is still weighted against 4 other loot chances. In a Pop-N-Pill crate, which only includes 2 magazines (medical and blades), putting 5 points into Physician for a 10x chance means that your chances would be 10/11 for a medical mag. But in something like a crack a book crate, which has a chance at EVERY magazine, your odds start at 1/23. If you invest 5 points into a perk for a 10x multiplier, your chance is still only at 10/32.

Well done. I'm glad you had the patience to run through 250 boxes per test because I sure didn't.

I think we can say two things with relative certainty at this point:

1) The increase to loot chance for each skill point is much higher than previously reported
2) The type of container and it's loot pool is as important, and in some cases almost more important, than the skill.

[Cookie]

Originally posted by Pain Weasel:

Well done. I'm glad you had the patience to run through 250 boxes per test because I sure didn't.

I think we can say two things with relative certainty at this point:

1) The increase to loot chance for each skill point is much higher than previously reported
2) The type of container and it's loot pool is as important, and in some cases almost more important, than the skill.

1) Absolutely true
2) I would say the container is the most important piece in determining the probability of a magazine dropping.

Here's the base xml for the savage country crate that includes magazines:

 
<lootgroup name="groupSavageCountryCrate01">
	
	<item name="armorSkillMagazine" loot_prob_template="high"/>
	<item name="bowsSkillMagazine" loot_prob_template="high"/>
	<item name="spearsSkillMagazine" loot_prob_template="high"/>
	<item name="bladesSkillMagazine" loot_prob_template="high"/>
	<item name="clubsSkillMagazine" loot_prob_template="high"/>

	<item group="groupArmorScaled" loot_prob_template="low"/>
	<item group="groupArmorModsScaled" loot_prob_template="medLow"/>
	<item group="groupApparelClothes" loot_prob_template="low"/>
	<item group="groupApparelShoesAll" loot_prob_template="low"/>
	<item group="groupApparelFaceCover" loot_prob_template="veryLow"/>
	<item name="gunBowT1WoodenBowParts" count="1,4" loot_prob_template="medLow"/>
</lootgroup>

For reference, veryLow = .05, low = 0.2, medLow = 0.35, and high = 0.75.

So each magazine has a weight of 75, totaling 375. Armor has a weight of 20, Armor Mods 35, Clothes 20, Shoes 20, Face Cover 5, and Bow Parts 35. In total that's 510. At base value, a specific magazine has a 75/510 (simplified to 5/34, or a 15% chance). At level 2 in a skill with a 400% modifier, for example, you bump that to 300/735 (simplified to 20/49, or a 41% chance) for the magazine associated to it (in my example it would've been bows). So 2 out of every 5 boxes would likely contain that mag. That still seems like pretty low odds in my opinion. Even at max level, your chance would be 750/1185 (50/79, a 63% chance).

Now I understand that there are other sources for magazines outside of these crates specifically, but the values for those sources are usually low or very low. That's why I edited my xml's so that each crate would have a dedicated loot group for it's related magazines. It's now the one avenue of magazines for me that isn't entirely dominated by RNG. I will always find a magazine in the crate, but the RNG revolves purely around WHICH magazine I might get. And because there aren't other items within the loot pool, the probability of finding one depends entirely on what skills I've chosen to invest in.

[lexgeo]

cement mixers and broken workbenches seem to be giving a forge mag most of the time for me. maybe if you have some free time and are willing ? maybe some test results on those?
thanks for that info.

[Cookie]

Originally posted by lexgeo:

cement mixers and broken workbenches seem to be giving a forge mag most of the time for me. maybe if you have some free time and are willing ? maybe some test results on those?
thanks for that info.

Workbenches seem to have the workstation magazine listed in groupWorkbenchLoot04, which has a low (20%) chance of appearing. That's shown here:

 
<lootgroup name="groupWorkbenchLoot04">
	<item group="groupToolsTiered"/>
	<item group="groupRareToolsTiered"/>
	<item group="vehiclePartsTiered"/>
    <item name="workstationSkillMagazine" loot_prob_template="high"/> 
</lootgroup>

<lootgroup name="groupWorkbenchLoot" count="all">
	<item group="groupWorkbenchLoot01" count="1"/>
	<item group="groupWorkbenchLoot02" loot_prob_template="medLow" force_prob="true"/>
	<item group="groupWorkbenchLoot03" loot_prob_template="low" force_prob="true"/>
    <item group="groupWorkbenchLoot04" loot_prob_template="low" force_prob="true"/> 

So even though the workstation skill magazine has a high chance of appearing, that's ONLY if the lootgroup it's contained in is generated, which is a 1/5 chance.

The forge overall has a high (75%) chance of producing a workstation mag, as it's contained outside of the other existing loot groups on its own. As seen here:

<lootgroup name="groupForge" count="all">
	<item group="groupForge01" count="1"/>
	<item group="groupForge02" loot_prob_template="med" force_prob="true"/>
	<item group="workstationToolsScaled" loot_prob_template="veryLow" force_prob="true"/>
	<item name="workstationSkillMagazine" loot_prob_template="high"/> 

This is also true for the cement mixer.

Chemistry station is the same as the workbench in terms of probability. Also true for the dew collector.

I would assume that, at least for the forge and cement mixer, a single point into either Advanced Engineering or Lockpicking would guarantee a magazine every single time since it's not being weighed against the odds of any other items.