How do you compare the ratchet wrench to the plumbers wrench?

[ElCabong]

I have an orange ratchet but I just looted a blue plumbers. The block damage and stamina cost is better for the plumbers wrench but what I'm more interested in how much I get out of salvage. Do I salvage more with the wrench that does more block damage? Is it just quicker but I get more salvage from the ratchet because it's a tier 2 tool?

[meganoth]

Never tested it, but as far as I know the only thing influencing the salvage amount from a specific block is your perk.

 

This is similar to mining where block damage (boosted by Miner69er) only increases your mining speed while mother lode increases what you get out of each block. The difference in mining is that since there is no limit on blocks to mine and you can mine more blocks per minute with miner69er, even a speed increase helps.

[ElCabong]

I am getting more steel bars when I break down light posts. I am getting three bars frequently and no less than two whereas with the ratchet I would get two or one and occasionally none.

[meganoth]

Ok, that would surprise me. Are you sure this isn't just random or that you leveled up the bad mechanic perk inbetween?

 

[ElCabong]

I put one point in salvage operations very early on. That is it. I might have added one more level since the last time I used the ratchet but I put the perk point into something in the agility tree.

[Morloc]

I have no idea how the algorithms play out, but I always wondered when breaking down say...a car, if I was losing out on chances for a battery with a faster tool. I convinced myself it was just perception, but would love for a dev to weigh in on this.

 

 

-Morloc

[meganoth]

Ok. I just checked the wrench and ratchet in items.xml and at least I don't see any indication they have differences in harvest amount. Actually there is a parameter there called "HarvestCount" that is set to "1" for all of the harvest tools.

 

If you want to get to the bottom of this, use something harvestable that you can find in masses, for example toilets or shopping carts (cars are suboptimal because they give too many different resources, hard to compare).

Empty your inventory, then harvest lets say 10 of them with wrench. Write down what you get.

Do the same with ratchet. Compare the two amounts.

 

Then post your result.

 

[Liesel Weppen]

Erm, i tested this with an older version (probably A17?) with mining and chopping wood and there it could happen that a weaker tool or lower skills result in more ressources.

 

What you get out of an item afaik is calculated per hit. So if you need more hits because less blockdamage of your tool, you CAN get more. But that only happened in some border cases. Most times it works as expected giving not less ressources with a better tool.

 

That was also discussed back then.

[theFlu]

If I were to test this, I would just spawn the things in from creative menu. Comes with a risk of using a different block than what is in POIs, but at least you can get a decent sample size - getting a 20% difference from 10 toilets may still be hidden in the randomness.

25 minutes ago, meganoth said:

cars are suboptimal because they give too many different resources, hard to compare

Maybe, but it seems to me that the car salvage drops are not dependent of each other (you can get a battery AND an engine for example); if so, there would just be a larger number of results from the same effort (each item type would be its own data set)

[Canute]

31 minutes ago, Liesel Weppen said:

Erm, i tested this with an older version (probably A17?) with mining and chopping wood and there it could happen that a weaker tool or lower skills result in more ressources.

That was a known issue at older version.

Special when you butcher animals you notice you got much less meat out of them with machete then with knifes.

But that should been fixed allready.

 

[meganoth]

1 hour ago, theFlu said:

If I were to test this, I would just spawn the things in from creative menu. Comes with a risk of using a different block than what is in POIs, but at least you can get a decent sample size - getting a 20% difference from 10 toilets may still be hidden in the randomness.

 

I wasn't sure if wrenchable *blocks* could be placed from creative menue so I kept it simple and the numbers low. But probably even the player-craftable lamps should be harvestable, then this would be ideal. I would guess a simple lamp might be even better than toilets.

 

Quote

Maybe, but it seems to me that the car salvage drops are not dependent of each other (you can get a battery AND an engine for example); if so, there would just be a larger number of results from the same effort (each item type would be its own data set)

 

No, I already experienced this problem in a similar case.

 

How can you compare

a) 430 gas, 13 forged iron, 2 headlamps, an engine and 3 batteries

with

b) 490 gas, 8 forged iron, 2 engines,  2 batteries and 2 forged steel

 

You can't unless you know something about their drop chances and amounts from studying the xml. Once you sample say 100 cars you would probably get enough parts in each category to have a clear indication anyway, but the less salvageable items, the less samples he needs to see a pattern with some certainty in it.

 

If harvesting works similar to looting (which I assume) you might get say 12 "draws" from the car. if 7 of those draws are for forged iron you get less draws for other things like forged steel or headlamps and vice versa. But if you get forged iron you will get a higher number of iron per draw than you would get steel. (if cars don't give steel at all, substitute it with something else, the argument stays the same)

 

 

 

[theFlu]

1 hour ago, meganoth said:

If harvesting works similar to looting (which I assume)

There's where we differ; I'm assuming the salvage system is different. My reason for that is mainly engines, that drop from the exact middle of the car durability - ie, whichever swing hits that 150/300 point has a chance. (This may have changed without me noticing) That suggests that the "middle swing" is special, and I would assume the others as well. Also you can get several resources with one swing, and there seems to be some patterns there as well, but nothing I'd have recorded properly. But to me it seems it's more like harvesting the old boulders where you'd get a random amount of everything.

 

And I did mention, admittedly in an unorthodox wording, that the drops would need to be independent variables. IF that's the case, then your a) and b) would be compared gas to gas, forged iron to forged iron etc. And while 430 to 490 and 13 to 8 look quite different, that's easily within the randomness with that small a sample. I wouldn't predict percentages until I have at least 100 results, as in, 100 headlamps from each set.

 

And yes, I'm guessing a whole lot here; with all the conditionals there I shouldn't need to stress this, but I will anyway...

[meganoth]

3 hours ago, theFlu said:

There's where we differ; I'm assuming the salvage system is different. My reason for that is mainly engines, that drop from the exact middle of the car durability - ie, whichever swing hits that 150/300 point has a chance. (This may have changed without me noticing) That suggests that the "middle swing" is special, and I would assume the others as well.

 

As long as we don't know which theory is correct we can't get any conclusions from a small sample when the worst case theory means that it won't work with a small sample.

 

That you get the contents in bits and pieces doesn't mean that the whole contents isn't just a loot container. And the syntax in blocks.xml is almost identical to a loot container except for the missing framing xml specifying the container and for additional tags that differentiate between wrenching and other harvesting or damage types.

 

To me the xml looks like the harvesting amount of a car is specified as a total and then distributed by the game over the block health when you harvest it. This is confirmed by looking at terrain blocks, see below.

 

Quote

 

Also you can get several resources with one swing, and there seems to be some patterns there as well, but nothing I'd have recorded properly. But to me it seems it's more like harvesting the old boulders where you'd get a random amount of everything.

 

Meanwhile you can get several resources out of a loot box, no difference.

 

Naturally harvesting a car is probably working the same way as harvesting a simple block of earth or a stone boulder (developers want to solve the general case and not use 10 different routines for 10 slightly different cases). If you check blocks.xml you will see that all the blocks have harvest amounts. For example terrForrestGround has simply 22 clay and 2 fibers as "loot box content" and when you dig it out with a iron shovel it gives you, surprise!!!, 22 clay and 2 fibers.

 

Which by the way did not change with a measly stone shovel and a superior steel shovel (which is another hint that the OP has probably imagined the difference between wrench and impact driver)

 

Conclusion: My theory that harvest is similar to loot boxes is correct. But since there is no draw count and no loot groups on car harvests (meaning there is always just one draw) you are right that with cars each item is its own data set

 

Quote

 

And I did mention, admittedly in an unorthodox wording, that the drops would need to be independent variables. IF that's the case, then your a) and b) would be compared gas to gas, forged iron to forged iron etc. And while 430 to 490 and 13 to 8 look quite different, that's easily within the randomness with that small a sample. I wouldn't predict percentages until I have at least 100 results, as in, 100 headlamps from each set.

 

You misinterpret my inteniton here: 10 or 20 samples would not be enough to provide statistically sound evidence. But it would have been enough for ElCabong to get a hint whether he is right or not. And in case he seemed right it would be reason enough for me to investigate further.

 

Another way to make me investigate further is to entangle me into a discussion about it though, like we are doing now. Without you objecting I would have simply waited for ElCabongs result.

 

Quote

 

And yes, I'm guessing a whole lot here; with all the conditionals there I shouldn't need to stress this, but I will anyway...

 

[theFlu]

Too long to quote, so I won't, but a quick recap;

- without knowing the mechanics, we can't decide anything from a small sample. Agreed.

- the method of deciding the "contents of a car" and the method of giving them out as "harvest" are not related. Agreed.

- terrain harvesting is set up similarly. Yup. You can see the harvest numbers for cars set up in blocks.xml as well, under "cntCar03SedanDamage1Master" and friends. This doesn't really tell us much about how the car loot table is handled.

 

- Your conclusion; how did you decide there's no draw count for the car loot? I can't make that call, but I'm not that familiar with the xml:s. Could be included from somewhere

 

- 10 to 20 samples.. the problem is, it is enough to get "an idea", but the probability of getting the wrong idea is.. high.

 

For an obvious example, if you're trying to catch a 20% difference, and you get samples of 4 and 5, you're right on the money. But you're also at exactly "one lucky roll" away from being equal.

"Two lucky rolls" could hide a difference of 10 and 12, will be a little rarer, but by no means impossible.

 

My main point is, 10 shopping carts is just not enough data. The next step for the average joe is, "well, if I don't see the proper results, I'll just try again". And then you have two bad datasets and you've just confirmed your original bias by the latter being "right".

That's why an "actually significant" sample size is required (I'm not even pretending to care about science here, but just for one's regular in-your-life decision making..). When you think about "how much would randomness need to change things to change the result significantly", you end up having to go for a lot of data. It sucks. But having terrible data is worse than having no data. (And if you don't mind, tell that to the nutritionists.. )

[meganoth]

I did decide that cars don't have a loot count because:

1) Nothing in the xml of cntCar03Sedan... looked like or would work as a loot count. Occams razor: If there still was a loot count it would have to be somewhere else, a special case among blocks AND differently solved than for loot containers even though they just could have used a loot group if they really wanted a loot count. And that special case for what? A draw count doesn't seem to bring much too the table. The by far simpler explanation wins out: No loot count

2) The amount of loot listed there checked out well with what I saw from harvesting one car

 

The probability to get the wrong idea is high, so what? The probability to get the right idea is still higher and we are not discussing failure rates of Covid tests 😉. The only consequence was whether I would invest further time in investigating a question of which I was pretty sure I knew the answer already.

 

Your obvious example is also one where I would not bat an eyelid at 20%.  The obvious obivous example would be if the drop chance per draw would be 1/10 or lower, then 10 draws would likely have samples of 0 or 1, then even 100% difference would be expected.

 

But Toilets or lamps drop resources at higher variance. Well, I was just assuming they did, but I was right, at least for toilets. A difference like the OP mentioned in his second post would probably show up.

10 is still really really low, but it has a higher chance that someone I don't really know well would really do it if he has to search for toilets.

 

 

 

 

[theFlu]

Thanks for your patience, I hope you're getting paid for this And well, don't get me started on covid, although.. no, this would get political quick, I'll refrain

I asked for your reasoning mostly since you supported my original bias; seems absolutely reasonable, but not entirely conclusive .. good little confirmation for my bias though ...

 

Looking at toilets in the xml (cntToilet01), it'll drop pipes on "destroy", which is likely not affected by harvest boni. So the usable variable from those would be scrap iron. It changes between 5-10. Ten toilets would then drop something approximating a normal distribution between 50 and 100. There are people who can easily answer "how likely will I guess the direction right if there's a 20% difference" from that, but that's not me.. (I'm using 20 as a single rank in the skill, a difference in tool ranks might be more, but hey, would be trying to capture info here!)

 

I do want to note here, with "toilets & shopping carts" I originally ended up thinking of "pipes & mechanical parts", which are dropping on very low amounts - that makes it quite prone for random errors. Akin to the experience of the OP with the steel btw. The 50-100 range would likely be good enough to see a 20% diff pretty reliably, but I'd still increase the number a bit for comfort. Doing boring work sucks, but being wrong sucks more, or something like that

[ElCabong]

I just tested 10 coke machine with the orange ratchet and 10 with the blue wrench, it's essentially the same. I got one more steel bar with the wrench and a few more mechanical parts with the ratchet. This is not enough variation to signal a difference.  Perhaps I need a larger sample size, say 100, but I'm not interested in doing that.

 

I am harvesting more stuff than earlier probably because of my game stage. Perks haven't changed, I put one point into salvage operation shortly after I looted my first wrench.

[meganoth]

On 3/10/2021 at 12:11 AM, theFlu said:

Thanks for your patience, I hope you're getting paid for this

 

Maybe in the afterlife 😄

[Maharin]

3 hours ago, meganoth said:

 

Maybe in the afterlife 😄

 

Maybe the afterlife is your fully dream modded version of 7DTD?  Heaven or hell is, of course, your perception. 

[meganoth]

8 hours ago, Maharin said:

 

Maybe the afterlife is your fully dream modded version of 7DTD?  Heaven or hell is, of course, your perception. 

Mmmh, sounds like a strange religion if heaven were populated with lots of cannibalistic not so bright fellow humans

 

On the other hand, what are we really ourselves when we are nominally dead but still concious and walking around in some sort of paradise? Undead!

 

 

[Canute]

On 3/10/2021 at 2:43 PM, ElCabong said:

I just tested 10 coke machine with the orange ratchet and 10 with the blue wrench, it's essentially the same. I got one more steel bar with the wrench and a few more mechanical parts with the ratchet. This is not enough variation to signal a difference.  Perhaps I need a larger sample size, say 100, but I'm not interested in doing that.

 

I am harvesting more stuff than earlier probably because of my game stage. Perks haven't changed, I put one point into salvage operation shortly after I looted my first wrench.

While you allready on it. Could you maybe get impact driver and try it too ? 🙂

But i don't think it would change anything at the result.

The tool just made it faster to salvage the object but don't give more out. Only perks/traits affect it.