Quality

[half raptor]

I'm not trying to craft from normal to legendary, I'm just trying to go for upgrades on each step. The math works for crafting to because the math is asking how to go to that quality in one crafting step with recycling for failed results

(though I suppose the discussion is moot if you prod all the intermediates in your base and quality loop the finished machine)

[daring siren]

What is this "overflow" you're talking about?

[dense kraken]

when unused stuff piles up

I'd aim for something similar as well, minimising recycling is key

tho there isn't that many steps in factorio and 25% per step isn't that much

[half raptor]

so let's assume I have an imbalance in the production of and where I am now accumulating cable. My trains stations will have their priority set by the equation ((goal-inventory)*100/goal) so that when I have a large buffer, the priority will decrease, this is how I will balance delivery to my various request stations. However if P falls below zero, that also means that the train limits will also be zero. It won't happen right away, but eventually I will have "Destination Full" while trying to deliver my cables. I will set in the train interrupt "If Destination is Full AND Time Elapsed >10 seconds, go to Recycle Station"

My recycle station will be where my excess items go for recycling back down to and then the train that picks up the products from my various recycling will deliver back out to the stations. Keep in mind I'm going for quality in all machines with the exceptions of the lines that are requesting my Legendary ingredients, which will be handled via Productivity modules.

As my buffers stay full, the recycled copper has a chance of coming back upgraded and then I'll eventually begin to accumulate excess if i'm not getting enough uncommon iron plate to go with it. The same thing will happen again where I'm now recycling the uncommon wire and getting back at minimum uncommon plates back. Eventually I'll start accumulating Rares, then Epics, then Legendaries.

To counter the over production of I would have a secondary station in the same stop as my epic, rare, uncommon, and normal stations, prioritizing them in that order. Since assemblelrs can be set to accept higher quality inngredients in place of lower, it'll still work

By setting a priority splitter to favor the legendary overflow before using the correct tier ingredient, I'll make the destination full message appear more often and prompt more recycling, which would probably lead to more legendary cable, so eventually I would need to set up storage buffers for it all since there's no benefit to recycling it.

However I can set alerts that tell me I'm accumulating too much and using these overflow stations and I'll see "Oh, I need more production" and as I'm now consuming more cable, the buffer will go down again, and the next time I might start accumulating more iron plate than I can use. Granted it has more paths it can go down so it'd take a while for that to be a problem. I plan to set up a monitoring station that sends me global alerts to tell me what modular block I should build next as it would be based on 1) Reading that I have an accumulation of the resources and 2) I have a high demand for the product in question. If Another product of those resources have a higher demand it'll favor that alert.

An example of that would be if I start accumulating Iron Plates, it would read the relative demands for Steel, Gears, Pipes, Green Circuits and whichever of these has a higher demand wins. Granted the green circuit factory would require a high buffer of copper cable as well to be considered, so if I'm starved for copper, it'd only be looking at the first 3 for comparison.

The way I'll be measuring my average priority value will be simple. Using the new radar circuit network, I'll output the P value as (item icon) = P on the red wire and a constant combinator that outputs (item icon) = 1 on the green. At my monitoring station it'll just say ((item icon)(red wire))/((item icon)(green wire)) and if I have 12 stations requesting an item, it'll take the total priority value and the 12 from the green wire to give a live average priority value for that item across the whole base

sorry for the wall of text, I'm left alone at work too much and I've been thinking about this a great deal

[dense kraken]

sharing is caring

you seem to care a lot about quality

[tall sandal]

i liked reading it

[heavy ember]

Has anyone done the math on output per second per quality+prod module in assembler?
Rather than ingredient cost?

[tall sandal]

not that im aware

[charred gyro]

I wonder if they'll balance prod modules at all so they're not so clearly the better choice compared to quality modules

[tall sandal]

hm? quality and prod modules do different things

[charred gyro]

Yeah, but it seems like everyone's decided that productivity is superior with the stats we've seen so far.

I got the impression they want to make them more comparable

[uncut helm]

if your goal is productivity then yeah productivity module is superior but its not that complicated

[heavy ember]

Can't get quality items without quality modules.
Honestly the balance seems great if from a material efficiency standpoint it's best to have both quality and prod in the same machine.

If quality modules were better, we would forgo prod and just use quality modules to push things to Q5 for science. That doesn't seem balanced.

Also we now have machines like foundry that have built in prod, which is basically a prod module nerf on them. Same goes for productivity research

[rotund egret]

I'm most happy about the s

since slotting a couple legendary ones can give you amazing productivity at only a fraction of the power cost

which might end up being a bottleneck

though funnily enough a quality non-SA 1000 SPM factory will consume significantly less power than it does now 😄

and have a much lower footprint

[half raptor]

I'm sure a lot of people will dip their toes into SA by designing 2.0 only bases with quality turned on

Such a fun feature on its own

[jaunty citrus]

There is also the issue of drastically reducing your quality output speed, even if it is cheaper in the long run

When you only look at one stat, the material cost of quality looping, then yeah it's good to use prod modules. But that isn't the only factor

[half raptor]

Well you're trading chance to upgrade for more materials to recycle loop, math comes out very close if you go to either extreme

People are saying that a balance is going to yield best results

I'm going to be stubborn and go quality all the way

[jaunty citrus]

Yes, based on material in -> quality out recycle looping it's better to use a mix

you go from ~160x cost to ~80x cost

[half raptor]

Well read my wall of text above for how I'm going to approach quality

[jaunty citrus]

but a) that's with all modules and b) it doesn't take into consideration the machine speed

[half raptor]

That's why I'm going modular train block base with train interrupts to sort it all because it'll be a process to get modules, upgrade modules and everything else so the math will be always changing. Letting the trains sort and solving the base as I go is going to be my best solution rather than building "Perfect End Game Base" from the start

I'm going to let bots place modules across the whole base then when I start accumulating those upgrade planner them to and start recycling the normals that return back

[trail cradle]

I just implemented productivity modules into the quality cost calculator I’m making. Results surprised me a little so I want to check what the consensus is. When feeding base quality items into a recycling loop such as green chips, if you want to optimize for (legendary Green Chips OUT):(normal Green Chips IN). It’s actually best to use productivity modules in the EMP not quality. Quality modules are just in the recycler(since recyclers can’t take prod anyway).
Actually it’s like this for all the quality levels I’ve tested, prod modules in the EMP perform better than quality modules. I didn’t expect this

[half raptor]

I think that was others consensus too that 1qual 4prod was the optimal ratio assuming no other prod bonus other than the +50 from the EMP itself

But 5prod is a close 2nd

[kindred crater]

#friday-facts message

[half raptor]

Okay I was half right

Can't wait to see the math with each rank of researched prod bonus

[trail cradle]

from my testing, prod bonus has a inverse multiplicative effect on the cost, as in, each level of prod reduces the relative cost of higher quality by more and more each level. I guess this makes sence as once you hit the 300% cap they are just time and energy to produce, you won't lose any material

[half raptor]

Makes sense

[daring siren]

Interesting that we'll hit prod cap early on +50% prod buildings

[trail cradle]

Ok this is really good, confirms my thoughts. My numbers are a little different from theirs though, but not by too much. I'm using a random number generator for these calculation, I have a program for the exact mathematical expectation, but I havn't been able to figure in prod modules to that one yet

[half raptor]

I'm still putting my faith in the fact that most end game crafts are 6-8 crafts from ore and so that's a lot of chances for upgrading intermediates and I'd just set up a destination full interrupt to put my unused intermediates through the recycler.

A lot of the math you're looking at is going from ingredients to product

[trail cradle]

Ya, I'm specifically working on the recycling loops, which are by nature lossy and probably suboptimal until we get a really high prod from research

[trail cradle]

I'm also planning to do something like this, but I think the bottle neck will end up being the shortest recipe chain

so for modules, that would be red circuits, they are used in the T3 module recipe and the fact that they have plastic in the recipe, which come directly from coal, gives them the fewest steps in which to use quality modules

As in, we'll probably have lots of higher quality T2 modules, and blue chips at the quality T3 assembly area that went through many steps, but our red chips will be holding us back since it's the lowest quality among the ingredients that's the bottleneck.

[half raptor]

@trail cradle a realization someone else pointed out was that you can set the assembler to accept higher quality ingredients alongside the quality it's aiming for. Of course if you have and it'll craft a so if you have an accumulation of legendary materials you can set an overflow station that will bring excess to your stations to be used as lower tier ingredients

With priority settings you can tell it to overflow to epic, then rare then uncommon etc and only stress about it if you're dumping legendary materials into your crafting station

[daring siren]

Why though

[half raptor]

To keep production going

It's a good thing if you're over stocked on quality materials, and you can set alarms to tell you when it's happening and all you have to do to fix it is to increase your production of your limiting reagent

But having it automatically doing so will keep trains from clogging up when they can't deliver legendaries anymore

If I'm measuring a low priority value for both and and I'm also reading a high priority value for then it means I have the materials buffered, I just need more assemblers or EMPs and I just build another modular block for that

[jolly swift]

once we hit 300% productivity on a machine isnt 5q the only sensible option

[half raptor]

Yes

[jolly swift]

and also at 300% prod recycling for quality is free i think

[half raptor]

It might be that the meta will be to pump productivity until you get to 300% prod and then replace one module at a time as you research to quality

[jolly swift]

that sounds accurate

[half raptor]

Yes you get 4 items per craft and you get a quarter ingredients back and you can quality roll while recycling

It will probably be that as you infinite research you transition to quality modules but start with prod

[jolly swift]

that actually introduces a really intersting problem when optimising a megabase for UPS

because then you're debating between speed modules or quality modules

speed mods = less machines = higher ups

[half raptor]

That's the puzzle

[jolly swift]

quality feels like a gift that just keeps on giving

[half raptor]

It is so controversial we've been discussing this for almost 7 months and had to start a channel for it

[jolly swift]

💀

[half raptor]

At least it's not boring repeatable 8 beacon builds

[jolly swift]

^^ exactly

[rotund egret]

I think people underestimate how long that will take

[half raptor]

Oh it'll be an accomplishment that'll take months I hope

[rotund egret]

legendary modules aren't that hard to come by

[half raptor]

"why is your CPU glowing?"
"Gotta get to 300% prod on all intermediates. Science goes BRRRrrrr"

[jolly swift]

Also as you approach 300% prod it becomes cheaper and cheaper to use recycling loops

An EMP with 5 legendary quality mods already has 150% base prod before any recipe prod

[kindred crater]

a big question is what items specifically will be included in productivity research

[heavy ember]

Whatever those items may be, they'll be the ones to recycle loop on

[rotund egret]

we currently know for sure steel and blue circuits

[terse lagoon]

Fun fact: if ur max quality is ( modules too) and u loop from to , it's the exact same resources usage u get from looping from to using modules. Only in EMP and with no prod research

[unborn flax]

checks out! the difference being a mere 0.07 inputs

[daring siren]

0.07 inputs out of how many?

because if it's out of 1 then 7% is nontrivial

[terse lagoon]

It's 0.0073 difference in items u need to make 1 max quality item for me. Overall u still need just over 13.2 butches of items to input to get 1 thingy of max quality output

[daring siren]

OK so it's 0.07 out of 13.23~

[terse lagoon]

0.007

[daring siren]

ye

[unborn flax]

that's the one... and the equivalent calculation for rares with rare modules isn't far off either (13.866)

wait ignore that misreading my own output: free prod 50%, 5-module machines: 1 normal input -> 6480574181/39070148045 rare outputs (decimal ratio: 0.165870223310540, reciprocal 6.02880963226181) rare outputs are 179722681007/12961148362 == 13.8662621542022 times the expense of a normal output tier 0: prod * 1, qual * 4 tier 1: prod * 1, qual * 4

so only 6.029 for rares

[daring creek]

do we know how much space age will cost?

[sacred totem]

About the same as the base game

[pure spire]

cost is irrelevant game is forever

[neat shard]

@north yew I know we've talked a lot about all the facets of trying to use qual mods for intermediates, and how in pretty much every case it's just better to use prod mods, but I think there is one single niche use case where it makes sense to use qual mods for intermediates over prod mods

That's for the case where some of the ingredients for a recipe are prod-compatible and some aren't.

Take, for example, prod mod 3s themselves. The recipe requires prod mod 2s, red circuits, and blue circuits. The vast majority of the cost is in the prod mod 2s.

Prod mod 2s can't have prod mods applied to their production, so you might as well outfit them with qual mods. You'll end up with a small supply of high-quality prod 2s.

This represents an opportunity to make a high-quality prod 3 for cheap, by using the high-quality prod 2s you already have as ingredients...

...but you'll also need some red and blue circuits of the same quality tier.

Quality circuits are a bit painful to produce because qual modding them means passing up on prod modding them, but in this case the circuits represent only a very small fraction of the total cost of the item you want, so overpaying for them isn't that big a deal.

[north yew]

It would work, but only for one step. It also means having a separate red/blue chip manufacturing area w/ qual mods that specifically feeds only the T3 module production. Benefit may outweigh the cost in this specific case.

The one beef I'm starting to develop about strategies like this, though, is that it requires ratios to be coordinated. That means you have to upgrade all quality modules in the setup at the same time or the ratio goes off. Doing it only at the last step you can slowly upgrade to better quality modules one at a time.

[neat shard]

Does it require the ratios to be coordinated? Like, in this specific example, maybe you have your quality circuit assemblers running off to the side, with basic-quality circuits fed into the base's main circuit belt with priority input, and maintain a small buffer of quality circuits ready to use when quality prod 2s show up.

I think that works regardless of the exact quality level of the assembler(s) making quality circuits.

[north yew]

If the modules are coordinated, then for the T3 recipe you won't even need to mix red/blue chips from the quality section with ones from elsewhere. The ones coming out of the quality section will match the ratio of the T2 modules. Doing it the way you described would work, but it's costly compared to just coordinating the modules.

[neat shard]

I'm not sure how it's costlier to do it my way, assuming there's enough of an outlet for the low-quality circuits.

Can you explain that to me in more detail?

[north yew]

Buffering quality ingredients generally feels yucky to me. On top of that it's quite a bit of logistical complexity getting the quality chips to run as much as you need them to without running too much or too little.
On top of that, I broadly feel that quality is going to be a very complex mechanic to play around even if you engage with it in the simplest possible way. Anything that adds more complexity on top of that is beginning to feel unappealing to me.
That being said, if anything in the game is worth getting fancy for, it's T3 modules.

In other words, I see increases in logistical complexity coming in at a very high premium generally.

[neat shard]

That's fair.

And different people will have different levels of tolerance for the logistical complexity added by the quality system.

[half raptor]

Buffering quality is indeed yucky and honestly I see the argument for going prod on everything and eventually quality looping at the end. I think there are 3 simple solutions and a couple more complicated solutions but the ones that stand out as the easiest to understand are:

1. Productivity module all intermediates, keeping it all at normal quality and then only deal with quality on the final step, creating a loop. This will mean you'll need to transport a higher quantity of intermediates before the end.

2). Set up quality loops on intermediates early in the production chain, specifically where we can potentially get the 5th slot for EMP to assist and in one step go from normal to legendary and everything after that just uses legendary ingredients for crafting

3). Set up quality modules in your factories that output normal products for use in the rest of your factory and push any and all upgrades items through a recycle loop.

4). My favorite, we set up modular factories that are set up to handle assembly in every combination of quality and design the train stops with train interrupts to sort everything, using recycling to address buildup of unbalanced materials

I said 3 simple because #4 is far from simple

[opal haven]

3 sounds good, especially for recipes with no prods (although recycle looping can wait for quite a bit - a bit too wasteful even for a midgame)

[half raptor]

the great thing is all of those methods will work in the end

some better than others, but there's more than one correct path

[jaunty citrus]

#5 is to quality loop on select recipes that have a productivity research, and crafting everything from those items, fulgora recycling style.

[frank warren]

@jaunty citrus The recipes that return the same thing do give higher quality stuff

I found a source (pov: I asked pepperbox)

V453000 mentions that you can't recycle fish at all because it had to be hard to get quality fish

this sentence makes no sense unless those recipes do quality bump

[jaunty citrus]

I don't see why they wouldn't bump quality in the first place

If you have quality modules, it can increase quality

[frank warren]

Didn't you ask for a source of that behavior?

[jaunty citrus]

No, I asked for these rules #friday-facts message

[frank warren]

those are just the rules I have that put everything together

my bad

[jaunty citrus]

I think someone else might have

Huh, you did #friday-facts message

[frank warren]

I asked because someone else did

but them asking appears gone

[jaunty citrus]

Probably yeah

I wonder if the massive boiler setup in the Fulgora soundtrack video was because you can't recycle solid fuel, or because you might as well get some power from your excess ice and solid fuel

[frank warren]

I bet you can recycle solid fuel

well

idk

you can recycle plastic and coal

and solid fuel feels to me like a strange child of those two items

[jaunty citrus]

You probably can, it'll just recycle into itself.

[frank warren]

imagine if they implemented the actual recipe

It's one case where it would actually be worse

(fluids are deleted on recipe recycle)

[jaunty citrus]

It's just a little odd, although I imagine the power priorities when using accumulators/lightning rods make it relatively simple

Yeah I imagine battery/processing unit cycling to be best on for that reason

[frank warren]

rod probably has priority over engine

but engine has priority over accu

so any time there is a lack of power coming in (all the time on fulgora) - engines run

[frank warren]

I mean.. among other reasons

battery is great for the copper/iron

it's kinda meta

but honestly idk if it will be enough sulfuric acid

I get that it is 'free in the ground' and all

I suppose mining prod helps us out here, but we don't have a super pumpjack

OR DO WE?!

[jaunty citrus]

Yes:

[frank warren]

idk it doesn't make as much sense as big miner

that's fair

[jaunty citrus]

Make your own super pumpjack with quality!

I'm a little sad that pumpjacks are probably still not flippable

I think a big mining drill makes sense, but I don't see what a 'big pumpjack' would even do other than extract faster

[frank warren]

yeah

I guess mining prod could just be part of the cost of business there

[jaunty citrus]

Mining prod + modules + quality will scale pumpjacks pretty hard

[daring siren]

Fun fact, even with a recycler with max quality modules, you still need to recycle about 1000 fish to get one . That said, fish aren't recycleable.
We are not 100% certain about if sending returns in SA. And we don't know if sending items returns items.
These were confirmed and unconfirmed, including the unconfirmation.
If it is possible, then we can recycle 1000 into itself to get and send that.
TBF, I think they should just ditch or change the achievement rather than distorting many things to make it not easy.

[jaunty citrus]

The limited supply of fish isn't that much of an issue in 1.1, as you don't need 10k+ spidertrons

but if you now want say 50 , 10 for each planet? Now you pretty much do need 50k+ fish

[frank warren]

I have a subtle feeling fish and wood will be renewable from Bacchus

not as a main, important thing, but just as a side recipe 'here it is' sort of thing

but it just seems wrong..

but so does needing to fish 50k fish

[jaunty citrus]

There just needs to be a way of getting fish from 'regular' resources

well, not needs, but really probably should be

[frank warren]

We don't know where you unlock it yet

It could be that Fish are renewable in SA

[jaunty citrus]

I hope they are, as in 1.1

Maybe instead of needing to launch space science you need to launch the new end game science

[frank warren]

I wouldn't even really call them properly renewable

It's possible farming fish is just something you do on Bacchus

[daring siren]

Fish aren't limited in 1.1 though, you can get them by launching

[frank warren]

flashback to the past

[jaunty citrus]

Yeah I didn't finish the thought clearly.
Even if you couldn't get fish from space science, as JG confirmed/unconfirmed, the limited supply of fish isn't that much of an issue in 1.1, as you don't need 10k+ spidertrons.

Noone wants to have to drain tens of thousands of chunks worth of fish to satisfy their needs

[daring siren]

The sad fact is that we'll still need to train tens of thousands of chunks worth of fish to have enough to recycle

[frank warren]

What does this mean?

[daring siren]

I'll find it again

#friday-facts message #friday-facts message

[frank warren]

Do you really think they'd make us do this? That's one of the reasons I am thinking it might be worth speculating some kind of fish supply.

[daring siren]

If they didn't mind, they'd let us recycle

[charred gyro]

If we could recycle fish, what would stop us from recycling ourselves? The quest for the legendary engineer.

[daring siren]

The fact the engineer isn't an item

also we had this talk before, about recycling the engineer

[neat shard]

You have options to farm fish in SE, so it's not off the table for it to be farmable in SA too

[daring siren]

I don't see the connection

[neat shard]

Perhaps the idea is for quality fish to be mostly limited to the endgame when you're at Aquilo

Such that you can recycle a zillion spidertrons to get quality spidertrons earlier, but more realistically you'll wait to get quality spidertrons until you reach Aquilo and can farm them

[kindred crater]

if theres going to be a place to farm fish it will be

[final nimbus]

Fishing minigame when?

[half raptor]

Please no

[hoary current]

I feel like will play a bigger role in SA on

[quaint flower]

for small power polls
for spidertrons

[median verge]

i hope its a mini seablock

[half raptor]

I'm still excited over a potential lava block

Or scrap block (can't build on scrap and more items can be found in scrap that makes victory possible only mining scrap)

[terse lagoon]

Ah yes, explosive mixed ore challenge. Very "fun"

[neat shard]

DangOreUs

[half raptor]

But with scrap

[daring siren]

ScrapOreos

[daring siren]

Important quality information #friday-facts message
We now know the reduce quality by 1%, 1.5%, 2.5% respectively. Which isn't a ton.

[dense kraken]

it's still more than 0

which is thus, bad

[unborn flax]

hmm, 1.5 and 2.5 have a decimal point.

[daring siren]

Quality modules reduce speed by 5%. Lets say you have 5 of them that's -25%.
Add one and it completely negates it and doubles the speed, for a reduction of -2.5% quality. This means you need to pay for less than half the modules.

[unborn flax]

is it safe to assume the 2.5% penalty isn't halved in a beacon? i would have thought it would be

[daring siren]

It is, but we assume 2 modules in a beacon, so it's the effect of one module.

That said, putting a single module is also an option, just to negate the negative speed.

[unborn flax]

ok - i read your as 1

If they are halved then I think that with 4 beacons with 2x speed3s, the (4-module) 79.9 only goes up to 91.8. Could someone else double-check?

[daring siren]

79.9?

[unborn flax]

number of recipe inputs required to quality-loop a single legendary output, given you have a a 4-module machine, legendary modules and no productivity bonuses

[daring siren]

Ah you mean input items for output?

We're talking about a 5 module machine though 😉 with +50% built in prod

[unborn flax]

how many beacons do you want?

i was talking about the 4-module one

[daring siren]

Just one beacon, with 2 modules, is a very large speed bonus, because it negates negative speed

[unborn flax]

For EMP the inputs-for-a-legendary numbers are 13.2, 13.7, 15.5 for respectively 0, 1, 4 beacons containing 2 speed3s. Though someone else please check.

This is assuming you're also beaconing your recycling machines - which you don't have to.

The impact of speed modules seems so small, I can't see myself holding back on them in the name of quality.

[daring siren]

Really? It got down to 13.2? 😮

With what ratio of prod and qual?

[unborn flax]

we already knew that right? see the graph in my heretical post: #1215078107334057984 message

[daring siren]

Neat. So it's 4p1q in emp + 4q in recycler

[unborn flax]

yes - and a few speed modules in beacons don't change that

[daring siren]

Sounds like investing in to put in is actually the play, because it reduces the price of all other buildings/modules

In terms of order, not sure which module to start grinding quality for

[trail cradle]

Could we discuss this number? This suggests self recycling loops take ~1000 normal input for each 1 legendary output. Self recycling loops being any item that when recycled returns itself. For all the math I’m going to assume a recycler with 4xLegendary T3 Quality modules.

[daring siren]

I did this test several times on JQ

[trail cradle]

When I run the exact numbers I get that an input of 1000x normal items into a self recycling loop, while recycling everything that isn’t legendary, will yield exactly 0.3667155061529014 Legendary items. How was the 1000 normal input for 1 Legendary output number calculated? It could be that my math is wrong but I want to know why.

I did the exact math on paper, and then built a test in Python to test it as well

[daring siren]

I haven't calculated it exactly

I can calculate it in JQ with FP

BRB

[red kelp]

jq? fp?

[daring siren]

Janky Quality. Factory Planner.

[red kelp]

ah i was thinking of another jq

[daring siren]

FP doesn't show fish recycling for some reason

[trail cradle]

Are you able to use iron plates? Anykind of self recycling item should yeild the same result

[daring siren]

It doesn't show any item to itself recycling, weird

I'll check helmod

[trail cradle]

Do we know if any others have calculated the cost of self recycling loops? It would be nice to compare. I acheived a very simular result using an RNG python script I worte. So I came to the same number from two very different dirrections. That is why I feel like 1000:1 is off by a good bit

[kindred crater]

thats actually not bad at all, assuming those numbers are final

I would've expected even a small speed boost would be enough of a penalty to not make it worthwhile whatsoever

[daring siren]

Turns out it's still useful 🙂

[kindred crater]

just when I thought I was largely done thinking about quality, this pulls me right back in

[daring siren]

As expected, it gets more and more useful, the better you have

Better means more boost per quality reduction

Better means quality reduction is less of an issue

[winter saffron]

This means you need to pay for less than half the modules
could you expand on this?

[kindred crater]

very cool. I don't know how best to articulate it, but going big with when using and is more incentivized in some aspect vs doing that with and

[final nimbus]

Even if the quality penalty was half of what it is, combining them is still dumb.

[kindred crater]

I'm sure there are some set of circumstances that could make it worthwhile

[final nimbus]

There's really not.

[kindred crater]

space limited, and you have a big buffer of ingredients, and you quickly want to get the thing you're after. something in that direction

[daring siren]

make things faster, i.e. you need less buildings i.e. less modules in the buildings

[winter saffron]

now I feel stupid

[kindred crater]

you're sacrificing quality % for higher short term yield

[final nimbus]

At that point your solution is / buildings which don't kneecap your quality bonus.

[kindred crater]

I don't disagree

just trying to find utility in this

1%, 1.5%, and 2.5% quality reduction respectively. it would be easier to write this off if it was higher, but instead its low enough to bother me

paging @obsidian crescent you'd find this interesting

[winter saffron]

A perfect foundry will have 16*-2.5% = -40% quality reduction. What's the quality bonus for leg. qual. modules?

[trail cradle]

Legendary Quality 3 module is a 6.2% chance in game correct? It reads as +62% Quality

[winter saffron]

A perfect foundry will have 16*-2.5% = -40% quality reduction. With 4 qual modules: 6,25% * 4= 25 bonus quality

Unless I'm missing something it's not possible to have quality outputs when fully utilizing beacons for the foundry

[neat shard]

Qual mods of any tier reduce speed by a fixed 5%, right?

[terse lagoon]

yes

quality doesn't increase negatives

wait, u asked smt different

T1-3 quality modules will have different speed debuffs

but for Q1-5 of same tier, speed debuff is consistent

[kindred crater]

#1215078107334057984 message it was said a little earlier we found out that speed modules decrease quality by 1%, 1.5%, and 2.5%

@neat shard

[neat shard]

No, I'm asking about

What are their speed debuffs, again?

[kindred crater]

my bad, misread

that I don't know for certain. we all thought -5% for a while, but it might be different across tiers

someone deeper in quality than me would know

[terse lagoon]

5%, 7.5%, 12.5%?

seems a bit too high 🤔

2%, 3%, 5%?

seems more reasonable

[unborn flax]

I'm taking the 1.5% quality reduction to apply additively to the 62% you get from a - i.e. the same way that prod modules produce speed penalties. This means the effect on the probability of a quality bump is -0.0015. Is everyone else reading it this way?

[frank warren]

I don't know if it's 62% anymore

the fact that this 'bonus' is listed as a decimal means it might be the correct 62.5%

[unborn flax]

This is the point I was making earlier: EffectNumber may now be stored internally as something other than a normal integer.

If true, I hope they go further and state (e.g) the bonuses as [2.5%, 3.25%, 4%, 4.75%, 6.25%], i.e. ditching the 1/10 'base quality' multiplier.

[hoary current]

yeah, I think if they can do -2.5%, they can also do +6.25%

[frank warren]

I would not be so fast to assume that

[hoary current]

otherwise -2.5% applied to +62.5% is just too small as a penalty

at least 6.3% is doable

[rotund egret]

iirc the speed penalty is a constant -5% regardless of module tier

[hoary current]

we're talking about quality penalty of speed modules

[neat shard]

I was asking about speed penalty of quality modules, at least.

[unborn flax]

There's another possible explanation that has a conspiracy-theory level of plausibility: What if they have already got rid of the 1/10 multiplier? Then "1.5%" has an probability impact of 0.015, not 0.0015.

This would mean that instead of "haste makes waste but really not that much", it's "haste makes waste so think carefully about how many speed beacons you use".

ok looks like not: #friday-facts message

[frank warren]

I figured those were the kinds of questions that would be answered, so I just asked them.

it is more fun for us to speculate on the big stuff imo

[unborn flax]

you're smarter than i am

[frank warren]

of course now boskid is alluding to some super secret very big unnaounced changes, unclear if they're around the quality system

I'm skeptical because of the framing of this, but so far wube hasn't let me down so I am not so scared :P

[quaint flower]

I wonder what could be changed/added to quality other then changing some values

[hoary current]

sound effects

[daring siren]

Where is that?

[quaint flower]

for the recycler?

[frank warren]

#friday-facts message @daring siren

[hoary current]

for the item when u hold it. iirc they added different sound effects for different item types

[quaint flower]

I saw that fff. Like when you pick up a item in a GUI or put a item in a machine

it sounds nice

[daring siren]

Woah... are we getting stacking qualities?

[frank warren]

I have no idea what he means

Personally, I don't like the sound of it haha

But I may be biased from the threat-like nature of this feature's introduction :P

[half raptor]

I'm sure some madmen are already thinking of adding 5 more quality ranks beyond legendary as a mod

For people who want their inserters to turn into helicopters

[quaint flower]

Imagine watching your ulta mythic quality inverter fly away into the horizon....

[frank warren]

I think the change would be more mechanical

[heavy ember]

I bet there will be at least one mod with all 250 ish

Oh that give me an idea: each quality tier is 1% better, and there's 250 of them

[daring siren]

Sounds like a stack pain

[kindred crater]

quality part 2 FFF?

[daring siren]

inb4 quality was an elaborate April's fools prank

[half raptor]

After mapping out how my base is gonna go and my whole strategy is decided out, after devoting that much mental energy toward it if quality turned out to be an out of season April fools joke, i would start playing other games

[red kelp]

you could say it was a quality prank

[kindred crater]

do we know if concrete recycles into iron ore and stone bricks, or itself?

just thinking about how it can be used on fulgora

[obsidian crescent]

Concrete is an assembler recipe, so I can't think of a good reason why it wouldn't be recyclable into its components. However, considering that scrap already emits stone, and there are plenty of gears lying around, the only reason I can think of to recycle scrap is because you don't want it anymore.

That being said, it sounds like free quality iron and stone brick.

[kindred crater]

yep that would be nice to take advantage of. I could still see a huge overproduction of stone/bricks, but I see it as a hidden blessing- it would make / stone walls common enough to use (provided it runs long enough and you're moving good volume), which would be a nice upgrade for base security

fulgora is a pretty powerful planet for sure. its pretty adaptable to what materials you need depending on how you manage recycling the scrap output items

[kindred crater]

posting this here since this is the quality channel afterall- major rework to beacons, quality effects their output

#space-age message ongoing convo about this from then on

[kindred crater]

#friday-facts message read this and directly above for tl;dr

[weak vault]

i think we can't have a long-term reliable loop unless we use circuit counters or have a loop directly after a recycling step which ensures that all items in the system will be within ratio for the item that's getting crafted
it seems to accumulate some items more than others

this one is directly after recycling so it stays relatively empty and within ratios

[daring siren]

Law of big numbers though?

[frank warren]

I have found that belts are often not 'large number' enough to prevent jamming within their lifetime.

Those are at least what my tests with your mod showed.

[hoary current]

I believe u need at least a steel chest buffer room

[hoary current]

buffer room on belt or even in chest may not be large enough for the law to apply

and there’s always an astronomically small probability it can jam any finite size buffer

[frank warren]

chest it has no issues

I think we undervalue how much larger a chest is than the belt. Reminder: the effect is exponential with buffer size.

[unborn flax]

I'm starting to think this may be mostly right after all, based on the apparently clamped "-8% quality" from today. And it's not inconsistent with what b*skid said #friday-facts message as I originally thought.

Reason is that (with an assembler3 say) a probability hit of 0.008 to quality only increases inputs per legendary by 11.7% (79.7 -> 89.2), which really isn't enough to give you a bad day. Whereas a probability hit of 0.08 increases this value by 241% (79.9 -> 272.4), which sounds more like "haste makes waste".

[frank warren]

b*skid is a new one.

[unborn flax]

So I think the "base quality" of 1/10 has gone, or rather it's moved to be user-invisible. And will be 6.2% (you're still not getting the last decimal place)

[hoary current]

yeah, in practice theres no issue with chest, just a non zero astronomically small probability

[daring siren]

Pretty sure it's -8% out of 24.8% rather than 0.8%

[unborn flax]

indeed

[frank warren]

This follows with how boskid told me it worked the other day - it is rigged up to display 0.1x the internal value

btw @hoary current this is outdated now

years of academy training, wasted

[hoary current]

I wonder if its still 6.2% for tho

[frank warren]

It is.

[hoary current]

make it 0.01x internal please

[frank warren]

Remember, boskid said that is the 'least of my concerns' given some unpublished changes

If that was these new silly beacons, idk - not feeling it haha

the new beacons are awesome, the quality rounding is still very slightly uncool 😅

[hoary current]

the foundry recipe still bugs me

[frank warren]

yeah... add it to the 'space age pet peeeves' list

[obsidian crescent]

Should it not take tungsten carbide?

[frank warren]

I mean, the recipes are so much stronger speed wise that it's gonna be no big deal

[hoary current]

I mean the molten iron casting recipes

[hoary current]

this is true tho

[frank warren]

so it's more of a 'i don't like this fact about it even though it is not so important'

[hoary current]

just the feeling I guess

[frank warren]

quality one is actually more substantive

it means you can't rely on the 25%

[hoary current]

quality is also about feeling tbh

[frank warren]

maybe, but this one has actual impacts on my designs

any design that relies on 4 splits, 1 being quality items doesn't work

It's still a lot more psychological than real of a problem.

[hoary current]

but the 4 splits isn’t a stable split?

[frank warren]

belt buffering could make it work out well enough

the buffer size for the quality line is big enough

remember: a long belt buffer is a lot bigger than a loop

[hoary current]

I mean how would any of those designs not also work for a split close enough to 4?

[frank warren]

in the long run one belt will fill up, and it will constrain the other one

[hoary current]

why?

[frank warren]

It will output like 3.95 belts or something

because the 3 belts will be overloaded

because there won't be 3 belts of output, but 3 + what we lose from rounding (*3)

It will 99.9% work

realistically everything around quality is going to be 'iffy' anyways.

[hoary current]

but if u can process 3 full belts, why can’t u process 3 non full belts?

[frank warren]

It would be 3 + a bit more belts.

[hoary current]

?

[frank warren]

the non-quality lines would have slightly too much stuff on them

you'd have 3 non upgraded lines and 1 upgraded line

think like a quality smelter

[hoary current]

I see

[frank warren]

it would send off 1 belt of quality items

[hoary current]

but it will only jam about 1% time

[frank warren]

in this case it will instead send out like 0.99 belts of quality items

[hoary current]

I think thats fine

[frank warren]

I mean, I never said it was a big problem 😅

I just said it was an unnecesary problem that I could fix in an hour or so with source access once I was up to speed.

[hoary current]

I thought u meant it would cause the whole system to deadlock 😅

[frank warren]

(it will 99.9% work)

[weak vault]

problem is if it's fundamentally off it'll be off, my machine is likely to jam because i add quality to by doing -> molten copper -> +quality -> +quality
while the -> molten iron -> +quality

[burnt reef]

yeah, the Law of Very Large Numbers means if it can jam, it will, eventually

[iron root]

But will it likely jam within the reasonable (or even unreasonable) lifetime of a game

[burnt reef]

right. you just need to make sure that "eventually" is far enough away to not matter.

[iron root]

But a bias in the system does tend to produce impactful results in pretty short periods of time

Ratio mismatches for things like cables and plates into chips create those kind of biases that you’ll want a system for

[tall sandal]

multiply the odds of a jam by the number recipes you use the setup in times the number of ingredients per recipe times the number of hours in your playthru

[half raptor]

cablel is an odd one because while you will have an extra step to craft it over making iron plate, you can also use the cable for on top of so lower quality cables can go toward green circuits and your higher qualiity goes toward red circuits

Personally I plan to have modular train blocks for each item and quality, with a train schedule condition that says if my destination is full, go toward the recycler block

[neat shard]

...It suddenly occurs to me that the quality mechanic seems pretty darn unfriendly to direct insertion builds.

[iron root]

yeah... it would be

[uncut helm]

Broski-d

[dense kraken]

bro do be skiing

[frank warren]

way ahead of you

[uncut helm]

@neat nacelle

[frank warren]

yeah yeah it's his twin brother ore something

[unborn flax]

Since you're all too polite to say told-you-so: I was dead wrong about my suspicion EMPs may not take prod modules (based yesterday's PIC of EMP). So making legendaries with them is really cheap after all, mitigated by the fact that they can only produce a handful of recipes.

[frank warren]

I am not not polite enough to say it, I just am not attentive enough to remember you made the claim

I would've said it if I saw though. 50% prod on emp is lower than the 100% we can get in an assembler already.

Also it having 5 module slots is kinda one of the big selling points, so it would be confusing if it did not support prod

[unborn flax]

They would still have beaten assemblers if you're trying to make legendaries (even without prod modules) but the argument is dead now

[frank warren]

Yeah, fair

[neat nacelle]

Whut?

[frank warren]

oh no broskid is here

[heavy ember]

Are we censoring dev names to avoid pinging them now?

[frank warren]

People have done that for him for a long long time

[quaint flower]

--->
--->
this is right?

[frank warren]

for recycling?

[quaint flower]

yes

[frank warren]

yes, but it won't do circuit -> cable in 1 step

It would be a lot more torturous if it did have recursive recycling chance :P

[quaint flower]

🤔

--->

no copper cable?

[quaint flower]

i mean you would gets and right?

[frank warren]

yes

sorry

no circuit -> copper in 1 step

[quaint flower]

and with you can make

[frank warren]

yes

[quaint flower]

nice

[median verge]

Now that is a beacon rework i can get behind. I knew they would think of something better than overloading

[sacred totem]

i mean, it was pretty damn obvious of a way to do it

the implementation is the cool part

[median verge]

And the quality tie in? Perfection

[neat shard]

As I said before, I'm a little put out about machines getting irrational number speed multipliers, but that's a minor thing, really.

[obsidian crescent]

It turns out that the computations for what the scaling function is were incorrect. There is no function; it's an arbitrary table that maps number of beacons to a percentage. So the only 2 data points we actually know are the two we saw pictures of; the numbers could be anything between them.

If I recall correctly, there was even a suggestion that a single beacon affecting a machine might get a bonus.

[tall sandal]

nah its (currently) 1/sqrt(n). boskid implied that we dont "know" its 1/sqrt but it is

[burnt reef]

math nitpick: an "arbitrary table that maps [input] to [output]" is still a function. in some foundations of mathematics, that's literally how "function" is defined.

[tall sandal]

based pedantic mathchad

in base.lua:

local effectivity = {}

for i = 1, 100 do
  table.insert(effectivity, 1/math.sqrt(i))
end

[trail cradle]

Yo I just logged on and saw the beacon changes!!!! I can't believe they actually did it! #friday-facts message

[tall sandal]

here's the message alf is referencing btw

[tall sandal]

wtf ur some kind of prophet

[trail cradle]

that's how I would have balanced beacons but I thought it would be too complicated given the scaling would lead to irrational numbers.

[frank warren]

yeah it's actually not that though

[tall sandal]

WTF

[frank warren]

He confirmed it is a single line of of code

[tall sandal]

local effectivity = {1, 1/math.sqrt(2), 1/math.sqrt(3), 1/math.sqrt(4)}

[frank warren]

so actually:

effectivity = {1, 0.7071, ...} 

No, it's manually cut to 4 decimal places

[tall sandal]

okay dokay

[jaunty citrus]

I've got to wonder how many beacons they decided to put in the table

[frank warren]

685ish characters worth

so we estimate about 70 or so

my bet is 64

[jaunty citrus]

You could probably get a good guess cause I think them mentioned around 600 charactrers or something

Yeah

[iron root]

the initial ss we got implied 1/sqrt(n) but the reality is that it can become anything the devs please

1/sqrt(n) was probably just a placeholder function because its a good first thing to jump to the moment you hear 'diminishing returns'

which is a good place to start when bridging low beacon and high beacon setups

[daring siren]

1/sqrt(n) is the Space Exploration core mining rate too

[tall sandal]

which is really funny to me

[daring siren]

Why funny?

[tall sandal]

its hard for me to describe, except that it is such an earandel feature though and though

[iron root]

its more than just E that uses square roots

[heavy ember]

If it was up to me, it would be:

100%, 86.9%, 57.1% 28.3%, 10.6%, 3%.
S(n)=e^-[((n-1)*2.67)^2]

[opal haven]

it drops way too insanely at the end imo

[heavy ember]

I know, yes.

[opal haven]

formula at least should make sure that adding more beacons not making setup worse. your formula not adhering to that

[heavy ember]

Ah sorry. Those numbers is the effectiveness of n:th beacon

Two beacons would be 186.9%

I just remember the numbers in that form by heart at this point due to the amount of theorycrafting I've spent doing with them in an another game

[opal haven]

hm, that would work. but then order of beacons will matter, while it's not important with 1/sqrt(n)

[iron root]

so exponential approach to a given value

[tall sandal]

I know but the 1/sqrt drives it home lol

[heavy ember]

100%, 186.9%, 244%, 272.3%, 282.9%, 285.9%

Also if the order of the beacons matter, and the largest bonus goes first.
That would be good for mixed module layouts with different quality tiers

[opal haven]

using different powers instead of 1/2 (in range (0;1)) might give some fun numbers to play with

[iron root]

you could also make it order-agnostic by just getting the total value by summing the series and then dividing by n

[heavy ember]

I think wube has gone with the order doesn't matter

[heavy ember]

I wonder if anyone knows where these are from

[iron root]

yeah, it'd be a pain to make a system that does care about order

[burnt reef]

"placement order matters" works terribly with blueprints

[frank warren]

this is roughly how it works now

[iron root]

efficiently running a system in which order matters likely requires machines to update their beacon effect only whenever the beacons around the machine change
and that system does have drawbacks, notably fragility in the event that anything unexpected (code-wise) happens

[unborn flax]

Module stats that are a mixture of facts inferred from FFFs / 'leaks' and reasonable assumptions. Is this right based on what we know so far and do we know any of the 5 blanks?

[burnt reef]

all the known bonuses are affected by quality at +30% of their base value per "level" (of Q, in your chart), so it's reasonable to speculate efficiency modules being 30+9Q, 40+12Q, and 50+15Q, respectively.

also most bonuses follow a +2/3/5 pattern, so I'm gonna speculate that's the speed% penalty of qual mods

[unborn flax]

That's reasonable - and I agree 2, 3 are most likely for qual1, qual2 speed penalties. I just wanted to make sure I hadn't missed out on any leaked images that tell us for certain.

[burnt reef]

(I do not think the latter speculation is reasonable enough for you to put in your chart. just calling it now ahead of time.)

[unborn flax]

In fact we don't even know the 30,40,50 numbers in the eff modules, but given that all other values are unchanged from 1.1, it's a reasonable assumption to put them there

[burnt reef]

it could also be 3/4/5. that's the Efficiency module bonus pattern the prod mod speed penalty pattern.

[weak vault]

idea: put everything in the chart, but have a legend color that says "not 100% confirmed"

[burnt reef]

the 2/3/5 pattern includes assemblers, and if you include the EM plant, it becomes 2/3/5/8. machine with craft speed 3.25 when please

[unborn flax]

yeah - i'll change Q to (1, 1.3, 1.6, 1.9, 2.5) as well

[weak vault]

"Q is the quality multiplier {1, 1.3, 1.6, 1.9, 2.5}"

[unborn flax]

possibly better

[trail cradle]

This table is really awesome, great work. I think we can also move the prod3 pollution% of 10 into the "known values" thanks to the leak from kovarex

#space-age message

[stray marsh]

I wonder if high Q speeds worts it for Quality

[frank warren]

@hoary current We need to seriously crunch the numbers here. The cost-performance analysis for modules and beacons is very nontrivial now.

I'm not even immediately sure how to approach such a problem.

[hoary current]

I have no idea

especially when we also need to consider the cost of quality machines

[daring siren]

Hey @unborn flax the person of the hour, featured in today's FFF 😉

[frank warren]

yes, also the costs of modules are nonstatic too because you can use quality first on the quality making devices

[hoary current]

the problem is as complex as quality in general I think

[daring siren]

And modules can get as well

[frank warren]

more?

[hoary current]

yeah perhaps more

[frank warren]

this was part of my original statement

[hoary current]

I mean when u need to consider where and when to put quality first , it’s already a very complex problem

[frank warren]

it is harder because it contains the quality puzzle inside 😨

it's an extension

[hoary current]

how do u do quality is relatively easier than what item and when do u do quality

[daring siren]

3 parameters to add to. Buildings, modules, beacons.

Quality buildings mean you need less of everything else, so that's probably higher priority

However, with the 1.5x supercharge, it might be worth investing in before in modules

[hoary current]

high quality beacon is better for modules than high quality machine

[daring siren]

True... because each can affect 8~ machines

But the quality bonus on is a bit less. It's just 0.2 more

That said, 0.2 of +100% speed is 20% on 8 machines... better than just 30% on a single machine

[frank warren]

yes, but a beacon affects modules more than the machine does

[hoary current]

high quality machine also get higher amount of speed reduction from prod modules

because the base speed is higher

[daring siren]

Wait no

It's multiplicative

[hoary current]

hummm

yeah

final speed = machine quality x machine speed x (module quality x speed module - speed debuff)

the multiplier outside the bracket is more effective

[daring siren]

I'd say the multiplier inside the bracket is more effective. Considering a -50% debuff, you would have to increase speed outside by 100% while only by 50% to make it twice as fast

[hoary current]

I guess it depends on the magnitude of the debuff

[daring siren]

Increasing the smaller multiplicative gives the bigger bonus

[hoary current]

yeah

[daring siren]

So what is our priority?
Quality modules
Beacons
Other modules
Production machines

[unborn flax]

Oh wow. Heh, this had to be on the first FFF day I've been away from home...

[daring siren]

Did they contact you for permission to feature it?

It's a good looking base

[unborn flax]

i posted it on the SE discord in the "hey look at the pic of my base" channel - so permission definitely not required

[daring siren]

It's still common courtesy I'd say

[unborn flax]

i've got other fairly neat builds for the SE ores but I thought that one looked the best

honestly i'm just flattered, i really don't mind

[daring siren]

even not the image, but rather having your name in an FFF... it's a big deal!

[unborn flax]

It's true that a is strictly better than a if you care about quality, despite both being +50% speed. So I can see myself using , in certain endgame builds.

[daring siren]

Because it has less power issues?

That said, is probably cheaper to make than

[obsidian crescent]

Because you can put a beacon with a double Q5 speed1 around a bunch of assemblers making quality goods and not destroy their ability to make quality stuff. That being said, remember that beacons amplify all of the effects of a module, including quality degredation. So a 1.5x multipler on two speed 1s will give a -3 modifier. That gets worse with higher quality beacons.

[daring siren]

Ah for quality maluses. Yea that's a good point

[unborn flax]

yes that's what i was saying

Back-of-the-envelope calculation: a is 65x the cost of a , whereas a is only ~11x the cost of a . So the difference in cost feels like fair balance.

[stray marsh]

One speed3 offsets one qual3
So a single speed-speed beacon loses 15 qual in exchange for +300% speed 🤔

With your average 4x qual3q5 assemblers having 25% qual and 80% speed
... that's totally useless
Okay

100% speed costs 5% qual

[hoary current]

if u r recycling upgrading using mall recipes instead of intermediate recipes, u would probably just use full quality modules tho

in that case u don't need speed beacons as the recipe is already super fast

at least not worth for the reduced quality bonus

with all the new tools I'm pretty confident that we will be able to afford recycling loops easily

[daring siren]

is even cheaper if we consider EMP prod

[unborn flax]

i did a very rough 5x -> and 5x -> , divided by 1.5 * 1.5 to account for EMP prod - so 11.1-ish.

[tough tulip]

65x?? that's insane lmao

although yea it seems fair but still

the time it takes to have the ability to build the best factory just exploded

we're going to need an hours in a day dlc too otherwise bad things are going to happen when this drops

[neat shard]

Meanwhile, in actual IRL factories... https://xkcd.com/2204/

xkcd: Ksp 2

"The committee appreciates that your 2020 launch is on track, but the 'human capital/personnel retention' budget includes a lot more unmarked cash payments than usual. What are th--" "Public outreach."

^ this but space age instead of KSP2

[median verge]

Lol

Engineering firms be like:

[unborn flax]

Do we know much about the recycler stats? Some guesses based on FFF-399:

1. Time to recycle = max(base crafting time, 1s) / 8. Or if there isn't a minimum of 1s, I think the crafting times of , , have all doubled.
2. Recycling time ignores multiple outputs from the original recipe (it's not pro rata). So you have to run the recycler twice to invert 2x , and concrete is even more painful.
3. Recycling iron plates is comparatively slow (0.4s?) - see the recycler at the bottom. So (bit of a leap here) perhaps there's a voiding penalty where there isn't a crafting machine recipe. Stone and ice aren't slow, suggesting both can be produced with a recipe.

Also I think LDS crafting time has reduced from 20s->15s.

[jaunty citrus]

The recycler at the bottom finishes with the iron plate in much less than half a second, and the one above it takes a lot longer than that to recycle one LDS.

The bottom one also finishes recycling 3-4 copper wire in compatible time to one iron plate

Far too little information to be making strong claims though, you'd need to be assuming too much

[unborn flax]

I think that's a little unfair - I frame-counted everything I could. 1) was otherwise consistent with the crafting times we know for red ciruits, processing units, battery, etc. Most of the ambiguity I encountered was from unclear belt stack sizes.

For instance the LDS takes ~112 frames which (given my hypothesis of x8 speed for recycler) works out at 14.93s for the original recipe.

[vapid pier]

question: for quality module 3, it has +2.5% quality base. at Q5 it has a shown quality of 6.2%. is that jsut a visual rounding of 6.25% or is that its true value (e.g. an Ass3 with 4x quality 3 Q5 modules, is it +25% quality, or +24.8%?

[terse lagoon]

game stores these values is ints, so in reality +2.5% quality is 25 in memory. 6.2% is 62 in memory. And 6.25% can't exist cause it stored not as a floating point number. And yes 4 gives u +24.8%. We don't know if devs will change that behavior or not, but for now it works like that.

[final nimbus]

#friday-facts message

[half raptor]

Seems like a logic check that says "if 4 modules are present, make it 25%"

[dense kraken]

sounds very mod un-friendly

wube wouldn't do that

[charred gyro]

Sounds like it just doesn't have the precision to change to 25% until all the extra decimals get added together if that's the case. I recall older games simply multiplying their internal values by some value within formulas to gain extra precision. They could do the same thing here if that's a problem.

[neat shard]

The problem is that the +30%/+60%/+90%/+150% progression isn't a hard-and-fast rule. It's applied on a case-by-case basis.

So, there's no formula – just a lookup table.

[iron root]

lookup tables are great, theyre also fast as hell
and if you have a tool to generate a lookup table according to an equation (which the devs undoubtably have...)

but yeah like beacons transmission effect has an altered bonus, its goes from 1.5 to 2.5 which isnt +30%/etc
mining drill's depletion reduction also doesn't have a base effect in the regular drill's case, while there is for big drills

[plain wind]

the lookup table is generated with a formula

[obsidian crescent]

Is it even a lookup table? I thought it would just be "here's the bonus per quality level that this property gets" and there's something, somewhere that says "quality level 5 gets a double-bonus".

[charred gyro]

Without it being a table, how would anyone make a mod that adds the missing quality level between epic and legendary which WUBE clearly forgot about?

[jaunty citrus]

Epindary

[obsidian crescent]

They'd change the parameter so that the quality level 5 is not a double-bonus, and that qualities now stop at quality level 6. There would be no need to go to every single item prototype and change its quality scaling.

[jaunty citrus]

The way I understand it is each prototype(?) gets a defined bonus from quality (+30% speed, +1 reach and coverage, etc.), which is then multiplied by the quality 'level' of a given quality tier.
Normal would have a quality level of 0, = 1, = 2, = 3, = 5

[charred gyro]

Apparently there is a level parameter and they skipped 4: #friday-facts message

[jaunty citrus]

Yes, exactly

[charred gyro]

Normal
Good
Great
Excellent
Ideal

[jaunty citrus]

Back to this

[charred gyro]

If I had to add the missing level in, perhaps something like this:
Normal
Good
Great
Excellent
Flawless
Ideal

[obsidian crescent]

Why is there a level above "flawless"?

[jaunty citrus]

Not a bad set tbh though

I would probably not use ideal and flawless at the same time though

[charred gyro]

Flawless vs ideal makes sense in some ways, but not others, I agree. I'm imagining items which are configured slightly differently but still without flaws, whereas ideal goes beyond that to a singlular state of being. I guess I'm mostly thinking of how gems are cut and graded.

I suppose I could just cut out flawless for the 5 levels we know of and that issue would go away.

[jaunty citrus]

Yeah I see the vision and I get the difference, I think it's just a little too subtle

Something that is flawless doesn't necessarily have to reach an ideal

[charred gyro]

I don't suppose people have come to any kind of consensus on replacement terms, have they? I think with {normal, good, great, excellent, ideal} as a set, "great" feels like the weakest link. So many of the rating systems I've seen while searching include negative terms or terms which mix meanings in some way.

[obsidian crescent]

Not really. People have their own ideas about what would work, but there hasn't been any real consensus building.

[charred gyro]

And most of them seem to end up with at least one rank which just puts an adjective down. Like gem grading uses "very good" hah.

[jaunty citrus]

Good, very good, very very good, very very gooder, very very goodest.

[charred gyro]

Maybe they could just rename the system from "quality" to "rarity" and be done with it. The terms would already fit... mostly

[median verge]

They said they had a renaming in the works, no?

[obsidian crescent]

I don't recall seeing anything about that.

[jaunty citrus]

No, they said they're open to renaming them if the community could come up with something better

But no concensus has been reached and the majority of people are at least okay with/used to the current names

[charred gyro]

The names are the easiest thing to change. Maybe I don't take myself and the game too seriously and found it amusing, but if we had a very good counter-proposal which feels good and is clear when it comes to tiers, we can still change it.
I don't know that they would need consensus. Perhaps a "negative consensus" would be enough - people agreeing that something else could be better. Given all the terms that have been suggested, coming up with a better set of terms shouldn't be too bad. Even if it seems like half the suggestions I saw were jokes

[obsidian crescent]

"Negative consensus" isn't helpful. The current names have one important advantage: they immediately communicate most of what the quality system means. Because of the ubiquity of these names in videogames, anyone who's an avid player of RPGs already gets what these terms mean and refer to. They immediately understand which one is better, they know that the system involves random probabilities, etc.

There's already a positive reason to use these names, so simply saying "not that" won't get them to change anything.

[charred gyro]

-shrug-
If that's the case then I can just mod it. Doesn't seem like it would be that complex so I suppose I'm not that invested either way.

[median verge]

I have a feeling most people will pick that option

[obsidian crescent]

I don't think it's going to be that many. Most people probably don't care. This isn't Squeak Through; it's purely cosmetic.

[jaunty citrus]

What people think it will be like: normal, uncommon, rare, epic, legendary
How you'll actually interact with it:

[charred gyro]

I'll laugh if the common terminology ends up being color based due to too many people using different names

Though perhaps Q1, Q2, Q3, etc would be just as easy and more expandable to fit modded stuff too

[heavy ember]

I see it sticking for modules at least.
"Q5T3 prod" is just way too convenient

[kindred crater]

easy to type, easy to readily understand

[jaunty citrus]

I want to recycle my blue blue chips so I can get some purple green chips to make some orange blue inserters

[kindred crater]

that would be awful

lol

[cerulean harbor]

try: getting friends

[jaunty citrus]

Excuse me?

[frank warren]

That seems rude for no reason at all

[daring siren]

I think they mean you need words to convey levels to your buddies?

[frank warren]

i see

'gimmie the purple ones' is something I expect to hear

most people already avoid the names of things with different colors anyways

Try getting someone to say 'automation science'

[jaunty citrus]

Most of the time context should make it clear enough

[daring siren]

That's how ColWill's gang does it... "We need more green green circuits for our blue red circuits... for the red blues..."

[jaunty citrus]

ColWill's gang?

[daring siren]

ColonelWill is a long time Factorio streamer

Those are the names they use in their 2.0-ish runs

[jaunty citrus]

Ah yeah, forgot about them. Don't really watch streams

[charred gyro]

Oh no the precedent has already been set 👀

[heavy ember]

Ahh shit, it's too late now.
Train has already left the station

[jaunty citrus]

Whether you call it legendary, orange, level 5, or whatever I expect most people will understand

There is some humour in calling them green green circuits or blue red circuits though

[terse lagoon]

Dunno, i either use an emote or call it Q5 (for )

[jaunty citrus]

Verbal conversation will be different than text, but yes for text Q5 and work quite well

[terse lagoon]

Agree

[unborn flax]

First attempt at a 'simple' quality build to produce just under 1.4 from 240 /s. To keep quality reasonably high and build size reasonably low, I'm only using a speed module in the processing.

I suspect it may have jamming problems, though I'm more interested in the ratios given I'm doing this blind. The RHS loop needs to sustain 61.8 items/s so I'm relying on the recyclers outputting the occasional stack of 2.

Visualisation is using @daring siren 's Janky Quality and cut'n'pasting from FFFs.

[frank warren]

@unborn flax One thing you should note: you can actually achieve reasonable compression ratios with the recyclers. As the backup fills up, the recyclers will fill more - this will cause the recycler to output more the next time it is able.

[unborn flax]

I don't get what you mean by compression - could you elaborate?

I'm relying on them helping a bit with belt compression by sometimes outputting a stack of 2 iron plates - I guess this happens 1/16 of the time.

[frank warren]

Yeah, I am saying you are dramatically underestimating the stack sizes.

A recycler which cannot output will buffer items in it's inventory.. allowing it to output a larger stack next time it can output

so the system naturally can reach higher stack sizes.

[unborn flax]

I see what you mean - that is a useful property. Though from experience of other builds, in this case I think more output stacking translates to "takes longer to jam" rather than "less likely to jam".

[frank warren]

Most recycler setups work better with a chest somewhere, that's one of the reasons I am avoiding belt-based recycler loops.

[burnt reef]

"less likely to jam" and "takes longer to jam" mean the same thing. according to the Law of Very Large Numbers, given sufficient time, anything that has a constant chance of happening will happen, eventually.

[unborn flax]

This is language not mathematics: by "less likely to jam" I mean less likely to jam ever, i.e. more confident that it's jam-proof, given incomplete information.

Jamming is also the least interesting thing about the build given we can't currently measure it.

[dense kraken]

I wonder, if you bluid a recycler outputing to a chest which feeds a stack inserter, can the inserter become stuck waiting for 15 other items if it ever gets a high quality one?

[unborn flax]

I think that could happen if the chest fills up due to buffering.

In the build above I try and solve almost the same problem (though without chests) by having a stack inserter for the majority output item and a fast inserter for all items. It felt sensible at the time.

[dense kraken]

oh right, I forgot about the chest's buffer

[frank warren]

I believe these builds can always jam.

[burnt reef]

unless you have it set up to destroy excess byproduct. storage is always finite, and the discrepancy between various recycled components can become arbitrarily large, so any buffer you create will eventually become overwhelmed. but if you destroy returned items beyond a certain point, that won't get backed up.

(... it might be possible to use Recursive Blueprints to create a self-expanding buffer system, but even then the space it can use is ultimately finite. unless you're using something like Factorissimo.)

[terse lagoon]

U forgetting u can just dump items into lava/space ot recycle them to get rid of

[jaunty citrus]

That would be destroying excess byproduct

[opal haven]

it's fine as last line of defense. but you can always re rout some of the quality resources to science

[charred gyro]

For anyone interested, there's a poll on reddit regarding the quality names: https://www.reddit.com/r/factorio/comments/1clq5fd/some_time_has_passed_are_you_ok_now_with_the/
It has about 800 votes so far.

[frank warren]

I hate how this poll is missing the 'I don't care' option

That is my actual opinion. I would support them changing. I would support them not changing.

oh my god I am blind

[hoary current]

[frank warren]

I will leave this here as a testament to humanity's ignorance. - and as a testament to my own humility.

[opal haven]

it's a bit surprising to see so many people wanting to change names

[frank warren]

My real opinion is more of an 'they are a problem, but I don't have a solution'

It's a really well-recognized 5 tier system...

[tall sandal]

I feel like its not actually that hard to change the names. Maybe a day's work? I dunno maybe I'm underestimating it

[tall sandal]

Yes, they are the best names I've seen so far

[red kelp]

at this point ive internalized the names and they sound fine to me

[tall sandal]

I get not liking the names but I think they are best taken with a grain of seriouslessness and humor

[charred gyro]

I'm not sure I understand that comment from Rseding91. If we need to use something like quality.epic to refer to that quality level, how would we reference a hypothetical sixth quality level above legendary?

[red kelp]

so definitely in the "i don't care" camp

[frank warren]

It still feels crazy to me to say an EPIC INSERTER

It feels off-theme

But every alternative proposal I have seen is garbage

[tall sandal]

I think it was kovarex who said he doesnt take it seriously and laughs at that kind of stuff

[red kelp]

i mean you can refer to it as q4 and people will know what you're talking about

so its not like you have to use the names

[opal haven]

why not? it would make factorio stories sounds kind of cooler

[red kelp]

in fact i will almost certainly mix up epic and rare

[frank warren]

why?

[red kelp]

i dont have a good answer for why I will forget something

i wont really think of the names very often and just look at the dots

[tall sandal]

based dot looker-atter

[obsidian crescent]

If you create your own quality level, you can name it whatever you like. But the point is that they want the default displayed names to match the actual under-the-hood names. So if they change the display text, they would change everything internally to match.

[frank warren]

it took me a second to get this. I couldn't for the life of me figure out what it had to do with the other dot.

[charred gyro]

Several quality-grading scales use the terms "good" and "excellent". They're used for both natural resources like gems and manufactured items like cards. Why not use those? Then we can use "normal" or whatever for the first level. For the top of the scale, "flawless", "perfect", or "ideal" all seem fine. That leaves just 1 remaining level.

[tall sandal]

propose it and I'll add it to my list of options for a mod I might make that's loaded with preset alternative names

[opal haven]

cuz it will be easier to forgot or mix qualities

[charred gyro]

Do we need to make proposals in a particular way? I figured the devs would just do a quick search for the topic and read through posts/threads about 'em, looking for terms that people prefer.

[obsidian crescent]

Why bother with presets? Just give the player the option to pick a different name.

[tall sandal]

I mean write it down and send it as a message in this channel, then I will add it to a google doc. Then when SA comes out, I may make a mod that uses your proposal as a preset

[tall sandal]

because most people are not creative. you'll be able to customize it too

my list has ideas you'll never think of

[red kelp]

can we see the list

[charred gyro]

Okay, for posterity then. Instead of Common, Uncommon, Rare, Epic, and Legendary as quality names, I suggest:
Normal
Good
Great
Excellent
Ideal

[red kelp]

ideal doesn't feel better than excellent but that might just be me

[tall sandal]

my favorite item is

Cringe, mid, bussin, rizzy, goated

[red kelp]

rizzy lmao

that's a good one

[sage vapor]

Perfect

[charred gyro]

That'd also be fine

[sage vapor]

is the thing I'd change on that lineup of names

I need to collect all the different variants people have proposed

[red kelp]

if you make this mod i will download it day 1

"pass me that goated inserter, will you?"

[tall sandal]

hehehe

[opal haven]

I like this one - it's properly embracing the silliness of the game

[frank warren]

@tall sandal can you bring me 5 rizzy inserters at the mall? our gear throughput is not so good here.

idk, it kinda works!

[red kelp]

those inserters have mad rizz

[tall sandal]

hehe

i kind of love it

[charred gyro]

Those inserters are so charismatic

They just have a certain way of handling things

[red kelp]

dude this circuit block is cringe we gotta build a new one

[dense kraken]

your proposition is
now let's see paul allen's card mine:

• normal (the baseline, with no negative connotation)
• good (obviously better than the baseline)
• superior (better than most)
• exceptional (convey how unlikely it is to get this good)
• perfect (cannot be better)

[median verge]

its less of a day's work and more just tracking down every instance and reference to it and changing it. doable, but they don't wanna do it seven times so they're waiting until someone finds a really good alternative to the current system

[tall sandal]

I guess the push back is that its not "trivial"

[median verge]

yes, not trivial

their search function also runs on the internal names, which is why they mention train stops

[charred gyro]

It does make me wonder how it's implemented if it's not just a localization thing. Maybe they have multiple ways to refer to quality levels? Both quality.level == 3 and quality.name == epic perhaps? If someone makes a mod which names the quality level between epic and legendary (quality level 4) then can other mods refer to it by name? I would've expected the quality level to be the identifier rather than the name itself.

[dense kraken]

speaking from experience, it's both

for example you have an item prototype named iron-ore and the locale files converts that to Iron Ore

so you can easily have a mod that justs converts legendary prototype name to the locale Perfect or something

that's how translations work anyway

[tall sandal]

but then the internal name will still be "quality-legendary"

[charred gyro]

Oh. If it's following exactly like other items, then it'd be more like iron-ore-epic then?

[dense kraken]

yup

[dense kraken]

very unlikely

I mean, the quality levels have probably their own prototype

[sage vapor]

yes

quality is entirely independent from recipes / items / entities

which is how it can scale so easily

[dense kraken]

like

{
  name = "legendary",
  type = "quality-level",
  ...
}

[tall sandal]

my understanding is an instance of an item has a quality. so an item might look like

{ 
  name: "iron-plate" 
  quality: { 
    name: "quality-legendary" 
  } 
} 

[sage vapor]

internally in the game state yes

but not in the prototypes

[dense kraken]

and in the locale

[quality-level-name]
legendary=Perfect

[tall sandal]

hows it look in the prottype?

[sage vapor]

item prototypes don't know about qualities at all

its separate

[dense kraken]

they probably have behaviors depending on the quality

remember that each quality level has a scaling value

[sage vapor]

yes

1 fixed value that then scales specific things on the engine side

[tall sandal]

I'm not sure how the scaling works exactly

[dense kraken]

so you could say that the productivity module has effect of "+25% * quality_factor" in the module proto without ever having to care how many quality levels there are and how they scale

[tall sandal]

can you specify what quality does? for example if you wanted quality beacons to have more range, could you do that?

[sage vapor]

unknown if we can affect what qualities will change

we do know for certain that we can add more quality levels with different multipliers

or remove levels

[tall sandal]

are negative levels possible?

[sage vapor]

no

[tall sandal]

rip

[sage vapor]

1-254

0 and 255 are reserved iirc

what would "negative" do anyways

just reduce the initial values

[tall sandal]

yeah!

[charred gyro]

But how would you acquire them? Are you thinking of a mod which would change everything to have a native chance to output different qualities, rather than needing quality modules, and then making the worse "negative" qualities increasingly less common? So you often get stuff that is a little worse, and rarely get stuff that is useless?

[tall sandal]

my understanding is you can say the odds of advancing a quality. my understanding is you could even have diverging quality paths, where you have 10% to go to "legendary" and 10% to go to "alien" (and its impossible to get to "alien" from legendary)

[dense kraken]

maybe something like that
(assuming you want to add 2 tile to the range per quality level)

{
  name = "beacon",
  type = "beacon",
  ...
  supply_area_distance = {
    base_value = 3,
    scaling_factor = 2
  },
  ...
}

[tall sandal]

that would be cool!

[dense kraken]

and the engine would detect that it's an object instead of a value, thus it must be a quality-depending one

a bit like all the different ways to define an icon

or expensive recipes in 1.1

[tall sandal]

I wish it could work like

supply_area_distance: func compute(int quality): blah blah

[sage vapor]

yeah not gonna happen

[tall sandal]

🥺

[charred gyro]

So are we expecting to have fields for starting value & additional value per quality level, rather than a table which lets you define different rates of change, like how the beacons will work with diminishing returns?

[sage vapor]

I am expecting not being able to define anything regarding quality scaling on an entity affected by it apart from maybe disabling it

[charred gyro]

I feel like they wouldn't need to skip quality level 4 if they had tables which could define each level. So I think the former option is more likely.

[dense kraken]

I am not expecting a table, because not having any would allow any amount of quality levels

[charred gyro]

I'm not expecting we'll be able to choose which properties are affected by quality, but we'd probably need a way to adjust how much each level adds, right?

[dense kraken]

nah, they obviously move that stuff around a lot during dev

knowing wube they must have made this easy

[terse lagoon]

There is quality position (1, 2, 3, 5) which is it identifier. There is quality multiplicator (mb different name) that also (1, 2, 3, 5), and there is name, which is just name

[daring siren]

Quality is a property of item stacks

[tall sandal]

I read that in bill nye the science guy voice

https://www.youtube.com/watch?v=HDhnx0c1w70

[dense kraken]

https://www.youtube.com/watch?v=edXOIJZrLPo

[tall sandal]

I never really liked that video. Parts are funny (the avengers part) but I think its a decent into depressed nihilism. the truth is much more nihilistic. in his goal to find out what inertia "really" is, he forgets that inertia is just a model for reality and isnt real. none of the formulations are fully correct

in that sence, p really does equal mv and that is just as valid as some wavefunction

[dense kraken]

alternatively you can also ignore your existential crysis for a sec and have a laugh

[frank warren]

I do not know how to feel about that video, but it did make me feel something.

[tall sandal]

its the experience of someone who misunderstood science

[sacred totem]

its actually kind of interesting seeing all teh ways someone can make a momentum

also, i would like to point out the the first time i ran into the symmetric inertia matrix was in the easiest class of this semester

never touched it outside of "this is where it is used" though

and i dont think its used for linear momentum...?

[burnt reef]

I find the juxtaposition of "this is a descent into depressed nihilism" and "the solution is more nihilism" here very funny. I largely agree, mind you, it's just amusing.

[tall sandal]

hehe i added that as a joke, glad you caught it

its funny AND true

[sacred totem]

hmmm... is inertia "real"?

i would say that is is, because things have inertia

[tall sandal]

yes, inertia is real

[burnt reef]

is it "real" if different observers can disagree about an object's momentum, and both be correct?

[sage vapor]

its "real" as long as it has no imaginary component

[fast gull]

Given the new beacon changes is there a discussion i can refer to in how it impacts quality decisions?

[opal haven]

quality beacons are OP now, but that's about it IIRC

there's no other changes to quality from beacons

[daring siren]

Basically beacons are now one of the best things to focus the quality production on

The quality priority for me is: Quality modules, beacons, other modules, buildings

[opal haven]

not sure beacons this op, but having 2 bonuses from quailty is quite convenient

[daring siren]

They 1.5x beacon power, on many machines

[opal haven]

beacons grow 20% more power from quality with 150% base power
modules grow 30% more power with 100% base power
beacons cost only small in comparison with t3 modules
so my priorities will be something like:
Quality modules, other modules t1, beacons/production buildings (whichever cheaper), other modules t2/t3
where to put t3 prod modules kind of depends on situation, not sure for now

[daring siren]

20% over 150% which is less than 20% change

[opal haven]

I mean 20% addictive
150%->170%->190%...

idea is t1 modules grow significantly faster from quality (considering the cost ofc), so better to get them first

[daring siren]

Yea me too, just noting the bonus is a bit lower, but still applies to many buildings around it, so super strong

We're going to go through many t1 modules while making t2 and t3. Then the question is... Do you use them as a good t1, or try to roll for better t2s?

[opal haven]

I probably will use them as good t1's , especially prod modules - blue chips/high tier sciences will appreciate a bit more productivity

[obsidian crescent]

I plan to do neither. I'll be making my early quality gear from intermediates built through ores and furnaces with quality modules in them. This includes coal for quality plastic.

[daring siren]

You will put in your machines still though

[opal haven]

getting higher tier modules before visiting fulgora sounds suboptimal anyway

[obsidian crescent]

Yes, but that's just for an occassional bonus.

[daring siren]

In the module assemblers I mean

We're going to have a lot of from

[opal haven]

that's the ones I'm 100% will use - easy pay off while I'm in space

[obsidian crescent]

Don't most of the infinite productivity techs use purple science too?

[daring siren]

We're not going to do infinites at this early stage

The whole priority listing I suggested is for midgame

Eventually? Yeah sure

[obsidian crescent]

Yes, but I'm talking about when you actually leave the planet. That gap of transport time + initial setup on a new world is a perfect time to build up a small stockpile of quality prod 1s and quality furnaces.

[daring siren]

We don't consume anymore though, because no

Yea good call on

[opal haven]

that one is a bit of a shame - will be nice to get some along the way

[obsidian crescent]

Yellow science, assuming its recipe is unchanged, looks like a pretty good way to get some quality bots though. Makes a nice sink for non-quality robot frames.

[opal haven]

adding some might actually be nuts - bots can't pile up on one roboport like crazy if their flying distance will be a bit different

[obsidian crescent]

They've already made AI fixes to minimize bots crowding a single roboport.

[opal haven]

true - I think it will still improve crowding on top of what fixes can achieve alone

bots not gonna search for free roboport as much

[jaunty citrus]

You can also distribute bots evenly through a base using roboport requests

[iron root]

in order of returns per quality upgrade its basically modules >= beacons > ??? i think now

the >= is because modules and beacons multiply together, so as you quality 1 up, qualitying the other becomes better and vice versa

and its close enough that when both are , modules are better but the moment you throw in an module, getting the beacon to gives more returns than upgrading the other module in it to

so theyll wanna be boosted hand in hand

[opal haven]

it depends a bit on tier of the modules, t3 has huge cost in comparison to beacons

[jaunty citrus]

you can get some through crafting , not quite the same but it's something

[daring siren]

Anything that isn't sinked into science/rockets has a limit on it

That said, when going for T3s, a lot of T1s and T2s are created

[fast gull]

One thing i liked with space exploration mod was the alien areas that had top tier modules. If you got those before you had them set up, it was fun choosing where to place the new module.

Would be cool for something similar in the xpac, as you explore space with the platform you can find wreckage with quality mods :3

Probly for a mod instead of base game

[daring siren]

The "Story Missions" mod campaign does things like that. It's nice to reward exploration

[tall sandal]

I thought it was very very cool, but it pretty much went like: okay I'll put the production one in labs, speed and eff ones I dont really care about much

[iron root]

the big speed mods you get make a comeback in mid-lategame when you wanna juice a lot of high-value recipies

but they arent as immediately useful as prod for sure

i made a lot of good on those t9 speeds when I made a gigabeacon to throw all my t7-t8 prod mods under for space material postprocessing/naq refining

the efficiency ones.... ehhhh, the best use I found for em was out in the asteroid fields where it's a bit harder to power things

SA could have similar applications though. big speed modules will be used to leverage big prod modules (one might actually do this with labs, because they dont have the innate 10x speed SE space labs do so beaconing them will be important)
big efficiency modules will be really strong in power-restricted areas like platforms

[unborn flax]

The quality modules make low-energy builds much more viable than in 1.1.

Kovarex posted an EMP build with 5x and 8 beacons full of (everything ). Relative energy per craft is 14.90 / 89.65 = 0.166

If instead you fill the with 7x , 6x , 3x then speed drops to 39.91 and energy drops to 20%, so relative energy per craft is 0.2/39.91 = 0.005

So this is 44% of the speed but 3% of the energy per craft. This is assuming that the eff module values are as in 1.1 (with quality bonuses) and it ignores the unknown energy required for the beacons. Though we do know that higher quality beacons consume less power.

In the corresponding situation in 1.1 (say an full of with 8 beacons) there's no real choice: if you want speed then put in the beacons, and if you want low energy then put in the beacons.

[daring siren]

In 1.1 power is free at that stage. In Space Age, power isn't free on platforms and some planets.

Indeed that makes in beacons attractive.

[unborn flax]

I like the idea that a build on (say) fulgora could by necessity look different from the equivalent build on nauvis

[daring siren]

That's what they are going for, and I love it.
Hopefully it's like that with enemies/defences as well

[obsidian crescent]

Considering Fulgora's unique resourcing setup, that's pretty much guaranteed.

[daring siren]

Also unique power setup

[unborn flax]

Even a single beacon with a is crazy (for energy consumption): put 2x , 2x in the machine and 2x in the beacon. This gives 7.25x the speed and 2.8% of the energy consumption per craft, compared to no modules.

[pure spire]

Isn’t there a minimum energy consumption still?

Like 20%?

[neat shard]

Yeah but if you have 20% energy consumption and 7.25x speed, you need 7.25x fewer assemblers, which takes 7.25x less power

20% / 7.25 = 2.759%

[pure spire]

ahhhh per craft I get what you’re saying now

yeah

[unborn flax]

Similarly quality essentially removes the disadvantage of some speed modules, in that if you have an empty machine, there is no energy penalty (per craft) of inserting a or or

It's true when (speed gain) * (quality multiplier) exceeds the energy penalty.

So if you have a machine with an infinite number of module slots, as you insert (say) many , the energy/craft approaches 70 / (50 * 1.6) = 0.875 which is less than 1

[unborn flax]

Graph of the 'best' low-energy builds for a 1-beacon assembler3, given various combinations of unlocked modules and quality tiers. Both axes are logarithmic. The cost function assumes with EMP quality-cycling, so some of the early-game builds may be a bit more expensive than necessary. This doesn't affect speed though.

Some of the rounding may be a bit off as I haven't worked out the exact order of operations and truncation in calculating beaconed bonuses.

"s2e3q5 | 123" means "the fastest energy-efficient build when you have unlocked , , has 1.23x the speed".
"bq2" means , and "s2q4" means .

[unborn flax]

The 7.25x from yesterday is in the top-right of the graph - the cost function worked out that you can make do with + in the machine instead of the more expensive 2x

Assuming my cost function is correct for the single-recipe-recycling quality model: in the endgame gets obsoleted by , which is better and 80% of the cost (this is the only such case)

[half raptor]

Considering the efficiency cap, that would make sense

[obsidian crescent]

That's not possible. We don't know what the recipe for prod 3s is in SA. We know that the recipe for qual 3s involves superconducting wire (in addition to red and blue circuits). But there's a good chance that each module comes from a different planet; if that's the case, then each one likely uses a different intermediate.

[daring siren]

And Eff3 likely uses bluranium

[unborn flax]

If you're asking how many normal prod3s you need, the answer is 64.6. But this means that you didn't take advantage of quality when making your prod11 or prod2s, which is a waste of module slots. To work out the 'real' best answer you'll need to both know all the recipes, and be able to produce quality intermediates for each ingredient of prod1/prod2/prod3.

[neat shard]

Can I see the math that goes into that 64.6 number?

[unborn flax]

I've posted a couple of versions of my linalg script; basically you write down the transformation matrix, then find an eigenvector for eigenvalue 1 in order to compute the infinite sum. Other people have come up with similar numbers with numerical approaches. Here's the output for the case you care about: (edit: my bad, the 64.6 is actually 45.1.)

legendary outputs are 336650780594603581568/7456286275322196443 == 45.1499269426933 times the expense of a normal output
    tier 0: prod * 0, qual * 5
    tier 1: prod * 0, qual * 5
    tier 2: prod * 0, qual * 5
    tier 3: prod * 0, qual * 5```

[neat shard]

Fair enough

[opal haven]

where does the idea of "prods into intermediates, quality into finished products" even come from? the more I think about it, the less sense it makes (if you wanna quality loop anyway). sure, it's simpler and cheaper to build, but the efficiency seems atrocious and just putting everywhere before actual loop and mix of at the loop just seems to be MUCH better option

[daring siren]

The idea came from the question of prods vs quality. Of course, if a recipe can't get prods, we'd want to qual tehm.

With and a recycler, the optimal is 2 2 , by a huge factor

In terms of simplicity, putting into intermediates saves a lot of complexity and quality imbalances

Additionally, with we get more items, but of lower quality

[opal haven]

true, but if all intermediates have 4 - it's basically a way to skip ~1 loop on quality looping and not pay 4 times as much once (for the cost of losing +100% from prods, which is still good value)

of course with different builds that do stuff other than simple quality looping priorities might be different, but i'm recalling in intermediates were discussed in looping builds (iirc)

oh, and with quality imbalances - is there any cases that can't be solved with recyclers and mixing small amount of into intermediates?

[daring siren]

Bonuses from prod are larger than bonuses on quality

It's almost always better to have more prod than quality

[opal haven]

hm, this one is incorrect - if final product is recycle looped, skip will be ~0.5 instead of ~1 loop (due to~ 8 )- this one might derail all subsequent math and make difference quite less significant

[unborn flax]

Say we're building prod3s. Two possible strategies are:
Strategy A: Quality-cycle all your ingredients to legendary, then have no modules in the EMPs making prod1/prod2/prod3.
Strategy B: Fill your prod1/prod2/prod3 EMPs with qual modules, start with normal ingredients then quality-cycle at the prod3 level.

A has better quality-cycling efficiency as it can use prod modules, but B has the benefit of 'free' quality steps. I think the optimal (but complex!) solution will be a combination of the two: run a limited version of strategy A to quality-cycle your intermediates up to a particular tier, then run strategy B and cycle up to legendary in the prod3 step.

[daring siren]

The problem with doing more quality steps, is that we need additional quality ingredients, so everything has to be quality'd

[unborn flax]

I said it would be complex 😕 You'll need to be able to produce intermediates of all qualities, yes.

[rotund egret]

With the cost of modules and quality machines, having to scale production for each specific quality might prove a bigger hassle than expected

I'm a big fan of the mixed approach above for stuff you want to mass-produce but can't prod all the way through

to be fair, all 3 approaches have their own benefits:

A. for end-products with ingredients you're already making a ton of and don't mind the loss of efficiency (e.g. Nauvis mall)

B. for stuff that's a bit isolated and making a quality chain is more trouble than it's worth (e.g. Vulcanus Foundry production)

C. for expensive stuff you want a ton of (e.g. Q5T3 module production)

[daring siren]

That said, intermediates can be skimmed from science in some cases

For example in we'll want to put in ♻️ so we'll have intermediates in a reasonable count

[opal haven]

i'm planning to do purple on nauvis asap, and getting basically for free is one of the key reasons

[daring siren]

that's the easy step

in general all final items

[neat shard]

@north yew might have something to add here

[north yew]

how much am I scrolling up

[unborn flax]

With , I think we can get some of the way of working out the benefit of two previous free quality steps from the and EMPs.

In the EMP, suppose inputs are normal quality, so [1 , 0 , 0 , 0 , 0 ] as a vector. Then we need 30.1 sets of inputs (prod2s etc) to make one legendary output (this is a correction from earlier, hopefully no-one noticed).

Suppose instead that our input is the unit vector [0.4761 , 0.38502 , 0.116343 , 0.0194184 , 0.0031186 ] - which is what you get when you put 1 normal item going through two EMPS full of . Then I believe you only need 10.5 sets of mixed-quality inputs per legendary output.

This isn't an exact science as at the and stages you still need to quality-match the other (non-module) ingredients, which comes with its own cost. But the saving of ~x3 looks most useful.

[opal haven]

I like it exactly because it's easy step. and is just profit printer

[north yew]

B does not really have the benefit of free quality steps. T1 module is bound by the ratio of green chip, T2 module is bound by the ratio of red chips. T3 module is bound also by the ratio of red chips

unless you plan to use quality T2 modules there's no point in putting quality mods in those machines

[daring siren]

And then it's sad we lost because it loses us a "free" ...

Though we can just recycle loops these 🙂

[opal haven]

yeah, it's one use of i will actually miss)

[opal haven]

get T2 modules with quality and recycle loop a bit other ingredients of T3 modules -> profit

[daring siren]

Also, please consider doing some extreme recycle looping early to make further looping more efficient

(early after that is)

[north yew]

you don't get as much quality as you think from putting quality modules in a production line. when quality modules are used in a crafting step, that changes the ratio of high to low quality from the inputs to the outputs. the more crafting steps an item has gone through, the more high quality there's gonna be. the length of the chain matters. the problem is introduced because most crafting recipes will have some ingredients with longer chains and then one or two ingredients with very short chains. in that assembler, any quality rolling that happened on the longer-chain ingredients is completely wasted, the recipe is entirely bounded by the shortest chain of its constituent ingredients.

like the example I pointed out earlier: T3 modules and T2 modules both use red chips, so rolling quality on the T2 modules is entirely wasted

the problem repeats itself over and over and over throughout a supply chain and ends up wasting most of your quality rolls

[daring siren]

You are correct in the bounding, but it can be balanced by configurably not using quality and/or loop the shorter chains, considering they are cheaper.

[north yew]

sounds like a nightmare to set up

[daring siren]

i.e. something we'll enjoy watching Dosh doing in a video 😄

[north yew]

lol true

[opal haven]

sure. that's why I pick simple example
all T3 modules ingredients can use 4 to get the same ratios and extract simple profit

[daring siren]

Looping cheap items is also a good option though

[opal haven]

going deeper will surely be more challenging

[daring siren]

Considering , the T3 modules will use 5 modules in an EMP 🙂

[north yew]

on paper yeah looping the cheap items to improve their ratio sounds great but I'm trying to fathom how you actually get that working properly in a factory and it sounds like a very bad plan

jamming a bunch of prod is much simpler and also makes items extremely cheap and therefore recyclable

[opal haven]

it's difficult, but profitable. i'm planning to at least give it a try

in small simple recipes it will be simple enough

[daring siren]

Agreed. Prod has a 4x larger bonus compared to qual.

[north yew]

I'm not exactly sure what the curve looks like on repeated quality rolls, but I know that prod is a steep exponential curve

[unborn flax]

You're only saying this is hard (which it is), not that the theory is wrong. No-one is suggesting you should mix with other ingredients.

[opal haven]

qual will save you some recycle looping, if your goal is to get more quality - it's great
if your goal is to get more stuff - sure, prods are the best by far

[opal haven]

i think it's just exponential as well

[daring siren]

The balance between and is important, especially with +50% built in

[unborn flax]

Basic game mechanics like splitter priorites can help produce a factory that produces (say) of all qualities. You don't need to hardcode the probability distribution you want into the factory design.

[daring siren]

Is it better to recycle loop final items (with full qual) or intermediates (with full prod and/or mixed qual)?

Assuming you want that specific item/intermediate

[north yew]

well it reaches a point where we start getting in to game strategy. I'd be willing to bet that if:

• player A just uses prod and recycle loops at the end
• player B tries to set up all of these fancy, complicated approaches to squeeze out a little more quality
  player A is gonna get a lot more legendary stuff a lot faster. player B wastes a bunch of time making a convoluted setup based on complex math meanwhile player A's setup has been chugging the whole time

[opal haven]

mix of strategies A and B from here is probably the best

[opal haven]

player B can buffer stuff and overbuild quality build tho))

[daring siren]

Not to mention, some extreme recycle looping early will quickly pay for itself

e.g. getting 1 that affects 8 with expensive modules is cheaper than more expensive modules

[unborn flax]

That's fine - and I don't even disagree - but it's a distraction from a theoretical analysis. Saying "if you're not careful you may end up playing Factorio for a long time before you win" won't persuade me to not try things.

[daring siren]

Even with 2 , this is a huge speed boost

[frank warren]

Depends. I thought this too, but JG told me it wasn't always the case.

[daring siren]

Also, we don't know if there's some super simple way to do all those complex chains

Last time I trusted JG he unconfirmed my whole existence

[frank warren]

Fair, but I think they were opting for speed modules.

[opal haven]

bots will crazy simplify a lot of complicated quality setups

[daring siren]

Speed modules in recyclers sounds nuts to me

[frank warren]

I think it is interesting at least that you cannot use prod in them, even for that recipe.

I hope it is kept that way.

[opal haven]

yeah, current state of fulgora is quite unique and interesting

[daring siren]

We haven't seen and Aquilo yet... but we should assume some other interesting mechanics

[daring siren]

How fast are BMD compared to recyclers? Can they be easily 1:1ed?

[obsidian crescent]

Mining productivity messes with those calculations. And it'll probably require a lot of productivity before recyclers are used at a 1:1 ratio, since the recycling recipe for scrap is something like 0.12 sec

[daring siren]

and though

Unless we choose to instead

We can in the BMD and in the recycler to slow it down

[unborn flax]

i corrected that number to 45.1, though this hasn't been double-checked by anyone else

[frank warren]

that number is ~1024

[uncut helm]

Actually I've done that.

Recyclers are slow and space on fulgora islands is limited

Also speed modules are cheaper to make then recyclers