Quality

[tall sandal]

absolutely masochist, you'll have a fun time with that

[north yew]

you kinda breezed past the logistical complexity and scale problems, but I think you should take some more time to think about them

you already need a pretty bananas number of assemblers for , for example

now you can't use prod or speed beacons in them by the time your base gets running

[gritty umbra]

Things I never use, so that's not a limitation.

[north yew]

so you're looking at assembler counts in the triple digits

why don't you use prod or speed beacons

[tall sandal]

the good news is the high quality blocks can be a lot smaller

[gritty umbra]

I dislike beacon mechanics for being magic radio broadcasts, and I just never use productivity.

[tall sandal]

like, you'll likely only need 1 or 2 assemblers for legendary advanced circuits

[north yew]

oh my

okay now I get why you're hard to convince

you haven't seen the light of productivity

prod modules are very, very strong

[tall sandal]

and if you are going hard on quality, you'll also have fast assemblers just because

[gritty umbra]

Also very messy.

[north yew]

yes that's the tradeoff, but there's a reason they have such a large penalty and are still the most used modules in the game

also quality will be 1000x more messy

[tall sandal]

you can even sushi your qualities, so your normal though legendary assemblers can be contained in one production block

[north yew]

fun fact: in the pay for themselves before you even launch a single rocket

[gritty umbra]

Not massively so. It's just five copies of a block and splitters for the inputs and outputs.

[tall sandal]

so just five times more messy :)

[north yew]

it's increasing the number of connections by 25x

5x as many sections for each product, 5x as many possible destinations for those products

[gritty umbra]

It's only one destination for each quality, not five. The input for that quality in the next block.

[north yew]

there's five different places that a common item can go after being crafted b/c it can be turned in to five different qualities potentially

[gritty umbra]

Only one. The common input line.

Any higher quality outputs would end up on the higher quality output belts.

Twenty five outputs is also wrong even if you count the possible qualities from each copy as going to different outputs. At most it's fifteen.

[north yew]

another fun fact: for a crafting recipe that's five steps away from furnaces, if you assume all of those steps have 4x , then that means you only need about ~19% of the drills/furnaces running as it would take for that same recipe without any prod

[gritty umbra]

Q5 inputs can only produce Q5 outputs, Q4 Q4 and Q5, and so on.

[north yew]

fair point. it's only a 15x increase in the number of connections. very doable.

[gritty umbra]

And even that is an overstatement, since I'd be filtering it down to five outputs.

And doing that wouldn't require any more than four splitters.

[north yew]

and 5x as many belts and all the belt spaghetti that would take

main bus with six lanes dedicated purely to different qualities of

[gritty umbra]

For that I'd just use trains to connect the blocks.

[tall sandal]

you can sushi your qualities

[gritty umbra]

Can, but never would.

[tall sandal]

y not

[gritty umbra]

It would require delving into circuitry that I don't want to deal with.

[north yew]

sushi only works if it loops back around. you'd have to draw a big loop around all of the things that take that product

otherwise you'll eventually hit a jam

[tall sandal]

i believe you that it would jam but im noy visualizing in my head why it would

[north yew]

inserter at end of belt needs [specific thing], but the only [specific thing] is behind other products on the belt

eventually happens to all of the inserters over time

sushi needs to constantly flow in order to make sure that every inserter gets access to every thing even in cases where bad luck throws the ratio off

[tall sandal]

I mean your bus would have all qualities on it in 1 belt. when you pulled from the bus you'd priority split just that quality off

[north yew]

if that belt is backed up and the only items in the splitter are of that quality, the whole belt stops

if there's a scenario where the only way to make the item causing the backup get consumed is for that belt to flow to another product, it's a deadlock

[tall sandal]

but you could loop that belt back to the bus if you wanted

[north yew]

same thing happens it just takes a little longer

the reason you learn as a newbie to generally keep belts (or at least specific belt lanes) as their own product is because it causes backups any time it hits a case where consumption doesn't perfectly match the belt. eventually something gets starved even though it has products in front of it.
sushi manages to subvert the typical logic of one item per belt/lane because it doesn't have an end. it never stops. if an inserter can't grab the items in front of it, that doesn't matter because more items will just keep shuffling through until the inserter finds the one it wants.

if a belt does eventually end, then the only way that new products can move on to the belt is if the ones already there get consumed.

[half raptor]

The 5 phases of ideas

1. Wake up with a new idea
2. Figuring how you'd use it
3. Recognizing the flaws
4. Realize you're getting more complicated than it needs to be
5. Realize that is the way and go to sleep

[north yew]

yeah it do be like that

[half raptor]

Today's idea that I'm workshopping is instead of sending all overflow quality items to work as lower quality ingredients replacements, go with the original idea and recycle any overflow and allow it to build upwards. However we can use any eventual buildup of legendary items as overflow for previous steps to pair up in epic lines (obviously the machine would go with lower quality ingredients), then rare etc. This way you can go for quality upgrades as you start accumulating ingredients and you have a release valve at the legendary level.

If you have too many legendaries, you're doing good

[north yew]

I mean I think the way to look at it is that prod is the shoe-in candidate for intermediates and quality is the shoe-in candidate for buildings and other final products

[north yew]

and that this is 100% okay

excess of high quality items means unnecessary quality rolls happened

[half raptor]

True enough, but considering that I'm always sending my products to either deliver or recycle, I should always be crafting, making a resource hungry base and as long as it's hungry I would have little reason to recycle. This concept is to automatically solve the almost certain imbalances until I can get around to building a new block to balance it again

I'm going for a large input buffer before the station is satisfied, so my recycler would be idle most of the time until imbalance starts to show

I'm fine with chaos as long as I have safeguards that will handle overflow. My oil design reflects this. If I start accumulating too much petroleum clogging the outputs, I cant get any more light or heavy oil, so I set a release valve of petroleum to solid fuel that feeds to my rocket fuel. This clears the outputs long enough for light and heavy oil to come again

I can deal with imbalance if I have a release valve

(Also with researchable prod bonus for intermediates, chaos will become the norm)

[tall sandal]

Its very exciting that the optimal solution is very complex and dynamic.

I think that people wont be able to play quality optimally -- its too hard. So I think there will be a lot of rule of thumb solutions. I wonder what they will be.

[half raptor]

My rule of thumb I want is "If drop off station is destination full and time elapsed is 10 sec, go to recycle drop off station" and "if legendary drop off is destination full and 10 sec elapsed, go to epic overflow, then rare overflow, and in those overflow situations it'll mix the legendary with lower quality ingredients which will functionally treat it as a lower quality.

That way things always migrate toward legendary and legendary has a means of "recycling" if it overflows

[kindred crater]

and that in itself is another point of discussion- what is optimum:

• material efficiency?
• getting the desired item(s) fastest?
• minimal time setting it up?

[tall sandal]

ease of playthough

[kindred crater]

unless you're working with refined/proven BPs, it seems like you can only pick 2 of those 3

[half raptor]

What about "fun"

[kindred crater]

and that too lol

[iron root]

depends what you define as 'material efficiency'

[daring siren]

Then it's no quality modules, except in either final items (+recycle) or miners (and store).

[iron root]

my reading on that is just maxing out the use of . If you can beacon spaghetti you can get a bit of all 3 pretty well

for quality material efficiency... yeah the 2/3 makes more sense

[daring siren]

Basically, the % on prod is higher than on quals

[iron root]

a lot of simple 'no waste' setups (quality at ores and not much else) will be pretty slow burning

[daring siren]

Enough until we get to I guess

[kindred crater]

probably minimal recycling, since you're losing 75% of material

[iron root]

yeah.
Also wondering if 'time setting up' can be changed to something better, because setup time isnt that relevant once you have a blueprint

so likely design time? Which can get pretty nutty

[daring siren]

A neat way to use recycling is to put on all mall items, and recycle the ones that got quality, but don't get interesting bonuses. It gives quality ingredients at a fixed cost of 4x

[iron root]

or if we talk the timespan of the entire game, how much tech you need before you can do some dumb things like >250% prod/recycle looping to efficiently spit out legendaries

[kindred crater]

not just setup, but "maintenance" as well (tinkering it over time). but with good bps like you said, that should hopefully shrink

[daring siren]

Don't forget param BPs and circuit recipes which help a lot

I do hope param BPs get "stack size of"

[iron root]

we'll at least get something close with selector combinators

i think

[daring siren]

We do, but Kovarex was all about "compile time optimization", which I like

And you can still manually put the stack size in the computation, but it'll be nicer when we have a computational param for it, like the "ingredient of"

[iron root]

and that 'could manually put thing in but it'd be nice to automate' is exactly where parameterized bps began

[daring siren]

Indeed. So I don't doubt they will eventually add this feature as well

[unborn flax]

Quality recycling intermediates for science packs is interesting. Suppose you have a normal LDS and this is part of a science recipe (e.g. in 1.1). Then from 65% infinite prod research bonus for LDS, you're better off recycling-looping normal LDS until you get >= uncommon. This assumes you can match the qualities to those of the other science pack ingredients. But the threshold is low enough to be maybe in scope for a real game.

The point is that I think partial quality-recycling of intermediates is worth thinking about for science (I'm not proposing a specific build for )

[iron root]

I think that's going to require having everything in said science pack with that much prod

or the other non-prod research materials need to get their quality from elsewhere

[neat shard]

I mean, the optimal solution is "prod mods in any recipe that can make use of them, qual mods in any recipe that can't"

I personally wouldn't call that "complex" or "dynamic" but you do you I guess

[north yew]

I think you're doing your math wrong. I just did the math on this assuming 100% prod from research and 50% from foundry, and recycling the normal quality science causes a loss in the total amount of science put out

well, recycling the normal quality LDS

[final nimbus]

Is recycle-looping intermediates such a bad idea?
If you don't, you end up needing to recycle every item you want quality of. And that's going to be a lot more recycling setups in total.

[north yew]

it's ~21% loss per recycle loop

[neat shard]

Recycling a module or building is no more expensive than recycling each of its ingredients once

[north yew]

well it goes back to the question of whether or not you're only accepting best-possible-quality. if you value not-best-possible-quality at all, then full prod your intermediates absolutely. otherwise, it's more debatable

[neat shard]

And recycling the intermediates means you're missing out on prod

[north yew]

it's arguably slightly more efficient because of the 2P2Q thing to recycle the ingredients that the final product is made out of

[tall sandal]

its not but that's okay

[north yew]

if you're only accepting best possible quality

still somewhat inadvisable, though, for reasons

[tall sandal]

for one, prod modules in everything is not the optimal solution in vanilla for all stages of the game

[final nimbus]

Productivity research would make certain intermediates more disposable.

[north yew]

generally though only accepting best-possible-quality is gonna be a very bad strat outside of a postgame ultrabase imho

[tall sandal]

for two, prod modules everywhere you can is not optimal if you are interested in getting higher qualities for cheap, and its certianly not optimal for the spidertron achivement

[north yew]

spidertrons are a bad example

there's no benefit to quality intermediates in those

because

the only way to upgrade is to recycle spiders

[tall sandal]

but you can get an epic fish

and so you want epic ingredients to go with that epic fish

[neat shard]

How?

[tall sandal]

by recycling an epic spidertron

or recycling a common spidertron

[final nimbus]

In theory you get the other epic ingredients at the same rate

In practice, RNG isn't nice.

[uncut plank]

Assuming there is no other recipe/source of fish. Another possibility is fish are treated like a fluid ingredient and don't affect quality at all.

[north yew]

high-quality ingredients would let you capitalize on a statistically unlikely fish roll, that's true, but on average it's not going to affect the outcome

[kindred crater]

looking at it another way, you just need to get a legendary fish. once you have that, you can recycle loop whatever else without being constrained as much

[north yew]

for every person that gets a legendary spider faster because they made quality ingredients, there will be a person who wasted the time making those ingredients because the fish was the bottleneck

[final nimbus]

You don't know which person you're gonna be though, and need to be prepared for any outcome.
Averages appear on paper, annoying bottlenecks appear in practice.

[north yew]

theoretically it's possible to liquidate your assets, bet it all on a roulette table, and come out much richer

is it a good idea?

[neat shard]

If you wind up getting the legendary fish first, maybe then you have something to gain by recycle looping the remaining intermediates. But there's no point in recycle looping the intermediates ahead of time

You lose nothing by waiting until you know you'll need them

[north yew]

additionally, you're gonna want to make quality spiders anyway. quality spiders are really good. if your base can support spidertron recycling, why not just let it do its thing beyond the point where you get the cheevo

[final nimbus]

Well imagine that you go ahead and build a spidertron recycler.
They have 8 ingredients to handle, each of which is coming out of the recyclers at random and with 5 possible levels of quality.
Any of those could start backing up, getting all of them produced at the same rate of consumption is not guaranteed.
To build a recycle-loop that can't deadlock you need an overflow valve that gets rid of anything you have too much of. Not necessarily difficult, just buffer each and send back to recycling if the buffer fills.
But now you have to deal with everything that that recycles into, unless you're okay with voiding it all.

[north yew]

on average what's coming out of the recycler is 25% of the ingredients of a spidertron. there will be some peaks and valleys for sure, but it'll average out to equal amounts for each ingredient

[final nimbus]

On paper.

[north yew]

yeah, some buffer space will pave over the issue

you're going to need to feed ingredients in to the setup constantly anyway, so you just make sure that recycler output gets prioritized over the new ingredients

[final nimbus]

Quality stuff is an issue tho

[north yew]

it's definitely the case that spiders take a cartoonish number of ingredients, both in count and quantity, so avoiding backup is gonna take more buffer space than a simpler recipe

[final nimbus]

Anything that can go wrong, will go wrong.

[north yew]

at the end of the day though it's just gonna take buffer space, prioritizing recycler output to new ingredients, and letting the law of large numbers do its thing

[final nimbus]

is the one likely to get you, with how many of them you need and the small stacks.

[kindred crater]

in janky quality with recycling looping expensive items and each grade had one assembler, I simply had inserter -> box -> inserter assembler so things kept flowing, even if it wasn't balanced enough to fit all inside of the assembler

[north yew]

you'd probably want to do spider recycling with bots tbh. then just have the inserter taking new ingredient input only swing when amount in logistic network = 0

[kindred crater]

and to be clear, the first inserter took all intermediates of that given quality

[final nimbus]

Point is

The Law of Large Numbers is descriptive, not prescriptive.
And anything that can go wrong, will go wrong.

I don't think it'll be an awful problem in the end though

[north yew]

right but even one is enough buffer space to make the probability of a bad enough result to clog the system approach zero

so if you had like ten of them

there's not a lot of room for it to go wrong

[final nimbus]

Might want a few for but sure, it's easily solvable.

I'm still gonna want an overflow valve because trusting RNG to be nice to me is a lesson I have learned.

[north yew]

I think you could actually do it really simply by just having recyclers output to , new inputs go in to , then have a little 5x5 chunk of

[final nimbus]

5x8

[north yew]

yeah each has enough space to store the ingredients for 2x spiders

with room to spare

so theoretically you'd probably be fine with like five

but they're cheap so why not do like 20

[final nimbus]

I said 5x8 because that's how many different ingredients there will be and bots like to give each item its own little box.

[north yew]

they try to but it's not enforced unless you filter

they can be trusted to fill up stacks before spreading them out

[half raptor]

time to get some sleep before FFF day

[red kelp]

why not sleep

+150% more sleep for your sleep

[rare thistle]

Sleep quality:
Tossing
Turning
Crashed
Snoozing
REM

[chrome mauve]

Now here me out. There is supposed to be another way to get quality things and what "if" that is technology that adds quality same way as we get productivity tech?

[daring siren]

Devs said clearly that the only way to get quality items is by putting quality modules in buildings, and those buildings producing something

[rotund egret]

it's already a simple system with crazy consequences

making it more might make it go too far down the rabbithole and be too complicated

[chrome mauve]

f, I was pretty sure that there was one sentence where it was said that modules wasn't only one way to do it. Now that means my memory is failing me a lot recently

[daring siren]

Mods could have buildings with built-in quality increases, which can allow for certain recipes to always result in better quality. This could help in things like Galdoc's Manufacturing, where there are alternate recipes for different characteristics of intermediates. But it's not something we'll have in vanilla.

[unborn flax]

Here's my working:

q = 0.062 * 4 

# [normal, uncommon, rare, epic, legendary] after recycling 1 normal LDS
v = [0.25*(1-q), 0.25*0.9*q, 0.25*0.09*q, 0.25*0.009*q, 0.25*0.001*q]

# [normal, uncommon, rare, epic, legendary] after Foundry with 4xprod3t5 and +65% prod infinite research
v = [(1 + 0.25*4 + 0.5 + 0.65) * a for a in v]

# dot-product resulting vector with science bonuses
print(sum([a * b for a, b in zip(v, [1, 2, 3, 4, 6])]))```

giving 1.0046736, which is > 1, i.e. more than we started with. Of course in practice we'd do this iteratively by looping any resulting normals, but I've just done 1 iteration to keep the maths simple. The point is that while our total items have dropped by 21% - or 52% if you keep looping the resulting normal LDSs - they now give us more effective research.

Could you please show me how you reached your conclusions?

On the point of best-possible-quality: I argue that for science this isn't a good idea. In my head the science pack quality tier I'm aiming for is uncommon, not anything higher. This is because you have the same chance of a quality bump starting from normal as you do from (say) epic. But while normal->uncommon gives a relative research gain of 100%, epic->legendary only gives a relative gain of 50% (from the ratio 400:600). So if you're making quality science from scratch (as opposed to using imported spare parts of quality from elsewhere in the factory), I don't see any point theorycrafting to obtain science packs higher than uncommon until you have this problem completely solved and profitable - which it definitely isn't yet.

[opal haven]

Multiplying by 0.9 in quality formula (while q is ~0.2) looks incorrect. Does v adds up to 1 after you calculated quality probabilities?

In the example with 0.9 quality probability was 0.1

[unborn flax]

They do add up to 1 (or rather, 0.25). The numbers in the original FFF were a bit unclear and were later clarified by devs. The way I work it out does match the 'for %248 quality strength' numbers in https://wiki.factorio.com/Quality

[0.752, 0.2232, 0.02232, 0.0022319999999999996, 0.000248]```

[opal haven]

Hm, looks like it's really gonna work like that - quite cool

[unborn flax]

Thanks for checking; scrutiny is always a good thing.

Also: I've run the numbers for putting quality modules in final-stage crafting machines for science packs. I believe that in this case, quality modules only become useful in the extreme / theoretical late game (assuming there is no controversial revelation that science packs can no longer be prodded). So this basically agrees with @north yew 's assertion and I can further give specific numbers. Note that assuming all inputs are normal is the best case.

• For a 4-module machine with normal inputs, one is useful at 190% bonus productivity (when the total prod is close to maxed out at 300%). The best this ever gets is 27.6% extra research, when you have so much bonus prod that there's no advantage to use prod modules.
• For a 5-module machine with normal inputs, one is useful at 165% bonus productivity (50% of which could be the EMP, assuming they take prod modules). The best this ever gets is 34.5% extra research.

[half raptor]

Prod bonus research is going to make calculations fun, that's why I'll just be going for builds that aren't too sensitive about exact ratios and just rely on some simple rule of thumb logic to sort things and just run around chasing my limiting reagents

[rotund egret]

extra productivity can also mean "this machine consumes less input for the same benefit"

so with the same setup, increasing productivity by 10% will just mean it consumes 10% less

kinda like mining productivity now

it's not like we're redesigning mining setups with each research

[unborn flax]

That's always a good plan.

I suspect that if bonus research thresholds are too high they may never result in a quality build being useful, i.e. by the time someone has accumulated a large amount of endgame prod research by running a factory for x years, UPS could be more of an issue than resource efficiency.

[half raptor]

with 300% prod ceiling, it's not technically an infinite tech, but I knonw a lot of megabasers will assume max research in that when designing builds

[north yew]

I assumed was also going in the foundry. your math is correct. some notable complications to the strategy, though:

• it's only really accessible to very late game builds with the best possible modules. even just going down to makes this strategy completely unfeasible
• it's only worthwhile if every single ingredient for the science pack has an infinite prod research. for example, in we haven't been given any reason to believe roboframes have an infinite research. you'd need to send the roboframes through the same loop to match the quality ratio of the and , but roboframes would be losing so much material that I'm pretty sure it'd outweigh the gains you get from recycling the other two.

[quaint flower]

i wonder if they have this DLC in a recycler loop....

[north yew]

reading your messages it seems like you're under the impression that every recipe is getting its own individual prod research, which is definitely not the case

[unborn flax]

I'm certainly not, and I'm not proposing a specific strategy (in fact I said explicitly "I'm not proposing a specific build for yellow science").

[north yew]

I do think that for non-proddable science recipes ( , , ) using for the ingredients eventually becomes a good idea in the lategame

(early and mid game it's probably more hassle than it's worth)

[neat shard]

They've definitely been investing a lot of resources to make sure it's high quality

[north yew]

it is definitely correct that in the case where a science pack's ingredients all have infinite prod research and you have high levels of that research and you're willing to invest a large quantity of legendary T3 modules in to squeezing out a few percent of efficiency, recycle looping the normal ingredients and only using uncommon or better for the science pack is best

[quaint flower]

indeed. it might take a while to become , but it will be worth it in the wait.

[neat shard]

I mean, with +300% prod from infinite research, you can just recycle loop infinitely to make everything max quality for no cost at all, right?

Other than the inherent UPS cost of spamming recyclers I guess

[unborn flax]

My calculations do show that writing off quality intermediates (as you have done more than once) may not be universally correct, so I'd encourage people working on this kind of thing to show numbers and working as much as they can. It's possible that there's a good strategy yet to be found.

[neat shard]

Share those calculations?

[sick dock]

This actually a great point. If an entity needs a very exotic item you can't easily scale with quality, if at all, then intermediate quality is pointless (at least for that craft in particular)

[daring siren]

Kinda... not everything is getting that prod increase, but it's enough so you can get most ingredients in high quality

[neat shard]

Well, yeah, for those things that get infinite prod research

[daring siren]

Steel, LDS, processing units, (not) RCU

[north yew]

understand that quality intermediates has been my white whale for like four or five months now, and I've had the argument with approaching a dozen people at this point. none have yet provided a convincing argument for quality intermediates, so that's why I tend to get a little dismissive of it these days.

[daring siren]

So that gives us steel, copper, plastic, red circuits, green circuits, and in double recycling, also iron plates

[unborn flax]

i mean e.g. #1215078107334057984 message - if you play with an open hand than everyone can check your working and either agree or disagree with numerical evidence. whereas statements like "i've checked this out and it doesn't work, trust me" just aren't useful in a roomful of smart people with slightly different ideas of what they want to achieve

[sick dock]

we're all assuming this is something relatively achievable...

[daring siren]

The main thing is the gives much more % than , so in total it's a better use of materials.

[sick dock]

In vanilla, bullet dmg lvl 30 would take ~1,000 days with 10k SPM

But i get we're in theory-crafting mode

[daring siren]

There's no reason to go lv30 on because foundry gives 50% already, 25 is already enough for max

[sick dock]

we have assemblers as well

and lvl 25 is still 30 days with 10k spm

[daring siren]

With 150% prod from base + modules, you only need prod 15 to max out (fixed)

[sick dock]

ah, yes you can use prod mod in the machine, duh.

It's still going to be a very late-game feature, which is good.

[daring siren]

With 175% prod from base + modules, you only need prod 13 to max out

[sick dock]

Inf quality looping is abit...counterintuitive? idk the right word

[daring siren]

Regardless, even at prod 5, you gain a ton of value from it

[sick dock]

For sure

[north yew]

this is a fair point, but the main reason I stay away from numbers at this point is that it tends to be a distraction. what most people are missing when it comes to quality intermediates is the way products move through a supply chain. it's easy to look at it in terms of a single step and just assume it works out, but it's the relationships between steps that undermines the premise. the big reason a lot of people tend to overlook or ignore the problem is because it's really hard to get the numbers for. I mathcrafted it out one single time for a simple supply chain with all assumptions made to keep the example as simple as possible, and even that took a massive spreadsheet

[sick dock]

I think similar to oil, people will just ignore maths and do what they want with quality, and go by feel of how much they're noticing their resources draining

[daring siren]

In ColWill's base they had a ton of issues of unbalanced qualities

Oil is very very easy though, and pretty much solved in vanilla

[sick dock]

Mmmm, with productivity cracking it's not that easy, at least for me

I still do it, but my point is many don't

Quality is more complicated in comparison, surely

[daring siren]

Oil has a single path - down to and into sciences

Quality is multi-result outputs at every stage

[sick dock]

make , down to , make & , down to

[daring siren]

Yea basically

[sick dock]

i'm not attesting to me personally. as i said, i'm fine doing the maths. i'm attesting to the newer player 😛

[neat shard]

Soul-Burn's point is that no "backflow" occurs

[sick dock]

and yeah, i think quality maths is going to be way more complicated

[neat shard]

No step of this process outputs a previous stage as a byproduct

[sick dock]

yeah, that's true

although, if they made a -> recipe, it'd have to be real good in order to use

but idk what that would add to the game

other then the potential for oil inf, but then oil is inf anyway...

[neat shard]

To be fair, an infinite oil loop would be rate-unlimited infinite oil, which you can't really get just from pumpjacks

[daring siren]

No backflow, and no real choices of what to crack, compared to recycling which is going to be messy

[neat shard]

If there's an oil-positive loop, you can always get more oil by building more cracking and uncracking, rather than having to track down more oil patches

[sick dock]

Yeah, it's a lil different

[sick dock]

I'm so excited for ...it's going to feel so different

[daring siren]

In K2 there's a recipe to extract from . With enough prod, you get a positive loop which is kinda funny

Heeba did convince me that I'm not going to do quality where we can use prod... but I'm still going to quality in my mall, and science ingredients that can't be prodded

[sick dock]

No reason not to use on final products, unless you'd prefer or

[north yew]

mall assemblers are idle 90% of the time anyway

[sick dock]

or just not to make as much mess with varying

[north yew]

so very little benefit from speed or eff

[daring siren]

Well, except for logistics, but I think it's worth it

[daring siren]

Not if we're gonna recipe circuit them 😉

[neat shard]

Also, in case anyone else is wondering about recycling... it's still not quiiiite positive even with Q5T3 prod mods, if my math serves.

[daring siren]

It's not, and it still costs

[neat shard]

1 used fuel cell turns into 0.6 * 1.5 = 0.9 U235, and disregarding the U238 requirement, 0.9 U235 turns into 0.947 fuel cells, which turns into 0.947 used fuel cells

You spend 5.3% of the uranium with each cycle

[sick dock]

can you a ? i've never tried

[neat shard]

Yes you can!

Centrifuges only have two module slots though.

[unborn flax]

That is what I mean about having different perspectives; it sounds a bit like theory vs practice. I like the numbers because they're an upper bound on how good things could ever get and they occasionally throw out surprises (for instance, how outrageously cheap it is to quality-cycle to legendary with the EMP, if it accepts prod modules). Considerations like "how do you actually turn this into a sensible factory?" are very important but don't affect the theory.

[sick dock]

I believe Heeba is saying they've done the maths and it's just overloading what could be explained in a few sentences

[north yew]

re: the numbers on that massive spreadsheet -

• , all normal ore goes to science, uncommon or better goes to quality section
• rare is highest quality unlocked
• and are being used
• all modules in assemblers/furnaces are t1 normal quality
• target product is t1 modules
• prod used on rare ingredients b/c there's no possibility for quality upgrade
• quality is used in final step in either case
• result is measured in T1 modules per 10,000 total consumed ( consumption not factored in to calculations)
  using all quality modules in the intermediate chain resulted in an output of 163 uncommon modules and 33 rare modules
  using all prod modules in the intermediate chain resulted in an output of 214 uncommon modules and 28 rare modules

this is a setup that is designed to tip the scales to the benefit of quality intermediates, and even then it doesn't look great. yes, you get a few more rare modules, but that's at a cost of almost a quarter of the uncommon modules. if you aren't starting from uncommon, the numbers are gonna look even worse. if you're in a longer chain with more opportunities for the ratio problem to bottleneck production, the numbers are gonna look even worse.

from a gameplay strategy standpoint, I think this example makes it pretty clear. would you rather have 51 uncommon modules or 5 rares?

[unborn flax]

I know you mean this rhetorically but I think you'd need to quantify that to be able to compare one hypothetical setup to another - otherwise who knows if (50 uncommon + 5 rare) is better than (40 uncommon + 7 rare)?

[daring siren]

The first couple of rares are very important, so the 51/5 numbers are not a good one. 163+33 and 214+28 shows it better - enough rares to get things started, and many more uncommons, so that's good.

[unborn flax]

It's also interesting that this is a mid-game problem you're trying to solve - way before things like prod3/qual3 modules and (probably) infinite research come into play.

[north yew]

tbh I'm at a point where I kinda don't like this example just because of how much work it does to tip things in favor of quality intermediates. it gives the wrong impression. a longer chain starting from normal quality with all five qualities in the mix would demonstrate the ratio problem much more clearly, but it would also be insanely hard to work out. even just keeping things down to two possible qualities required eight different calculations per ingredient

[heavy ember]

What I'm getting from the write up is that quality at the first and last step in early game seems to be the best idea?

[north yew]

the big part of what tips the scales is that we're starting one below best-possible-quality, and 10% of our input is already there. you're only seeing the opportunity cost of not using prod to a far smaller degree than you would in a less narrow setup.

[north yew]

last step is a freebie, you always do it there for sure. quality modules in drills is a bit debatable. if you have biters off, go for it, but if you don't, it means the power consumption and pollution of your drills is ~6x as high vs using . notable, though, is that if you're using a strategy of sending normal ore to science, sending uncommon or better to a specific quality section, and using prod modules on all intermediates in the quality section, the ratio problem does not arise. basically, quality modules in drills potentially a viable approach. costly, but might be worth it.

[neat shard]

Also because even though you can put prod mods in drills... there's really not much reason to.

The only thing prod does for a drill is make it deplete the patch slightly slower.

[kindred crater]

which is a non-issue if you're later game and have access to sufficient ore patches

[daring siren]

Imagine super late game, recycling ores for quality at the patch

[neat shard]

But why though

[unborn flax]

That would make for blissfully simple legendary factories

[north yew]

recycle looping ore for legendary is the most expensive way to ensure legendary only outputs, but it's also logistically the simplest way to do it

[kindred crater]

I'm veering from quality for a moment, I know planets are for practical purposes infinite, but for longer term games I'm kinda worried about "burning out" a planet in the sense you have to go pretty far away to find ore. when SA releases, a lot of thought will be put into minimizing this by choosing which planets (vulcanus, fulgora, etc) to 'sacrifice' in that sense lol

[north yew]

by the time you've got 10-15 levels of mining prod and legendary big drills reducing consumption to 1/6th, it might reach a point where you kinda don't care about resource efficiency and just want to build things faster and deal with fewer problems

[kindred crater]

where I don't mind using lower efficiency methods to get what I want

[daring siren]

It might be better for UPS to recycle at the patch

[heavy ember]

I wonder, with all the prod/buffs/new stuff, could mining prod actually create more ore than it needs for the level?

[unborn flax]

I can see people making DI builds at the patch, given the huge range of big mining drills

[daring siren]

Combined with lab prod? Possibly

[neat shard]

It's possible infinite mining prod research will be pushed back and made more expensive

[unborn flax]

Fulgora could be different though. Actually Fulgora feels like a pretty trolly planet if you want to use manufacture quality things. Because you have two ratio problems fighting each other: balancing scrap outputs, and balancing quality.

[heavy ember]

Because in dsp there's a flat number for which is infinite if you only research vein utilization

[north yew]

considering just : even just the innate bonuses of the foundry and big drills reduce your ore consumption to a tiny fraction of what it would be without them. for example, I estimate that just the innate bonuses get you for 1/5th of the cost in . throw in a few levels of mining prod in there, and a 20m iron vein (not something you don't have to go far to find) gets you the same amount as a 120m vein with no foundry/big drill/mining prod

[north yew]

there are some who have proposed putting quality modules in the scrap recyclers. I consider them mad

[neat shard]

Maybe if you're doing the "sushi megabase" approach? XD

[kindred crater]

yeah thats a good point. adding to that, before I start mass upgrading stuff via quality, upgrading productivity modules throughout the base will be massively helpful in reducing consumption, and will likely be the first thing I mass quality manufacture after the legendary quality modules themselves

[north yew]

quality makes for a considerable increase in resource consumption once you start recycle looping, but the rest of SA allows for a considerable increase in resource efficiency.

[unborn flax]

One technique that may be useful is having multiple production lines with different module setups, so you don't have to void items. I do this in my SE game to balance core mining outputs and have never had to turn anything into landfill. I'm planning to try the same thing in Fulgora to balance scrap outputs but I'm not sure how well it would extend to quality balancing, if at all (the linear algebra wouldn't be pleasant).

[north yew]

yeah, nailing down resource efficiency on without quality in the mix is tough enough. throwing quality in would be an exponential explosion in complexity

[heavy ember]

SE endgame stuff spoilered:
Would it be that different from ||folding logic||? Seems like a similar use case

[unborn flax]

sorry i haven't got that far... i don't want to de-spoiltag 😕

[iron root]

nah that doesnt sound like the SE thingy

you can use linear algebra for the SE ||arcospheres but they have a conservation law, this doesnt and thats a pretty big fundamental change I think||

but if you ask me, running a line with fewer productivity modules to use more material is equivalent to just voiding

the difference is with SE core mining, you dont have the option to automate voiding until lategame
in SA you do

[unborn flax]

i agree entirely - for some reason I find it more elegant than landfill, even if the landfill solution is easier to implement

[heavy ember]

I wouldn't say delivery cannon pointed at a warehouse that destroys it whenever it gets full is exactly late game

[old lake]

Like belts, lamps, walls?

Eh tough walls could be nice

[half raptor]

replaceable walls are better than tough walls

[old lake]

But tough belts and lamps are 99% useless

[half raptor]

also repairable

[old lake]

True

So walls for stone, belts for iron and gears, and lamps for circuits?

[final nimbus]

I've proposed using in everything that doesn't allow
The other modules are for beacons.

[old lake]

I like that

[half raptor]

I propose using in everything and for lines handling ingredientts

and you just route the unlucky to making belts and lamps

[old lake]

Won’t get much prod that way

[half raptor]

buut you will get qual

[old lake]

That is true

There’s also the strategy of “ooh quality modules! Let me grab a few of those and… *toss* into the assembler mall you go!” And no other modules

Not a GOOD strategy but it gets you faster machines sometimes

[north yew]

(spoiler alert: that's the best strategy actually)

[half raptor]

not a bad strategy either, I know that putting prod everywhere and qual at the end step (with recycle loop) will also work eventually

[north yew]

it has the best ratio of effort to return

[half raptor]

I plan to go qual the whole way, but more for the fun challenge.... productivity 99% of the way and quality loop for the last 1% is probably best though

[old lake]

Agreed

[half raptor]

easier to "balance" too

[old lake]

I’m probably gonna go full qual but that’s just cause I like the organized chaos of multiple things being thrown onto the same belt and then getting sorted later

[half raptor]

I have sevral fun ideas, but Wube won't give me a beta key for some stupid reason

[old lake]

You should see me when I’m sorting the ores from SE core processing. It’s one enormously fast belt running along with like 5 ores being loadered onto it, loadered into a warehouse, and sorted out. Plus a combinator to ring an alarm when it’s gonna break in less than an hour

[half raptor]

I've already come up with a solution to imbalance of qualitiy materials for my theoretical base. In essence it goes like this

Every production city block (i'm so original guys) has 2 outputs, 1 for "expected results" which is the same quality as the ingredients where 75% of products go. Then I have an Upgrade pickup station and because of how the train interrupts work, they function as generic pickups. When a train reads its contents it will deliver to a drop station, train limits controlled via circuits and so will go to a drop station. With these small drop offs from our upgrade trains, it'll eventually buildl up enough product to fill a full train in the rare production block and will filter upward in production.

If there is "Destination Full AND Duration 10-30 seconds (haven't decided yet) then go to a Recycle station. The recycle stations are simple and repeatable so it can handle many different kinds of things being recycled out of the base. Since my request stations are going to have a buffer of 6 train loads (6 chests = 7.2 trains) if I get that much buffer then I really don't need this junk, do I? This will recycle back a step and maybe upgrade the items. It's fine for items to void and maybe upgrade over time which would then be picked up by another generic train to deliver.

If the stop is for items and faces the Destination Full and 10 seconds, I have a plan for a secondary train stop to be at each block that acts as an overflow valve. You can set your assembler to use higher grade material in combination with lower grade and it'll just craft it as if the material was lower grade. This handles the eventual 300 hour problem of too much building up

I can set priorities on those overflow stations so it'll try to go to epic then rare, then uncommon and lastly normal. If for some ungodly reason I wind up with full overflow stations even after all of that.... then I've played too much SA

I might even at that rare moment just send it to a depot because if I'm that full on something.... obviously my base needs more of another ingredient and when I increase production of that, it'll correct itself. Because I'm either always delivering goods or recycling goods, my output buffer should never completely fill and it'll be my input buffers that will be the test of production

[final nimbus]

All those new train functions will make Quality not as much of a pain to organize.
Circuit stuff too.

[half raptor]

I like to think that SA features qualify each other. Why do we have redesigned rail segments? Because they wanted to go elevated

Why do we have a wiring rework? Because they're going to change combinators

Why is uranium now so essential when it wasn't previously? Because each planet has their special resource and uranium just made sense for nauvis

[final nimbus]

Why have elevated rails? Because they're really cool Unique planet terrains that require bridges.

[half raptor]

it's like we have a 100 piece jigsaw puzzle and each week Wube gives us a new piece to play with

it's so satisfying when a new piece clicks with another

That's why I got overexcited this morning when I saw more train stuff.... eh, it's okay but it's not super game-changing

It was more of a pollish than something new

[north yew]

testament to how high Wube sets the bar for themselves

I know many games where that kind of specific QoL update would be a godsend to be celebrated for years

but for us it's just like "yeah neat 👍 "

[half raptor]

FFXIV announced a couple years back that you can change menu windows to the classic JRPG format to look like SNES final fantasy and the fanbase loved it dearly. Factorio casually drops "we're making editing your train stations easier and simpler to avoid massive problems and headaches for not doing it correctly and we're like "ok"

to steer back to the topic of the thread, I hope for FFFs that will give another logistical piece of the puzzle for making quality stuff, train interrupts definitely helps, as does changes to circuitry

[daring siren]

Yes. I do hope lamps get a bigger light radius though

[old lake]

Is JG the only dev who watches this thread?

[kindred crater]

who talks at least

[neat shard]

Honestly, I think it's just a matter of us already being maximally hyped for 2.0 and SA.

There's an upper limit to how much hype the human brain can process and I think this community has just about reached it.

[kindred crater]

yeah this place is the epicenter for SA hype 😎

[daring siren]

@obsidian crescent "nutshell" - writes 4 page dissertation 😉

[north yew]

Kudos from me for not once bringing up the concept of quality modules in intermediate production though

[tall sandal]

quality modules in miners

[iron root]

Huh
Quality modding miners will reduce their speed which sucks
But then you can make quality drills which more than compensates

[daring siren]

it compensates for a different thing though

[iron root]

Oh right drills don’t run faster with quality

Speed penalties on miners is rough but I still feel like the ability to make everything is worthwhile still

[daring siren]

Interestingly, removal of means we're not sinking , so putting quality in them isn't "free"

For , we're cycling through a lot of which can be qualitied

[iron root]

Huh

Following that line of thought the quality mall fed with quality ore doesn’t need to make everything, cause some of those factory machines are used in science

[daring siren]

My and makers are the same that make . Same goes for with , walls, and

[iron root]

Yeah, I forgot about stuff like that when planning my quality setups but that’s a good shout

Mom can we have
We have at home
at home:

[north yew]

I think is gonna be the big one you want to quality IMHO. It has the lowest idle drain and power consumption of all inserters, which are both gonna be relevant on space platforms. There's definitely gonna be some cases where isn't enough but or is way too much

[iron root]

I’m not a fan of micromanaging my inserter types, and they don’t use much power compared to machines usually

But who knows with in everything on a platform maybe they will

And quality might be more useful

[north yew]

Most of the time when I check the power overview I'm surprised by how high up inserters are in the list. They use a not-insubstantial amount of power.

Platforms are a place where squeezing out a little bit more efficiency pays off

[iron root]

I suppose after prod/speed stacking to hell and back for 200 hrs I’m just not used to the ratios

[north yew]

Yeah it's easy to forget our little yellow friend when all your machines are running at 400% speed

[dense kraken]

I just noticed something, heat pipes won't have quality, right?

while reactors and boilers will

that is bound to be annoying interesting

[sacred totem]

they will

everything has quality

but uh, well, not everything has a useful quality

[north yew]

I may just need to think about this more but I'm pretty sure with the way heat works, as long as all of the turbines, heat exchangers, and reactors get an equivalent quality upgrade, no changes need to be made to the design

assuming heat pipes don't get any sort of upgrade

fluid throughput becomes the issue, actually

[neat shard]

I wonder why pipes don't benefit from quality...

Like, belts I get. Belts are optimized and aggregated, having mixed quality belts would be bad for performance, and also would mean belts that go so fast that it breaks stuff.

But pipes already have to be calculated on an individual basis, so what's the harm in letting quality increase pipe capacity?

[north yew]

it's also just that they suffer from the same problem of it only being worth it to use a quality upgrade when you do the entire line in that same quality

if your chain of legendary belts/pipes has a single normal pipe in it, it only goes as fast as the normal pipe

[sacred totem]

no, you will run into heat throughput issues

obviously water/steam as well, but you cant discout heat

[north yew]

yeah you right

[median verge]

Not necessarily, I don't think. It will bump up average pipe throughput iirc

[rotund egret]

the quality mall also doesn't have to make a lot of stuff that you'd make in the common mall, like belts, pipes, chests, trains et cetera

[iron root]

my thought is the common mall wont even exist once you get quality ore setup

though you can still make things that dont benefit much from quality as commons ie those belts/pipes/chests

[rotund egret]

it depends, you might want to deliver quality ingredients to the mall, so just maintaining the old mall could be a viable solution

[daring siren]

You would want to circuit stuff up when you work you quality mall. Before that, it's not worth the effort

Also, you are likely to have several malls, one per planet

[rotund egret]

I anticipate some sourness on my part when the influx of

"EASY CIRCUIT LEGENDARY MALL HUB BLUEPRINT" where people just copy-paste it

well I guess it's not that much different from what it is now

[charred gyro]

It's going to be harder than ever to avoid the temptation to just grab a blueprint for malls

Maybe I ought to prioritize making one of those fancy new automated malls so I can avoid that temptation.

[weak vault]

I feel like they very intentionally added recipe productivity research. This allows to offset the need for productivity modules in machines and choose quality instead. Plus machines are faster so you offset the need for speed. That's going to be incredibly fun to play with because it really matters what module you put where and it's not an easy choice!

[rotund egret]

and it throws off perfect ratios

[jaunty citrus]

There really just isn't a chance that the increased logistical complexity and inability to use speed/productivity modules makes using quality production throughout a factory worth it.

Prod+speed is just too powerful

You can get really cheap, but specific quality production using medium levels of productivity research and high quality , it just won't scale anywhere near as much as +

[daring siren]

Good.

People keep asking for rate calculator in vanilla, and I disagree wholeheartedly because it only works with perfect ratios

[rotund egret]

exactly

people forget that free productivity can just read "this machine consumes 10% fewer resources"

so even if you keep the same ratios, the end result is that you need way fewer early inputs

so your iron patches will be able to service more stuff

if I manage to do another playthrough until SA my goal will be to make minimalist purple and yellow science builds just for a few researches. I don't make any blue circuits until my first full-scale yellow science build.

just make one assembler for each and let it run dammit

speed-moduled too

[daring siren]

Nothing is good as actually running the simulation. If there was a vanilla way to easily copy a build to a sandbox, with an easy way to set infinity belts, run it in for a bit, and give out statistics of that. Then that is something I would approve.

The blueprint GUI needs work anyway, so this is a good place to work on.

Tips & Tricks already have live simulations, so it's not out of the picture

A surface where you can design blueprints, with creative tools that don't get pasted in the world, and live simulation and statistics.

Works well with the idea of production per surface.

Also, this whole conversation isn't relevant quality, so I'll move it to another place.

This discussion moved to #1177597309572366396

[sacred totem]

no, 1/(1 + .1) != 1 - .1

prod also increases the machien output rate

[quaint flower]

speed does that too i think

[sacred totem]

really? no way

/s

[quaint flower]

yeah ill need to research it more.

[rotund egret]

okay, 10% productivity can just read "this machine consumes 9.(09)% fewer resources"

[north yew]

what if you deconstruct the machine before it fills up the purple bar

[rotund egret]

sucks to be you then

[daring siren]

But it shouldn't, because of how effects stack

[rotund egret]

my whole point was that you don't need to worry too much about ratios when you get a new prod research, since you can treat it as "prod saves resources upstream"

[daring siren]

Worry no, but every once in a while you can optimize your builds for size

[rotund egret]

yeah

this started from the convo about people being too fixated on blueprints with specific ratios

[quaint flower]

optimize, shmotimize, build it big and if its too big make it smaller

[rotund egret]

"too big" the astronaut says

[sacred totem]

L + ratio + biters broke your walls + no oil

[daring siren]

But

But space platform

[sacred totem]

total rebuild every time you unlock a new tier of prod

no exceptions

not even the rails will survive

[rotund egret]

legendary rails or bust

[daring siren]

Do yourself a favor, and don't play FactoryIdle

[quaint flower]

[quaint flower]

i used to play that game lol.
i did not get too far but it was like one of the first factory games i played

[daring siren]

It's horrible. I got relatively far and then realized I'm not enjoying my time with it. Thing is, it gets so slow, that you always have to restructure everything when you get upgrades, and it's always different builds.

[quaint flower]

i unlocked a second area but that's how far i got

i dedicated that are to resertch

[half raptor]

One fun thing about idle/tycoon game is the exponential currency model where you're spending 6 trillion coins to upgrade a lemonade stand.

[daring siren]

In this case, it just got very slow and not viable without mtx

[north yew]

Generally speaking resource usage is going to go up over time, so while a productivity research might temporarily result in idle machines, that's not gonna last for very long

[final nimbus]

If they're adding productivity techs for a bunch of things
Why not do it like mining prod
Where the first few levels are done with earlier science packs, but are limited
Something like a +10% from green science or +10% from blue
Just a thought I had

[kindred crater]

I'd assume so, to tease the mechanic and its benefits

[north yew]

Part of the justification for it is to fill gaps in the tech tree where it'll be a bit before you can bring new science on line

[iron root]

id argue it works with more
i just use it to see if theres a major imbalance when slapping down rows of machines for a sequential set of recipes

[daring siren]

Please use the #1177597309572366396 thread for this, as this place is about quality

[daring siren]

It only gives correct/useful values if your goal is a perfect ratio, something that gets thrown out of the window for more advanced recipes, and then even more when adding modules.

Say I have a chain of 3 machines with modules, in a way they there isn't a nice perfect ratio between them e.g. 9:8:7 or 110:111:100. If I do 1:1:1, rate calculator will show that I'm missing buildings in the middle, and that my total output is higher than it really is. This is not useful data. I know that is the case and I want to see the total expect value across the 3 machines.

More complex tools like Factory Planner have a matrix solver in them, which gives better results, but also has its own issues.

[iron root]

maybe im just thinking about it differently then i dunno. I've been doing rough corrections on the final output based on intermediate material chokes, and using it to find said intermediate chokes too

[daring siren]

I honestly don't usually care how much something can/actually makes. I do care to know if it makes what I need.

[iron root]

what you need is usually some kind of number, and rate calculator spits out a number
You can't take said number at face value when things get crazy with modules/recipes but it can be adjusted up/down based on the shenanigans in the crafting chain. I was using it to pretty great effect in this way in a SE run
Like I was running into fluid shortages, could rate calculator the consuming machines, and the producing machines in another part of the factory to figure out roughly how much more I needed
So... i guess more or less I was using it to shortcut the very basic math and was doing the more in-depth compensations myself

now
will most people use it like this?
Probably not, and that is a whole other can of worms (and is relevant to whether or not itll make it into the main game)
but to say its only useful with perfect ratios isnt necessarily correct I think, based on my experience

[daring siren]

I would use it if it showed me the actual calculated rate, and how many extra buildings I have.

Instead it first assumes everything is running at a maximum rate, and then shows deviations from this - not useful.

And even this actual rate is very hard to calculate, as you have belts and inserters

Anything that involves loops now requires a matrix solver and that's a whole can of worms

[iron root]

showing actual rate/extra buildings would be more usable I think, but you pointed out the technical hurdles to doing so (matrix solver)
and we have those in other mods but theyre not that clean click and drag ui
Far as the belt/inserter thing goes too I was also using rate calculator on single machines just to check input/output numbers too, again shortcutting the low-level math when machines get crazy with modules

[daring siren]

A "real rate" on the hover panel for a single building would be great.

[iron root]

'real rate' as in like, those basic input/output numbers based on machine speed/prod? Or something more empirical based on a machines actual crafts over time

[daring siren]

Both would be nice, but the first is more likely

https://mods.factorio.com/mod/assemblyanalyst
does the latter

[sick dock]

This is only true if the machine’s output capacity is backed up. A machine will eat resources as its speed allows regardless of its prod level

[iron root]

fewer resources per unit output is really the exact phrasing meant I think, but yeah

if you throw prod onto a machine but don't decrease its speed, it takes in the same amount of input, it just makes more output

[sick dock]

Yes. Esp in belt bases, things will eventually backup to meet your bottlenecks speed, but this can take awhile depending on how much buffers you have

Although with recycling loops you’ll have to be more careful 🤣
You can easily eat up a whole mine making legendary substations if you forget to set limits

[half raptor]

There is always one resource that will be the limiting reagent, if all ingredients are backed up, you need more production machines and/or output belts

[sick dock]

Do we have confirmation of the % power of , , ?

[daring siren]

25% for

10% for

Following other entity strengths, should be 15%

[sick dock]

See this is where I get confused because 10% seems to be a 10x factor

[daring siren]

It is

[sick dock]

has 1% chance to up a tier, yes?

[daring siren]

Yes

They do it 10x because it is integers

[sick dock]

The mechanics reveal showed the EMP had a 20% chance to up with 5 *

20% / 5 / 1.6 = 2.5%

[daring siren]

When 25% is multiplied by 2.5x, it becomes 62.5, rounded down to 62. That's why we get 248% or 24.8% for 4 legendaries.

[sick dock]

Always round down to interger, even if .5 or greater?

[daring siren]

That's how they did it

Truncated

[sick dock]

Never mind, I’m blind. They’re

[half raptor]

So each is 2.5% chance but at status that goes up to 6.25, whichll probably be rounded to 6.2% and I bet the game knows that that two of them is 12.5% instead of 12.4, seems like an easy check in the game

So many RNG rolls per tick.... This is going to be crazy

[daring siren]

No. It literally is 12.4 instead of 12.5, which is why we get 24.8% instead of 25% for 4

[half raptor]

It's a small difference.... But an aggravating one

Whatever it's all RNG anyway, it'll go between 23% to 27% over the course of a few seconds, you'd have to measure for a few hours to confirm the 24.8%

[tall sandal]

honestly the way it works is so fucked I can barely believe they allowed it

[jaunty citrus]

One value is rounded by 0.05: "My disappointment is immeasurable, and my day is ruined."

It is pretty jank, but is probably not worth the effort of changing all the other values in the module logic to fix it.

[tall sandal]

not just the rounding, but taking the displayed % value and diving it by 10 to get the real % value

[jaunty citrus]

It does confuse me that given they already are dividing by 10 they didn't just make it 10x more just to fix the rounding for

[tall sandal]

its all very strange

it would make so much more sense if +25% quality meant a 25% chance to get a higher quality item

instead of 2.5% for some reason

it feels like something that will get fixed post release

the rounding bit is the chrry on top for weirdness

[terse lagoon]

The problem is how they represent that value in memory. The obvious solution is to just multiply it by 10 (or 100 by that matter). So instead of 25 internally it would be 250. But it also will change literally every module value. And they don't want to do it, cause it will break oh so many mods

Tho they probably should

[unborn flax]

Internally the game already uses fixed point numbers a lot (typically multiples of 1/256) for things like map positions.

So changing module bonus resolution to multiples of 1/256 would work, e.g. 6.25 == 1600/256. However this still wouldn't be entirely clean in that you'd be mixing percents (i.e. decimal) with hexadecimal.

[remote scaffold]

[sacred totem]

Didn’t this whole thing happen because modules are already fixed point uint8?

[unborn flax]

Do you mean https://lua-api.factorio.com/latest/types/EffectValue.html ? I don't want to make assertions beyond my understanding, but I believe that bonus percentages are internally stored as integers - so the fact you can specify 20% as 0.2 is just syntactic sugar at the prototype definition level.

[sacred totem]

fixed point numbers are just integers with sugar on top :V

but yes, i do mean the effectvalue

[sick dock]

Thankfully quality is a very uncontroversial mechanic besides this minor point 😎

And yeah, I think both points are super weird. Like why is it listed as 10x it's "up chance" value, and why is it 24.8%

[fallow flint]

It's a 1 in 500 loss comparatively

If it was actually 25%

Then no loss

[jaunty citrus]

What?

Oh, the rounding difference

[half raptor]

We need exact numbers for our randommess

[pure spire]

cant wait for better quality bot stuff

[kindred crater]

its going to be crucial for high spm megabases around the landing pad

it'll be a fun minigame maxing out the small area around it to unload

[pure spire]

we can set logistics for qualities right

[half raptor]

Yes, when selecting an item there is a little drop down menu on that window for quality

You can set a specific quality or you can select any quality

[pure spire]

can you select 2 at once in the same logistic slot or do you need to use 2 slots

[terse lagoon]

aren't there are 5 slots now?

[daring siren]

From 375:

Here, for example, the logistic request allows setting either one specific quality, or a quality condition.
In this case, we are trashing all of the assemblers with lower than rare quality.
Requesting items must use the '=' condition, as using the other quality comparisons would be ambiguous and inefficient, hence the lower number is greyed out.
https://cdn.factorio.com/assets/blog-sync/fff-375-logistic-request-with-quality.png

[pure spire]

say i want filter to only pickup purple and orange assemblers

oh nice

[half raptor]

If you have the filter slots. You can also set filter splitters to select greater than or equal to quality levels

[pure spire]

didn’t know the >= stuff

[half raptor]

Picture is worth a thousand explanations

[limpid gorge]

quality over quantity?

[kindred crater]

also quantity has a quality of its own

[pure spire]

have we seen a photo of the speed effect of quality on assemblers?

[kindred crater]

not sure if theres been many pictures showing the tooltip, but we know it improves their speed in the standard way quality steps up: 30%, 60, 90, 150

[daring siren]

https://cdn.factorio.com/assets/blog-sync/fff-375-assembling-machine-tooltip.png

[unborn flax]

Without wanting to reopen a particular can of worms: There are a couple of things missing in this picture. For instance maybe the devs have redone the personal roboport mk2 and don't want to show us yet. I found the other one more interesting.

[jaunty citrus]

The other one?

[charred gyro]

I see the assembler is missing power consumption info

[jaunty citrus]

I'm pretty sure they just cropped it so that it has the stats that are affected by quality

Power consumption isn't affected by quality, unless it's directly tied to something like charge rate

[unborn flax]

this could well be true

[pure spire]

So, the quality of spidertrons

equipment grid

+1 in EACH direction for EACH level??

Because that is absolutely insane

That'd make it 20x16 equipment grid

So, legendary spidertron plus legendary reactors, roboports, and lasers would make an absolute unit

[median verge]

Theres a reason legendary equipment takes so long to make

[pure spire]

Well even non-legendary

Rare spider troms are gonna be insane as well compared to what we have now on top of the i@provemnts to them already

[jaunty citrus]

It would be 15x9 for

[pure spire]

How so?

Or am I misunderstanding the wording

Each direction seems to me like +1 NSEW for each level, or does it mean that one level is +1 N, next +1 S

[charred gyro]

I would presume the latter. Each direction to me means each axis.

I only expect it to expand in 2 directions. The upper left will stay the upper left.

[jaunty citrus]

There is only 2 directions, Vertical and horizontal, rows and columns

[terse lagoon]

What about bi-direction? Or attack helicopter direction?

[jaunty citrus]

x axis and y axis

[half raptor]

And z axis

[daring siren]

And my axis!

[half raptor]

I mean making cross sectional bases was controversial but it's easy to switch between layers so you can see how your factory tower will work. Multi elevation rails is gonna be a godsend

[jaunty citrus]

3 dimensional equipment grids would probably get difficult to manage

[trail cradle]

Does anyone know whether or not we will be able to place productivity modules in machines crafting intermediates of a higher quality? Looking back at the Quality post, prod modules are conspicuously absent from the legendary electronic circuit assembler. I haven't seen the answer anywhere yet.

[kindred crater]

you can

[daring siren]

Yes, it is possible to mix quality modules and prod modules in a machine

[kindred crater]

and in fact, I think its been proven having both in at the same time is optimal

[daring siren]

It is optimal in assembler-recycler loops, for very high quality modules of both kinds

[trail cradle]

This is really interesting. I've heard people propose doing a recycling loop for the end product. But it might be worth doing the math on using a recycling loop on each of the lowest intermediaries that can be recycled into components to be crafted again. For example if you want a legendary productivity module 3. Instead of doing a recycling loop on the module, you run a recycling loop on green circuits and redcircuits (much cheaper than recycling a T3 modules) and pass those legendary circuits up the chain as well as using them for the blue circuits to ensure everything “up chain” is legendary by default. You could use prod modules on the blue circuit assembler and since you have legendary reds and greens you always get legendary blues.

This isn't the best example because we cant use prod mods in the module assemblers. But any long multi-step recipe could benefit from productivity multiplying your legendary stuff up the chain.

[half raptor]

So recycle loops does lose you material. Another approach is to allow each step of crafting a chance for quality, sorting the different quality ingredients, putting prod on anything that is legendary, and then at the end step you have x , y , z etc and then you recycle loop at the end.

From ore -> plate -> cable -> green circuit -> red circuit -> blue circuit -> module mk 2 -> module mk 3, that one possible chain had 7 steps that could have gained quality, 8 if you put quality modules on your miners

The logistical challenge is dealing with buildup of certain qualities that doesn't balance right and those buildup you can recycle and if you wind up with a buildup of legendary ingredients you can use the legendary as a replacement ingredient for a lower quality recipe, which the game will treat as the lowest quality ingredients in the machine

[trail cradle]

There are sooo many different approaches, that's what i really love about quality!

[half raptor]

Another interesting and simple approach is to recycle loop the ore

It means that a mine will spit out very few ores as it is recycle looping for legendary, but when you have all your mines spitting out legendary ore, then you are able to prod module the rest of the base

A recycled ore will return an ore 25% of the time and you can try for quality in the process

[trail cradle]

I never considered starting with ore, but I did think of plates in a foundry. I wrote a program that just crunches a huge numbers and tells you how many legendaries you get at the end if you recycle loop everything that isn't legendary. (In this case recycling the plates to get a 25% chance of a plate back but with the quality chances for that plate as well)

Throwing 100,000 iron plates into a recycler with 4 Legendary Quality modules returns something in the ballpark of 170ish legendary plates on average, so it's very lossy

I think using the recycler on things that just return themselves (25% of the time) to upgrade their quality is less efficient than on things that will return their ingredients (like green circuits) so that you can run those ingredients through a crafter with Quality mods and have a bigger chance of moving the Quality needle upwards.

But the ore strategy sure does make things simple, even if you only get a fraction of the ore you mine out of the ground to your factory, you can just know what you make will be legendary with prod all the way up.

[rotund egret]

yep, it's been one of the dominant theorycrafts since Quality was revealed. It was quite a contentious topic back then.

The lowest intermediate that you should recycle does depend on a few factors. Recycling ore is hampered by the fact that the Foundry exists, and you want to make use of its innate productivity. Since liquid ores don't have quality, it's best to recycle loop plates.

for plastic it's best to recycle coal, but I'm betting the unique building on Bacchus will throw a wrench in this plan

but that's the general idea. quality is in a way opposite to productivity. Prod is best used as late in the production chain as possible, while Quality is best used as early in the production chain as possible

the main advantage of early recycle looping is that it significantly reduces the footprint of the rest of your base

you frontload all the effort at the beginning of the chain, and then you can have a couple of machines working with quality stuff exclusively

[half raptor]

Ignoring quality, prod is most useful when you use it on every step since the effect becomes multiplicative, but yeah someone will theorycraft the correct answer and then when we can start researching prod bonus for ingredients the math will change all over again.

I personally plan to just go all in on quality and send stuff to a recycle station if destination is full for greater than 10 seconds. I designed an overflow station for legendary which will go to epic factories, then rare and so on since + 3 will yield base with chance for legendary. If my input buffer for is full across my whole base, then I don't see anything wrong with "downgrading"

I know it'll take a long time for the problem to happen, but like thanos it is inevitable and having a plan for possible use cases will be more beneficial than "I'll solve it later"

[rotund egret]

yeah, I meant prod has its greatest effect the later in the chain you do it -- prod in miners has lower value than prod in rocket silos

I'm probably just going to purge the quality setup whenever I unlock a new quality tier.

in the early game, I might use uncommons by leeching off of final product assemblers, but my first "real" setup for mass-production of items would be rares

and then I'd have to purge the system for the transitions to epic and legendary

[half raptor]

My starter base quality strategy is to only put normal junk on the bus and have any RNG quality items filter off to a storage chest. Fact is that with normal MK1 modules, it'll take a while to accumulate a lot of uncommon and rare but I can use what little I get for armor and personal equipment

They didn't say we'd unlock epic on Fulgora so it might unlock at Bwuhuo

Or Aquilo might just have both epic and legendary unlocks

[rotund egret]

I think they said Epic is one of the first 3 and Legendary the last planet

[terse lagoon]

Epic is on

[half raptor]

I didn't see them say it's on fulgora

Screenshot?

[terse lagoon]

Hm, i indeed can't find it. They mentioned only T3 modules in FFF, not the quality lvls

[rotund egret]

yep

[trail cradle]

About recycle looping plates, if you recycle loop x plates until you have nothing but legendaries, you’ll be left with ~0.17% of x.
Example: Recycle Looping 10,000 common plates will leave you with generally around 17 legendary plates. That seems super expensive, although you then benefit off of productivity (which now goes much higher) up the chain from there so you can make some of that loss up. The math is really interesting.

[rotund egret]

yep

plus it's very easy to set up. I'd rather do that and have an inefficient setup than do a huge production chain and recycle loop at the end

this way I can get by with 1-2 buildings doing each intermediate at very low throughputs

[trail cradle]

I never thought about the advantage that because your lossy stuff is at the bottom of the chain, the entire production process can stay much much smaller as you said, even if it ends up being less efficient in the end it could definitely be worth it!

One thing for sure is the more steps in the process the more chances you have to prod your legendary intermediates, and the better and better starting with legendary at the bottom looks.

[rotund egret]

I just want to say I feel extremely smug right now seeing people more or less accept that this is one of the ways to go. When quality was first revealed I got so much flak for saying I'd rather recycle ore. 😎

[rotund egret]

exactly

[rotund egret]

it gets even better when you factor in the bonus innate productivity from buildings.

You recycle loop copper plates from liquid ore using a Foundry, but then you use the EMP for the copper cables and circuits

using molten ores also has the advantage of allowing you to use the same train network to feed the setups as the rest of the base

since you're gonna be making lots of molten ore anyway

[trail cradle]

This is such a fun problem to solve, love how there are so many approaches with pros and cons for each

[rotund egret]

yeah!

while for some of the unique buildings you might want to start off recycling the end-products after all

since on Vulcanus you're only really interested in making higher-quality foundries, because big miners don't benefit that much from higher qualities (I'm probably going to leave them at uncommon), and belts don't benefit at all

it's not worth doubling up the chain just for them

plus, the more different intermediates they have, the more recycling loops you have to set up

[trail cradle]

Just came up with this, Since recycling plates is so incredibly lossy, you could recycle loop to just epic. Then you’d start with 10xmore plates then if you started with legendary. Then you prod up the production chain, and for the FINAL step where you can’t use prod, you use quality modules. With 4x legendary quality modules you could expect ~25% of your end product to jump up to legendary. So it would be about 2 and ½ times less resources used per legendary product then if you went legendary from the bottom.

[rotund egret]

yep!

though that still means setting up recycling loops for all your products

so it might be useful if you have, say, one thing that you're making in bulk, like module 3s

but not that useful for a mall

[trail cradle]

Ya, you're right. The starting from the bottom method does assume a lot of loops. And each of those loops would have a recycler(or many) with quality modules. so it would be expensive to start

[rotund egret]

it's definitely useful on Nauvis, where you have so much stuff to make

but for the other planets it might depend

definitely excited to see what's what

[trail cradle]

Indeed, once we get out hand on the expansion and actually start playing, we can get a feel for what is best

[half raptor]

So that's why my thing is going to be that I'm going to set a train interrupt that says "if destination full and 10 seconds have elapsed, go to Recycle station" that way i don't have to build a recycler for every intermediate, the trains will sort it out and then I can have trains pick up the resulting recycled intermediates to go deliver again by train interrupts

[trail cradle]

That's a really good idea! Then we can concentrate our best quality modules all into that one recycling station instead of spreading them over many loops

[half raptor]

Indeed and it's a good way to monitor what you're recycling most and adjust your base

[daring siren]

Interesting that and share some qualities, because both rely on conditional recycling

[half raptor]

It's also the planet where it'll be more common to have higher quality iron and copper plates than it is to have quality

The planet is backwards

[pure spire]

is quality ammo a thing?

the item I mean

because that seems very inefficient

[chrome mauve]

isnt ammo fall under consumables that will get durability bonus?

[pure spire]

“ammo does more damage”

there’s 0 point in going for any higher tiers of ammo unless you are too cheap to make red ammo and you have thousands of iron sitting around

[median verge]

Think about energy dense stacks of ammo for outposts or space platforms

[patent cradle]

Disagree... Ammo assemblers cannot have prod mods so you could just put quality ones in the final step. Could be very useful for Space Platforms. Now if you are talking about recycling ammo to only get the highest tier, then I do agree

[kindred crater]

I'm not sure if this has been figured out already but am curious to see what you guys say... suppose you are aiming to get a legendary power armor 2, and for the sake of this exercise lets just assume that you have all of the legendary 's you need for whatever setup.

would it be more efficient to recycle loop the power armor 2 until you roll , or to recycle loop all of the required items to legendary and then craft it? or are they equal?

[median verge]

More efficient space or resource wise

Or time

[kindred crater]

I meant resource wise, but time is could be another important dimension if the difference between the two is substantial, although I don't expect it to be

[rotund egret]

yeah, ammo for personal use would benefit from quality

shooting up biter nests is much easier

and I'm making hundreds of thousands of ammo cartridges anyway

uncommons or rares are welcome, especially since the quality effect is multiplicative, not additive

[median verge]

You cant prod the armor, so prod in the base resources and cycle for quality with those?

Use quality recyclers to recycle circuts etc, then use epic/legendary items to craft a quality moduled armor. If it doesnt become legendary then scrap it

[half raptor]

Quality GUN makes sense

Though even a stack of would be nice but I wouldn't hold my breath for legendary bullets

[charred gyro]

Doesn't seem like there would be much point anyway if there aren't legendary enemies

[half raptor]

[charred gyro]

More variability among biters could be neat. They're very predictable in vanilla

[half raptor]

There's a blueprint military wall you can use on your whole base and never have to worry again, and just about everyone can make a decent defense blueprint that'll work

[charred gyro]

Maybe they could get stronger the further away from the center of the map you explore. No need to increase the challenge for those who aren't interested.

[half raptor]

Or you can have a new tier unlock every time you go to a new planet

So if your evolution is on medium biters, unlocking a planet could give us medium biters

Still on the same evo, but the quality goes up

[charred gyro]

I'm thinking evolution alone may be too simple to have a wider range of difficulties.

[half raptor]

Sure, I don't think we will hear about biters until after Bwuhuo reveal

Day 1 mod, rename Bacchus to Bwuhuo

[charred gyro]

I'm not sure we'll hear about them at all. I'd expect to see some teasers about enemies on the middle 3 planets before they cover biters, and they seem to be avoiding military stuff entirely.

Makes some sense as the more predictable enemies are, the less functional they would be at their role.

[half raptor]

Just suggesting.... Swimmers

[charred gyro]

I do hope they implement some of the stuff from rampant. Biters swimming short distances would be neat and not that much of a burden if it's an infrequent thing.

Same with tunneling

But I'd take just the improved AI if they don't want to add extra abilities like that.

[little orchid]

They could even do the corpse landfill thing, because landfill is now removable

[daring siren]

Hmmmmm

Using biters as a source for landfill?

Which can be recycled down to stone

[little orchid]

Oh dear...

[daring siren]

worms get larger range (as they are a turret), and nests have a chance to produce quality biters/spitters.

When an expansion squad has a unit turning into a base, it creates a nest/worm of that quality

Also... negative quality feature: All buildings have a big base quality bonus, but higher qualities make things worse. Speed/prod modules increase this number, while quality modules reduce it, possibly down to 0 to keep the base quality. So you keep getting junk on your belts, until you get them better.

[old lake]

It’s a bit annoying that the math always ends up upside down

[half raptor]

"I can't do reverse math" ~Travis Willingham

[uncut helm]

press undo on the calculator

[half raptor]

I'd post the YouTube clip that quote comes from but I know that YouTube spamming is discouraged here

[rotund egret]

do the <hyperlink>

[tall sandal]

I think its okay

[trail cradle]

Do we know the math for the 56 number that came from in the original quality fff#375?

“With this straightforward approach, if you want to produce items of legendary quality, and you already have enough legendary quality 3 modules . . . the legendary items are 56 times more expensive than normal items.”

My current understanding of the mechanics around quality modules and recyclers don’t lead to that number.

[tall sandal]

yes, 56x is a lie

its not real

it cant hurt you

#friday-facts message

numerous community members, myself included, have written mathematical models and simulations and none of them have reproduced it. even boskid using the game itself could not reproduce it

[trail cradle]

oh wow! that's actually a lot closer to what I'm calculating

something in the ball park for 250x more expensive when starting with common inputs

[tall sandal]

here's some other numbers you can use to check your results
#friday-facts message
#friday-facts message
#friday-facts message
#friday-facts message
#friday-facts message
(there are plenty more from various community members but these are from the developers themselves)

[trail cradle]

Currently I just have a program using an RNG, but I'm still working on the actual

Thanks for all the info! This is really useful

[tall sandal]

btw the table in the FFF is also subtly wrong #friday-facts message

1st row should be 9%, 0.9%, 0.09% .... uhhh whatever is left

[trail cradle]

Oh that's BIG. I'll have to redue my math.

[tall sandal]

yeah

[burnt reef]

9%, 0.9%, 0.09%, 0.01%

the table's odds would be correct if they were for "this rarity or better"

(with unmentioned odds at 100%)

[terse lagoon]

It is

Was confirmed by devs

[dense kraken]

didn't they also confirm they're going to fix the value rounding as well?

so that the max is 25% instead of 24.8%

[terse lagoon]

Pretty sure they confirmed they won't fix it. But mabe they later changed their opinion, dunno, if they did - i missed it

[daring siren]

They can fix it if they make it 1000x instead of 100x... that said, I hope that the UI is float 1x

[jolly swift]

after reading through the past ~1000 messages here my mind is absolutely blown

On the surface it’s such a simple feature but as soon as you start designing it becomes nightmarishly complicated and I love it

my take on how to handle quality on nauvis:
qmods in miners, have one smelter for quality copper and iron plates with qmod electric furnaces. uncommon copper/iron plates will go on their own dedicated 1/2 belt on bus, and rare ones will be stored for personal use.

any overflow quality items goes on main bus with common items to minimize complexity

[half raptor]

I couldn't imagine designing with quality intermediates in mind without trains and the new schedule interrupt system

Now I can put a mess of products on the same train and it'll sort it at the various drop stations, though I'm going to have a dedicated train for the production blocks expected level result and another train for any upgrades that'll run less frequently

[rotund egret]

that's a surefire way to brick your factory if you aren't careful.

remember that quality items are treated as completely separate items, meaning they don't stack in assemblers

and as anyone who's misplaced a damaged wall for military science can tell you, even one item with incompatible stacking can ruin your day

[heavy ember]

Didn't they say you could use higher quality items in lower quality recipes, they were just treated lower quality in that case?

[rotund egret]

you can, but different quality items still don't stack

[daring siren]

Only in recipes

[rotund egret]

for instance:

takes 3x

if the assembler first inserts 1x () , it would have to wait for 2 more ()

and you cannot insert any more common wires until the craft is finished

[daring siren]

Specifically requiring 3x

When it's set to , it can only accept

[sacred totem]

you could have dedicated assemblers to the task of doing high-quality stuff

like, q1 the whole thing, then q2, q3, etc at the end of the stack

[daring siren]

Recipe circuits will help reduce this

[jolly swift]

I think early game having manual buffers for overflow quality stuff is a lot easier than designing a base with perfect ratios

[jolly swift]

So modifying this, store all rare plates, prioritize uncommons for science, excess uncommon copper is combined with common iron to make common green chips

[half raptor]

My bus is going to be all and any quality upgrades are going to overflow chests for making my early personal equipment and first space platforms

[jolly swift]

honestly that’s probably better

2% more science isn’t worth using up quality stuff

[rotund egret]

yeah, science is arguably the worst place for quality

unless you're at the endgame and can afford the filtering required for some uncommon unproddable sciences

[rotund egret]

yeah, but you'd either need dedicated assemblers for each quality or some fancy inserter <> chest filtering

or else the line will get clogged up

mixing qualities like that only really works for science (since the labs take 1x of each science so you can't clog it)

[half raptor]

It's also pretty easy at the product output line to set a priority splitter to send or greater off to the side to load into chests. At chances, it's not going to fill the chest very quickly

Sushi factories will be able to handle quality pretty well though since there is no end of line on a sushi belt

Hmm... Sushi bus?

[rotund egret]

we have read full belt contents now 👀

[tall sandal]

and stacking, and super high speed belts

[rotund egret]

you'd still need separate assemblers to make sure they don't accidentally pick the epic green circuit and break

[half raptor]

Sushi does hurt belt optimizations since groups of similar items are processed together in groups. Sushi breaks that some

[tall sandal]

or just a few assemblers that automatically set their recipes based on need

[half raptor]

If you set an assembler to make an product, the inserters will only pick up ingredients unless you select allowing for any quality ingredient

[half raptor]

This!

[tall sandal]

cracked out legendary assemblers with 250% speed and beacons

surrounded by 4x sushi green belts stacked 4x tall

[rotund egret]

yeah, the issue is that leaves a lot of duplicate assemblers

[half raptor]

You can also set it based on a availability, measuring the belts you can say "if there's and , put the signal on a wire and depending on which signals get priority by the it'll set it because it has ingredients availabke

[rotund egret]

so I'd rather just funnel any quality resources to a special place to be processed

hmm, setting recipes is going to be such a big can of worms

I can't wait

[half raptor]

A bus and a sushi base that takes quality output from the bus line

Having an assembler set by circuit means that you can put your highest quality modules on the fewest assemblers

Since you can sort the signals by index, you can have a few assemblers reading index 1, a few reading index 2 and so on so that you can have machines dedicated to the the top X signals

Or if you're stubborn and old-school, you set one assembler for each recipe and quality and let your sushi base just work with permanent recipes.

But I like to figure out some circuit controls to choose based on what is present

[rotund egret]

yeah, that gets pretty insane

even moreso after you unlock bots

[half raptor]

Then just keep any and all normal items on the bus

[rotund egret]

the 1-assembler mall meta

[half raptor]

You joke, but I guarantee you there will be a week 1 blueprint for this very thing

[rotund egret]

no joke

[half raptor]

If you set up 10 assemblers and tell each to read indexes 1-10, you will have 10 different recipes and will always be crafting the top 10 signals (or bottom 10)

[rotund egret]

yeah, 1 assembler is a bit too barebones, but a handful with indexes will do great

actually it depends, if you don't do any end-chain fishing for quality and only use quality ingredients, an assembler with max speed modules and beacons, surrounded by requester chests and one active provider will probably be enough

actually, the biggest roadblock to the 1-assembler mall meta is that you'll want to use foundries/EMPs/etc for their specific items

leading to a ...4-assembler mall meta

[sacred totem]

nah, gotta parralelize

get that 16 mall meta going

four threads

[half raptor]

So you could set (for example) that if is greater than 0, set to 1

If and is greater than 0, set to 2 because I'd want to prioritize making that over gears.

For rare we can multiply this priority by 10 since wed love to craft rares over uncommons anyway, epic by 100 and legendary by 1000 to force higher qualities to have higher values.

Finally because we are indexing our signals, if we have 7 different signals and 10 assemblers reading the indexes, we'd be crafting all 7 of our signals and 3 would probably just retain their previous recipe

It doesn't lose its recipe, just changes at a signal

However you determine your priority values, you can multiply your qualities to artificially prioritize sushi belt crafting epic over uncommon, though any legendary materials should probably filter out of your sushi line to go to a line filled with prod modules

This whole idea is early to midgame, I don't see a megabase running well on this

Main point I was making is you don't set it based on what you need since it might take a while to get materials, rather it crafts intermediates based on what is present. You can always set other machines to craft the finished machines with any quality intermediates you wind up with

[chrome mauve]

just gib me Quality now so that I cant test my solutions if they work

[half raptor]

We have the janky quality mod that simulates the RNG factors

[daring siren]

Qualities in RimWorld have Normal and Legendary

So it's not weird they are called Legendary

[jaunty citrus]

New Rimworld expansion

[daring siren]

DARK SCIENCES

[jaunty citrus]

"Legendary muffalo wool baby romper"

My favourite is when my guy makes a legendary wooden dining chair

[abstract condor]

i wish it wasnt random

like for intermediates a quality 1 always makes , quality 2 always makes , quality 3 always makes

and if you use 2 modules it doubles

[jaunty citrus]

It being random is kinda the whole point

[abstract condor]

so 2 qual 1s make , 2 qual 2s make , 2 qual 3s make

[abstract condor]

how

Then for buildings and tertiary products, you need 70% of the ingredients to be quality N, and the other 30% to be quality N-1, to get a qualitg N product

[jaunty citrus]

Well I don't see how just putting some modules into a machine and having it always spit out is interesting

[abstract condor]

That way if i really want to make just one high quality roboport i know exactly what i need to make that

[abstract condor]

make the quality modules exponentially increase craft time

a qual 1 is +200%, then +400%, then +800%

[jaunty citrus]

That's even less interesting, now you just need to wait a long time.

[abstract condor]

as opposed to either waiting a long time OR getting lucky with the system as it is now

[jaunty citrus]

No, because you need to keep it fed with additional resources, it becomes a strain on your production

[abstract condor]

so add a negative production bonus on top of that

[jaunty citrus]

And there are many different approaches you can take, not just "you have to wait"

[abstract condor]

how?

the only different approach you can take now is to be lucky

[jaunty citrus]

Do you just put quality modules in the end product machines? do you try to skim off some quality intermediaries from regular production? do you use recycle loops to get high quality items as quickly as possible, but waste a lot of resources?

[abstract condor]

mathematically you can make it take the same time as the random system would take on average
and consume the same amount of resources

[jaunty citrus]

So what's the issue then

If you scale big enough they function the same anyway

[abstract condor]

identical for large scale quality operations, identical for smaller ones aside from no luck

also machines have deterministic output so no need to worry about sorting

[jaunty citrus]

Maybe the devs want sorting to be a part of dealing with quality?

It's not that hard, you can just use a filter inserter

[abstract condor]

could have using multiple quality modules multiplies the penalty

[jaunty citrus]

That's not how modules work though, the bonuses they apply stack additively

[abstract condor]

so applying 2 qual 1 modules would have the same effect as 1 qual 2, but significantly more resource and time intensive

[abstract condor]

make an exception for qual

ooo what if science could have quality
and when used in labs, a high quality science pack was researched at the same rate but counted for more

[jaunty citrus]

Quality affects every item in the game, including science packs

Quality science packs get additional durability, so they last longer in labs

up to +500% at

[abstract condor]

I thought science was excepted like belts?

[abstract condor]

thats boring

[jaunty citrus]

Belts also get quality, they get +150% health at

[abstract condor]

what

just for hp?

i thought there were a few exceptions

landfill?

[jaunty citrus]

There are no exceptions, every item can be given quality

Quality landfill would presumable have no effect, as once placed it becomes a tile. Tiles don't get quality

Unless landfill is used in a recipe, then the quality would be useful as an intermediary

[abstract condor]

fiddle with the ratios so that it is the same productivity to use legendary science packs but obv significantly faster as they count for more

a leg science pack would count for like 10 science packs

or something

[jaunty citrus]

A leg science pack counts for 6 science packs

[abstract condor]

yeah currently

wait ill get the ratios

[jaunty citrus]

Even if you could get guarunteed quality increase, it wouldn't be better than using productivity, and there is no 'perfect ratio' to make it better, because it depends on how many steps there are in crafting the science pack

Especially in space age, where productivity is getting multiplicatively better

[abstract condor]

ok i couldnt find a number for quality intermedates

but without them items are 56 times as expensive

[abstract condor]

it would be better because it would be faster

[rotund egret]

quality science isn't worth it except for quality-ing the intermediates of un-proddable sciences like green, military and purple. you get it essentially for free

it does take a bit of sorting though

[abstract condor]

for the same amount of labs and SPM

[rotund egret]

OH, you can actually use stack inserters with quality filters to make sure you don't clog the machines, since they only swing if their hands are full

so you can use custom hand sizes

[abstract condor]

what if the way they worked was that they take far more resources and power but if they are all of high quality the research speed is insane

good for infinite science

[rotund egret]

you're pretty much never bottlenecked by research speed, just by the quantity

[abstract condor]

such that it was still cheaper than building wide

[abstract condor]

artillery shooting speed 56 begs to differ

[rotund egret]

so you have the tens of thousands of resources ready to go?

[abstract condor]

no by speed i mean how long it takes to actually complete the technilogy

[rotund egret]

research speed only affects the number of labs

technologies require X amount of science packs to complete

[abstract condor]

im proposing that quality science packs count for more packs when researched

[rotund egret]

which is the main bottleneck, not how fast they're processed

they already do

but the amount isn't worth it

[abstract condor]

no they just last longer

which is the same thing if you dont care about time only prod

[jaunty citrus]

How would it be anything other than slower? It's the same number of crafting steps with added logistical complexity and no speed beacons allowed.

[jaunty citrus]

Yes, so one lasts 6x as long as one , doing 6x the research

[abstract condor]

my suggestion would make it take the same amount of time
so one is the same as N but all at once

[jaunty citrus]

That wouldn't be an improvement over just placing 6x as many labs

[abstract condor]

N being some number that doesnt completely break the game

[abstract condor]

yeah i think thats the main issue

maybe higher tier labs that are very expensive? i think its boring that for some items like radars and offshore pumps, the base item unlocked from the start is as good as you will ever get

aside from modules and beacons

[jaunty citrus]

It's to prevent clutter, you don't want multiple variants of offshore pumps that do the same thing, when just placing 2 of them would work.

And hey, quality fixes that, everything gets quality, and most entities are affected by said quality in more ways than just health

If you want a better offshore pump, make a higher quality one

[abstract condor]

thats true

[abstract condor]

no but a more advanced version
like one that takes coolant and has a larger footprint or something

[fading fern]

x6 value for one bottle doesn't sound like it's worth it compared to the cost of 6 bottles.
Maybe the potency should increase in multiple of 2 or something. 1, 2, 4, 8, and 16.

Unless there's more to quality than just using recycling loops.

[dense kraken]

without using recycling the cost of a legendary science pack is the same as the cost of a normal one

you should rather compare with the use of prod modules

they'll give you 40% more items (assuming assembling machine 3 and mk3 module) while a single step of quality will give you around 28% more effectivity

but successive steps are more beneficial to the quality modules than the prod ones actually it's not the case, for two steps find +96% for prod and +56% for qual

also you have less throughput to handle (something something quality over quantity)

[rotund egret]

yeah, a flaw in the "quality up purple science" is that you'd need more machines for the intermediates

since you can't use speed modules

[half raptor]

With modules you're looking at a 100% bonus meaning you get a base doubling of science packs. Having 4 quality modules only gives 25% chance and its only a doubling from normal to uncommon. Upgrades from higher qualities only gives a 1 pack upgrade.

That being said, having quality ingredients put through a machine filled with prod modules will give you the 100% bonus on getting quality science so it might be worth investing in quality ingredients but I think most people agree that a science will just be intermediate sinks

I still don't know if a science lab can handle multiple qualities, say for instance you have 2 and 1 in the lab researching a tech. Will it just research together with the have twice the durabiliy or will we need to pair with it to research?

[rotund egret]

it's probably compatible

my post above was referring to filling machines with quality ingredients for the recipes that can't take prod (e.g. green, purple and military science)

but the issue with that is you can't use speed modules on the machines

so I'm not sure it's worth it for the modest increase in science pack efficiency

[covert ridge]

Please add a quality tier "glitched" that's objectively better than the previous one, but inserters have a chance to fling whatever they're holding in a random direction and corner belts sometimes speed up and launch shit all over the place and bots have faulty memory and sometimes just drop whatever they're carrying

[jaunty citrus]

It isn't, that's my point

[jaunty citrus]

there is a lot more to quality than just recycle loops

[half raptor]

Honestly, filtering your unlucky intermediates toward science makes sense and send you quality intermediates toward machines and modules

[jaunty citrus]

You're going to get some crazy bias putting the poll here

[kindred crater]

yeah I think thats a valid point lol

[sacred totem]

im gonna think about it only as hard as googling "best qulity method factorio"

[half raptor]

We all have the methods we think will work, and then some genius is going to post the ideal setup on Reddit and everyone will just follow that

[kindred crater]

its certainly not going to be as straightforward as a bp book for balancers, but there will be some "good enough" bps that work for most players

programmable assemblers will probably be a big part of that

[jaunty citrus]

A programmable assembler bot setup is probably going to be as close to a universal solution as you can get

[heavy ember]

And it's going to get a stupid name on the first day that doesn't make sense and can't be changed because community already ran away with it

[cerulean harbor]

the cool thing is that the best quality method isn't a single thing

there are different things to optimize for

[heavy ember]

Just like there's no best beacon setup?
Reddit hivemind will take the first remotely good looking setup and start calling it optimal. And it will be nearly impossible to change once they decide on it

[cerulean harbor]

i don't think reddit does that lol

[jaunty citrus]

Some people might conflate most common with most optimal but i don't think everyone does that. 8 beacon setups are most common in my experience because they strike a good balance, but they're not optimal in any one category.

[half raptor]

I have decided that I would set up my train schedule to take stuff to a recycler train station if it's destination is full and 10 seconds have elapsed. When it comes to dealing with overflow items (what a problem to have) I decided to set up each of my drop stations with a secondary station in the same drop area that allows for overflow legendaries which you can set assemblers to use in place of lower quality ingredients. This way I have a recycle path for all qualities of materials. If all of my legendary overflow chests fill.... I'll burn that bridge when I get to it.

[rotund egret]

if all your legendary overflow chests are full then you don't really need the quality setup to work

[half raptor]

One thought is that I'd set a siren to send an alert when a blocks ingredients are full, it'll tell me "hey you should build another of these blocks or you're going to start recycling stuff"

[rotund egret]

maybe disable all quality stations if the overflow is fulll

[half raptor]

So if my production cell has nearly full ingredient stations and it's being controlled via variable priorities for dropoff, that means that the best way to prevent from recycling rare iron and rare cable is to set up another rare green circuit block

[half raptor]

You know that's not a terrible solution since belts can be flipped on and off according to signals. Honestly I know that it's inevitable to have overflow of legendaries if you recycle enough (and eat a lot of raw materials) so it'll probably need constant tweaking and babysitting. If you want a perfectly balanced system and you want to reduce the quantity of materials on the belt, recycle looping ore to make it all legendary is probably the only way to make everything simple yet legendary

Might do that particular solution on a creative save

My modular train block is built around a "I don't want to do math", I just want my stuff sorted according to calculated priority and to just decide based on buffer sizes. If my smelter/foundry lines are full of ore, build more until my ore buffer goes down again

The idea of that doesn't translate well if you have an abundance of copper plate since according to my previous logic you'd simply build more blocks of things you have enough ingredients for and unless you can make more that logic might tell you to make more even if you don't need it. The other solution is to let the copper plate recycle and hope for upgrades but now you're accumulating in that direction and could eventually wind up with too much legendary copper

The trick is to find a logical modular solution that will account for inattention and silly miscalculations

[dense kraken]

what I like about quality in this context is that it is new

setting up a factory for a new recipe is the usual process

while handling random quality stuff is very interesting

like, with new recipe you'd still use prod modules, but with quality you can ask yourself which is better where

a bit like nuclear and its non-even ratios, there are many different solutions with different pros and cons, which I enjoy analysing and comparing

I would even argue that satisfactory's alternate recipe would be better than new recipes for improved part, because if the part is improved you don't ask yourself if you would use the recipe or not

[opal haven]

Mostly same here. One problem with new recipes for stat improvements for me is kind of "lack of excitement" - setting new production chains for something radically new just feels much better in comparison, while quality can get bigger numbers just fine

[half raptor]

I like the multiple paths to the same destination discussions. Some people are going prod all the way and then quality loop at the end. Some are going to go for quality on every step and prod anything that uses legendary ingredients

At the end of the day, that legendary spidertron and legendary MK2 (3?) armor is going to be amazing

[dense kraken]

them with legendary equipment*

that's gonna be so expensive

but oh so strong

[rotund egret]

I've seen some Janky Quality clips of fully legendary spidertrons

more like Speedertrons

[daring siren]

Yah ColWill had a full PaMk2 and they had some walking, so they could easily fill them with equipment

[half raptor]

Getting full legendary personal equipment unmodded is going to be the Pinnacle of flexing

[trail cradle]

When I first saw quality I thought it looked weird and having new recipes for the better things would be a more interesting approach. But after thinking about it for a while (and playing with some mods) I realized a big problem with new recipes for improved things. If you have a recipe you can lock it further away in the tech tree, but once the player gets there the old recipe can easily become obsolete. You can try to keep the old recipe valid by making the improved recipe REALLY expensive, and this can work. But that causes other problems and you can end up with a bunch of recipes you don’t use and things start to get cluttered.

[daring siren]

You could say it will be an achievement.

[half raptor]

Instead of better recipes we get the same recipes but a stat bonus on the finished good and multiple ways to go for it. I know several people who are going to the entire process and set up a loop at the end, which will work and will make intermediate sorting easier I suppose

[daring siren]

I'm going to 2 prong it. on some of the the , in the middle, and on final products.

I agree that with enough in the middle, the price of a recycling loop can be worth it for the simplicity

It's important to remember that getting early on some items is worth it

[final nimbus]

I start with the decision to use wherever possible and in any other place (and miners)
Then... figure it out from there.

[daring siren]

Miners I'm still half-half on using or

[final nimbus]

up until I stop caring about pollution or power draw

[kindred crater]

until and get unlocked, I think I'll just stick with recycling loops to get what I want quality wise, without messing with quality in miners. once the good stuff gets unlocked, I'll invest into a more serious quality processing area and switch some ore outposts over to

and then from there, having -equipped outposts vs normal ones have a higher priority for their train stops so they're more jam resistant

[half raptor]

So many paths and choices, that's what I like in a game like this

Just give us beta keys already so we can try out our ideas

[heavy ember]

Will beta test for free

[half raptor]

Will pay to beta test

[iron root]

multi-objective optimization and pareto surfaces

i believe 8-8 beacon setups give the best production per unit area in your factory

and linked with that is the most production per resource invested in modules
but 12-1 setups are better for ups so if you go big big those are good (and then theres the abominations of beacon-maxing plus DI)

[daring siren]

12-1 costs more in but less in . Depending on the qualities you want, it may be worth it

[kindred crater]

if there is a canon on the progression of making an increasingly higher throughput/optimized megabase, it seems like 8-8 layouts come before 12-1, fully enclosing single machines

[opal haven]

8-8 gives the best return on investments, better than 12-1 and no beacon setups - really good choice to grow your factory the fastest

[kindred crater]

agreed. 12-1 certainly has its place, but going straight to that is like not using quality until you get or better

[half raptor]

If only speed beacons didn't put a quality penalty

Then again I guess it prevents people from setting up the same old builds

[kindred crater]

theres very little sidestepping when paying the price for a larger volume of quality stuff

quality machines make a difference, but not enough to make a huge difference

[half raptor]

Quality machines just increase speed, more speed reduces down the amount of machines needed

[kindred crater]

absolutely, I'm just saying that benefit isn't as strong as if we could beacon -equipped machines, like you were saying

30-150% vs 300% and well beyond

[daring siren]

Bulk production goes to science and can use beacons. Quality production is for mall stuff. That said, with the price of quality stuff, we get some equivalence between the chains

TBF, beaconed prodded production for everything but final items which is where you quality and recycle, is probably the most simple way to do it, with not many downsides

Interestingly, considering 4 and 4 , if the recipe accepts , then it doesn't matter if you put prod in the building, or quality.
If we start with ingredients for 8 items, using prod gives 16 items, recycled down to 3 ingredients of same quality and 1 higher quality. Then goes again through prod you get 6 + 2.
If we use qual, the 8 ingredients makes 6 + 2 directly.

With the prod method, we can use in , but not in the ♻️, but those are very fast and quite small too.

But yea, we calculated in the past that 2q2p is optimal in this case, by a large margin.

[half raptor]

How much of a margin vs 4q

[daring siren]

like 2x IIRC?

[kindred crater]

the exact numbers are buried somewhere in the FFF channel

[half raptor]

I'm just wondering over the math of that considering 4q and 4p are close

[daring siren]

4q and 4p are not just close, they are exact

[kindred crater]

btw, do you remember the optimal ratio for the 5 slot EMP? 3Q?

nevermind, found some of the convo #friday-facts message

[half raptor]

I'm trying to work through the math, are you supposing you're recycling the normal products once for this math with a receipt to get the final result?

[daring siren]

Anyways for 2q2p it's 79.5 for 1 , while 4p or 4q is ~154

[half raptor]

Okay you're supposing the average to make a legendary starting with normal. My thought process is to go all in quality and don't recycle until I get overflow so that each step of crafting to to has a chance to upgrade and only recycle if your request stations are full across the basr

Advanced machines can be 5-8 craft steps from ore