Quality

[plain wind]

required message to start thread

[whole jungle]

56

[bright topaz]

[sacred totem]

legendary

[final nimbus]

This is a high-quality thread

[north yew]

putting quality modules in your intermediate production is for big stinky dumb doo doo heads

[kindred crater]

4K Quality

[fast gull]

what were the odds that you made this quality thread?

[sacred totem]

.025%

[fast gull]

need to know for science :3

[plain wind]

50/50, it either happened or it didn't

[wary seal]

I'll move my question to here.. I understand quality modules make things better, but if I put quality 2 into things then does my chance of getting quality 2 go up?

[fast gull]

xD

[kindred crater]

looking forward to daily arguments taking place here

[plain wind]

hey, we call those discussions here

[north yew]

relative to eachother, quality chances always stay the same. quality modules affect the yes upgrade or no upgrade chance

[wary seal]

[final nimbus]

It's in the FFF

[wary seal]

Yeah but the FFF is so far away

[plain wind]

probably should have the quality FFF pinned

https://www.factorio.com/blog/post/fff-375

[north yew]

eg if all your ingredients are common, the proportion of uncommon, rare, epic, and legendary will be the same regardless of what quality modules are used. however, you'll get far more common outputs with lower quality modules

[whole jungle]

What if this was the real April Fools prank all along

[north yew]

the easy way to think about it is that the thing rolls for upgrade based on the quality modules in the machine. if the upgrade succeeds, it rolls again, but with a strict 10% chance this time. keeps going with 10% chances until it doesn't get an upgrade or hits max quality

[plain wind]

I wish

[uncut helm]

[fast gull]

Isnt quality optional tho?

[north yew]

yes

[plain wind]

it's not required

[north yew]

from what we know SA will be very hard without it though

[plain wind]

but that's like saying fast inserters are not required

[fast gull]

Yeh just something for us min maxxers 😛

[plain wind]

and beacons and modules are not required

[north yew]

iirc it was JG himself who said that one platform with all is worth 100 platforms with all

[opal haven]

Probably will be fine - just like you can easily beat 1.1 without ever using any beacons

[fast gull]

yeah I dont even start beacons until I am doing mega base stuff

[whole jungle]

I'm not sure why people think beacons are required. Is it not easier and more power efficient to build more machines?

[plain wind]

people don't think beacons are required

it's just generally the most efficient

not using them gets unwieldy due to the space it takes up

[north yew]

it's definitely simpler to just build more machines. you have to either plan ahead or rebuild your manufacturing to incorporate beacons

in terms of cheaper, though, that's not really the case

[opal haven]

With high tier prod modules - it's surprisingly much better to use beacons to get return of investments asap

[north yew]

assuming you want in your machines, beacons drastically reduce the number of modules you have to make

[plain wind]

here's an example, with an 8 beacon setup, you need like 13 furnaces to fill a blue belt
without modules, you need 72

6 times as much

[north yew]

(furnaces are the last thing that should get beaconed though)

(also the last thing that should get prod)

[plain wind]

yes, but for example's sake

simple math

[kindred crater]

imagine the setups required to fully saturate stacked belts now 😎

[plain wind]

dear god

384 furnaces to fill a stacked tier 4 belt

and you'd have to do some kajiggery with the stack inserters to actually do it

[kindred crater]

at that point, just use the effin beacons/modules lol

[north yew]

iirc fully beaconed em plants can saturate a (stacked) with only 2-3 machines making

[kindred crater]

the endgame builds are going to be absolute insanity

you won't necessarily be aiming for this, but some parts of production can/will shrink dramatically in size

maybe this is the true endgame of SA... as you approach singularity, production skyrockets as the footprint collapses into itself

[north yew]

even without quality footprints of everything shrink quite a bit

em plant's crafting speed is 2, for example. much fewer machines required to get the same output

[tall sandal]

finally, we will be able to count how many quality posts we get per day

(except we cant, discord doesnt let you search through a thread)

[sick dock]

I am excited for quality overall. Ik the randomness bothers people, but I think most will like it once they get their hands on it. It's good that if you have all the components at rare, you can guarantee craft a rare item. You'll just have to wait for all your intermediates to get qualitied up

[tall sandal]

I think its one of the deepest mechanics we've seen so far

[patent cradle]

its very veritcal

[north yew]

(you really don't want to quality intermediates)

[patent cradle]

agree and disagree

[jaunty citrus]

Quality intermediates are quite the challenge and probably not worth it until your scale warrants it.
Personally I don't see the benefits outweighing the drawbacks until then, but we'll see when the expansion drops.

[patent cradle]

Agreed... Quality at the end of the line only until you can have dedicated production and handling of quality from step 1 (quality in miners and every step of the chain after too).

at least thats how I currently plan to handle it but there is still a lot of unknowns that could change that mindset

[sick dock]

end product quality is only going to net you uncommons and the occassional rare

if you want to guarantee a rare early, you're going to need to use internediates

unless by end product you mean using qualiy in only the end product machines, then throwing them back into recyclers until you up-scale

ihaven't run the maths, but i don't think quality for intermediates is that bad

[north yew]

Three main arguments against quality intermediates:

• Logistical complexity: 5x as many different recipes, and each item needs to go to 5x as many places. Each point in the factory connects to 25x as many points as it does without quality. It's going to be extremely complicated to set up and will consume a massive amount of the most important resource in the game: the player's time and attention.
• Scale: You can't use speed modules or beacons, you aren't getting any productivity boosts, and the quality modules are slowing down all of your machines. Doing quality intermediates at any sort of scale requires massively more machines to be running at the same time. We're talking orders of magnitude here.
• The ratio problem: each ingredient takes a different number of steps. With each step, the ratio of high quality to low will change. Any individual step in the process can only utilize the ratio of its constituent ingredient with the shortest chain. Much of the rolling for quality will end up getting wasted because almost all recipes are bottlenecked by something with a short chain (eg steel).

[final nimbus]

It'll be easier to farm / in SA with the insane productivity increases available.
Less so with 2.0 + Quality, but you still get prod research to smooth things out.

[sick dock]

logistical: my quality area is quarantined to its own area. it will not be in the whole factory. i have no idea why people seem to think this would happen to anyone. no one would would have random quality stuff all over their factory.
scale: don't care, we have space. this is factorio, after all.
ratios: just wait....

[final nimbus]

you aren't getting any productivity boosts
lmao did you really just say that

[north yew]

even if it's just for a simple mall, it's still 25x as many connections from each point to each other

[sick dock]

what are you talking about

you roll to get what level you want, then you input it its next corresponding machine

i really don't think it's that complicamted

i don't think your'e going to care about all 4 quality levels, i'll put it that way

if yo uwant, you can just think about it as rare & legendary

yes, if you cared about all 5 levels of quality, it would get real complicated, but i don't think most people will play that way. i certainly won't

[north yew]

if you're planning to recycle loop then prod in intermediates is better

[sick dock]

i imagine most peopel will get to rare, leave nauvis, then either go into epic or legendary and start going megabasing, if they even get that far

[sick dock]

i haven't run the maths on this. i'l ltake your word for it, but how are you planning to get a legendary radar exactly?

[north yew]

prod in all intermediates, recycle loop radars

[sick dock]

i'm like 95% confident that rolling intermediates into legendary would be better, that is jus tmy intuition

i guess it behoove me to actually do the maths

[north yew]

it's the ratio problem

the math is really mindbending and complicated fair warning

but I've mathcrafted out some specific scenarios

[sick dock]

even if it is ultimately more efficient, i still appreciate we can guarantee a quality level if we have the components at that level

[north yew]

plus quality being the flashy new thing a lot of people seem to be forgetting that prod is really, really strong and there's a huge opportunity cost to not using it

[sick dock]

the only thing prod does is extend resources. if your resources are near infinite, i'm pretty sure you'd rather just use quality

in everything

[north yew]

nope, see point 2

[sick dock]

scale doesn't matter

[north yew]

with prod and speed beacons you can get the same output from like five em plants as 80 with quality modules

[sick dock]

ah, i msitook your point

[north yew]

doing quality intermediates is going to take a ton of extra work for little to no benefit

vs recycle loops being fairly easy and quick to set up

[kindred crater]

the hardest part is getting the legendary quality modules

[sick dock]

duh

[north yew]

em plant made that a lot more feasible though

like holy cow prod bonus on modules

[sick dock]

right?

most expensive item just got a hell of a sale

[kindred crater]

yeah it was 11x or something like that now to reach legendary

[sick dock]

i guess rocket silo is most expensive single item, but who counts that

i guess one point to intermediate qualities is, if yo udon't know what yo uwant to use them for, or have a later use for them

[north yew]

it's not just the prod on modules either, it's also that you're getting that same prod on green chips then red chips then blue chips

[sick dock]

you can preemtively set it up to get what you want later, rather then waiting to unlock a tech and then setup a recycler loop for it

esp very expensive items might take a while to stockpile resources for

rather then setting up: quality radar recycler loop, miner loop, interseter loop, pole loop, etc. etc

[north yew]

it's gonna take way more factory to get the same amount of research doing it like that

[sick dock]

i'm talking time, not resources

also, i don't think so. if we had 100 items we wanted to build, what is more space effiicient: a few intermediates builds, or 100 item recycler loops?

my bets on the intermediates

yes, you then have to hookup a few assemblers to actually make the end item, but you had to do that anyway

[north yew]

well the thing is even if you do quality intermediates you'll need the recycle loops anyway

[sick dock]

right, but like 10 loops instead of a 100

[north yew]

well generally I'm operating on the assumption that only caring about legendary and considering everything else to not exist is something that's only gonna be a good strategy for ultramegagigabases

generally I imagine the best approach strategically is to make smart decisions about each quality tier as you get them

and recycle looping purely for legendary is only something that's worth it in the super late game

[sick dock]

i think this is a mindset the deeply enfranchized have. its like peopel who say that everyone only cares aabout t3 modules. and they are wrong

i think t1 & t2 modules are very undeerrated. they are obv weaker, duh, but the paypack period for an iron gear assembler that is fully decked with t3 modules is rediculous

[north yew]

like as an easy example: I'm gonna get quality mods in my solar panel assembler ASAP and try to place down as many common ones on Nauvis as possible so that I have a stash of uncommon/rare for space platforms

[sick dock]

it will ofc eventually pay tiself back (even over the opportunity cost of the boost of the lower module), but it's not as needed as people seem to think it is

i think rares & epics will have their place. ofc legendary will alway be the best, and the megabasers will likely rush it regardless

[north yew]

and if I did quality intermediates to try to minmax how many rares I get, that would mean I'm getting far, far fewer commons to have on Nauvis and uncommons to use on less-imporant platforms or Vulcanus or something

[sick dock]

true. everyone will play different, quality is no exception. this is one of the beauties of factorio as a game. for me, i will likely scale all my stuff to a quality level once i get it, although how that'll play in practice we'll only find out while actually playing it

[north yew]

but I definitely agree w/ what you're saying. it's a pretty common gamer brain thing. there are two kinds of thing: the best thing that stands high above all the rest, and useless garbage

zero in between

[sick dock]

hah, yeah. i play a lot of competative games, i know the feeling quite well

i don't use t2 modules like at all. very rarely. but i will use t1 a lot. should never not use t3 for rockets, labs, probably yellow & purple science tho

3 tiers of module vs. 5 tiers of quality will feel different. that's a lot more varience. (well, really 4 additional quality levels). not to mention there are now 15 module levels now, effectively

and i can understand people not wanting to bother trying to quantify which one of those 15 prodmod levels is the good one, and instead just make the best one they can

certainly most people will deck their equipment & spidertrons out as much as they can. you could have like 8 legendary exoskeletons and be absolutely zoomin' around

[jaunty citrus]

The only clear 'best strategy' for quality is to use recycle loops on items you can get the highest productivity bonus on.
As soon as you can reach 300% prod (not as hard as you would think with prod research and EM plant). Recycle loops no longer void product.

[sick dock]

i was thinking about this. if you can actually get a machine to 300% productivty while still saving at least one module slot, you can use those extra slots for quality modules. so while you cannot make additional resources, you should (?) be able to infinitely scale those resources up in quality

[jaunty citrus]

You can put quality modules in recyclers, so you don't even need to extra slot in the machine

[sick dock]

ah, right

[jaunty citrus]

If you have legendary modules, you'd only need to get to Productivity level 13 to reach 300% prod in the EM plant, which is around ~530k science iirc

[final nimbus]

It's been noted by the devs that going straight for and ignoring intermediary levels is a trap.

[sick dock]

do we know the equation for inf research scaling?

right now it's a simple 2,500 * 2^x, i believe....

don't quote me on that

[jaunty citrus]

1000 * 1.5^k (iirc)

for prod research

[sick dock]

i think they meant early game

or at least for the majority of players, who aren't trying to megabase. it's a huge resource sink before all your lategame is setup with your nice producitivty bonuses

@north yew also i think i see what you mean about just using prodmod for intermediates if you're doing loops. just effectively get more chances to reroll

[jaunty citrus]

My interpretation is that trying to build a base using only legendary quality items is a massive waste of time, as you could be using resources to scale using lower quality or normal quality items instead of flushing them down the drain in a legendary recycle loop.

Which I would have to agree with

[sick dock]

for players who are only interested in legendary, they'll need to prepare for it

it's akin to saying you should make t1 modules to help you make t3 modules

[jaunty citrus]

Its similar to the beacon trap currently, you don't need to spam beacons as soon as you get access to them, you'll just cause yourself more issues than it fixes.

[sick dock]

beacons are power-hungry lil' bastards

[jaunty citrus]

Indeed! It rubs off on the machines around them too

Your 120MW base just turned into a 900MW base without really changing production numbers much

[sick dock]

on a related note, Nilaus echoed a thought of mine that efficiency modules might actually be almost-maybe viable on Fulgora, at least before you can landfill over the oil-sands

it'll really depend on how good the power generation of lightning bolts actually is, and i'm sure that's on the top of the devs minds when it comes to Fulgora

it is fitting that quality is....focalized on Fulgora, given that your island space is a little limited. higher quality machiens means you need less space for same output.

[jaunty citrus]

I had a similar discussion as well. Are way better in combination with other modules because the benefits scales but the negatives don't
So they become relatively more effective at their job when made a higher quality

[sick dock]

ngl, they could also just buff efficiency module t2 & t3. t3 gives....-50%. that is like nothing when using other modules

[jaunty citrus]

Efficiency modules are also bad because they have a huge opportunity cost for very little benefit, they have almost no impact a buildings power consumption when combined with other modules/beacons.
Because quality only increases the benefits, it increases the relative benefit that efficiency modules have when combined with other modules. You can have a and in a beacon and have a machine consume less power while being faster
From an earlier discussion I had

[sick dock]

they did have a lot of use in Mike Hendi's videos but that's cause he had a mountain of biters to contend with, and the reduction of pollution actually helped keep his base intact. most players obv wo'nt have such a big issue

[jaunty citrus]

If I remember my calcs correctly a would give machines around it 62.5% speed and -22.5% energy consumption

[wary seal]

I'm late to (re)join the conversation... but I can't wait for there to be situations where you do need lower tiers of quality

[sick dock]

10% productivity on a t3 prodmod...25% on legendary, right? that's insane value

[jaunty citrus]

and the would also consume relatively little power

[wary seal]

Like where the ratios are cleaner if you use the instead of items or something

[sick dock]

the devs might also consider making their new bulidings incredibly power hungry, which efficiency modules could help alleviate, at least in the midgame

[sick dock]

true, true

[jaunty citrus]

I expect there to be more cases where you just have a bunch of lower quality stuff to use, so you use it instead of the valuable items

[sick dock]

ik people meme on power poles, but there might be uses for the extended range in tight builds.

[jaunty citrus]

The did mention the foundry is very power hungry, so I'm expecting the 0.8-2MW range for that.

[wary seal]

I could also see solar/accumulator ratios being super refined now

[jaunty citrus]

A reaching the other side of an assembler will be 'legendary'

[sick dock]

i'd sure bloody hope that their em plant (EP?) is also powerhungry. it's a freaking energy conduit

[jaunty citrus]

I'm gonna stick to calling it the EM plant or the EMP.

[sick dock]

emp is decent

[jaunty citrus]

Context should help to not confuse it with electromagnetic pulse

[sick dock]

also harkens to actual EMPs, which also have to do with electricity

"Foundry" was such a short, fitting name

electromagnetic plant is just a big mouthful

[jaunty citrus]

Similar to why most people call the blue chips and not processing units

[sick dock]

yee

[jaunty citrus]

I'm interested to see what the ratios for other planets will be

Vulcanus has a 90 second day/night cycle and 400% solar production

[sick dock]

that whole thing makes me a little tilted at qualities naming conventions. i thought the devs already decided to give their entities more "serious" names, and let the community give colloquial names

the RPG names just sound too....video gamey. too muh like a mobile game

[kindred crater]

you can always just call it Q1-5

[jaunty citrus]

Count me in 'who else'

[sick dock]

me like quality 🙂

"legendary spidertrons" does sound like a hollywood movie plot ....i'd watch it, can't lie

[north yew]

the actual, practical argument for keeping the names as they are is that there's zero cognitive overhead to keeping track of the sequence

98% of people who play Factorio already have that exact sequence of names and colors memorized

[sick dock]

yeah. like 99% recognizable

yup

[wary seal]

I can get behind that... especially if the quality 0 (quality 1? Whatever the lowest tier is) was actually worse than normal

[jaunty citrus]

Every time I or someone else tries to come up with alternative names it still creates silly names for some items. Keeping them to 'normal' item rarity naming makes it simple for people who already play video games.

[north yew]

and tbh I like that it's a little bit silly there's room for fun in video games

[kindred crater]

there will surely be mods that change it if it bothers you that much anyway

[wary seal]

'Cause you could have it be like rudimentary, poor, average, great, and perfect

[jaunty citrus]

You just can't have 5 adjectives make sense for power armour, elevated rail support, and used uranium fuel cells at the same time

[sick dock]

i wouldn't say i hate silly...although i don't play factorio for goofiness. i got other game sfor that. it just felt like it was being conflated with other, very unrelated games

looty-shooty games are more then welcome to use the RPG naming. it's very fitted

[wary seal]

Also I really really hope someone makes a mod that causes structures with quality to have that quality degrade over time

[median verge]

That would be a script heavy mod

[jaunty citrus]

no

[sick dock]

i've heard so many people say this i'm sure it'll be a day 1 mod, if not day 0

[median verge]

But so cool

[north yew]

this. it's gonna be a bit silly no matter how you slice the cake, may as well embrace it

[sick dock]

double qualifiers are always a headache

[kindred crater]

the names kinda annoyed me at first too in all honesty, but after enough convo here and on the forums, I don't care anymore lol

[wary seal]

Idk if it would simplify it at all, but maybe just a hidden health bar and a spot of poison or something?

[sick dock]

i think there's a concern of UPS after a while

could potentialy be a ton of stats the game has to contantly track at all times

[wary seal]

When the hidden bar hits zero, downgrade one tier

Could also make it where the higher tiers last longer, highest tier is immune or something

[median verge]

Too many spare entities. It would effectively double the size of everythinf

[wary seal]

Both of you have good points. Just spitballin'

[median verge]

Double walls, double inserters, etc

[jaunty citrus]

Could be better to have it be based on machine cycles?

after 100k products finished, downgrade a tier?

wouldn't apply to a lot of things though

[median verge]

No good way to track products finished

[jaunty citrus]

The game already does though?

[median verge]

Not accessible to mods iirc

[sick dock]

ya, no shade. just thinking aloud. i cannot stand degredation mechanics, but more power to you

[kindred crater]

if all entities have a unique ID, you could have a relatively simple data structure to manage a list of items as well as the game tick they were placed. using that, you could have stuff incrementally degrade over time. I know nothing about lua, but surely that wouldn't be hard to implement that.

[median verge]

They can be fun but they are a pain to make and slow to run

[mortal junco]

each entity has a unit number actually, you just cant actually get them by unit number. its a personal gripe with the api for me

[wary seal]

That's not too huge of a hurdle - Wube's made changes to give access to modders before. Afaik it's usually just "give us a valid reason and we can have the conversation to see if it's possible"

[median verge]

Too many entities. It would balloon super fast especially if you iterate over them quickly

[wary seal]

Not that it's a guarantee, by any means, but it's not an impossible situation which is nice

[north yew]

why do you want your buildings to downgrade in quality that sounds miserable

[sick dock]

on the topic of mods, it is good that Wube revealed Quality mechanic very early. gave modders much time to think about how quality will integrate with their designs, esp the overhaul mods like Krastorio. alhotuhg idk how playable Krastorio or Space Ex will be with 2.0....i imagine they'll update it

[jaunty citrus]

I would expect not having a big list of all entities to be a precaution to bad mods

[median verge]

The main problem with any degredation is ballooning script runtime

[kindred crater]

for something like this, no way you'd have to calculate it every tick. every 15-60 sec could work, if not more

[wary seal]

Because I'm a glutton for punishment

[sick dock]

mmm, tasty masochism

[mortal junco]

ping me once 2.0 comes out and ill make you that mod. its not hard, the problematic part is remembering doing it

[north yew]

I love really difficult games too but there's a difference between fun difficulty and miserable difficulty and degrading buildings sounds like the latter to me

[jaunty citrus]

I find turning up pollution numbers on startup to be an effective way to have your factory degrade over time

[wary seal]

If I remember, sure lmao

[median verge]

Still too many

[mortal junco]

same problem for you i take it

there aint a remind me command in here either lol

[kindred crater]

personally I'd enjoy negative productivity if machines had a power deficit

[wary seal]

Wouldn't help without a release date lmao

[median verge]

Also brings the problem of how you combat the degredation. No point in having it if there is no solution

[kindred crater]

for instance, 90% power needs met = 9/10 items actually get crafted. you could see it as higher defect rate

[burnt mist]

It will release in 739 days... I think.

[mortal junco]

remind me on trigger defined by server would worok

[wary seal]

I was about to say, I don't think that's real... but I see it's a joke command

[sick dock]

in other survival games, degredation is no problem,but that's cause your degrading entity count is like the double digits....maybe tripple. we play factorio for taht juicy scalability, and could easily get 4 digits numbers of entities, 5, 6....

[kindred crater]

all good, I am not a fan of the idea anyway lol. just brainstorming

I'd find it fun 🤷‍♂️

[median verge]

So would i

[kindred crater]

suffering is part of my preferred gameplay

[median verge]

Just dont see a good way to do it

[sick dock]

that's a true gamer right there 😎

[wary seal]

One of my biggest problems with the game is that once it's done, it's done, so there's never any reason to revisit what you've done

[median verge]

But you dont wanna rebuild your factory every time

[north yew]

in like 30 or 40 years eventually you'll run out of ore and need to connect new patches

[wary seal]

It's part of why I build huge bases - anything to extend the time I can invest in doing something new rather than just building new mines and waiting for them to deplete

[sick dock]

(can't wait for big mining drills, although again i hope they make a more ...interesting name)

[wary seal]

(Then again I just launched a run with 1000x research cost & expensive recipes... I'm only 16 hours away from having splitters!)

[north yew]

idk my preferred way to get more out of Factorio is to play through to rocket launch again but on harder settings this time

[mortal junco]

with or without

[sick dock]

i'm a habitual restarter

[north yew]

expensive recipes makes the game so much harder than it seems at first

[mortal junco]

isnt it like 3 recipes total

[north yew]

it's like eight

but it's green chips and gears and steel

[kindred crater]

you have more patience than me lol. I felt like I was losing my mind on a 20x game

[sick dock]

which are p important, turns out

[wary seal]

Max frequency & max size, but evolution is almost fully disabled, polution is turned off, starting size maxed and peaceful mode on.

So basically when I'm ready to deal with them they'll be a real pain, but I don't have to put up with it while I'm dealing with the massive costs of starting

[median verge]

You would hate my 73x pY run

[mortal junco]

whats that number wtf

[median verge]

GF gave me a num between 20 and 100

[wary seal]

Wait wait we're offtopic!

What's important is that each of us plays in a way that improves the quality of our own experiences

[mortal junco]

i mean thats not that unusual

[wary seal]

Bam. Saved it.

[north yew]

not all who wander are lost

this applies to conversation too

[sick dock]

i'm so proud of this. puns make everything better...EVERYTHING

[wary seal]

You might even say they improve the quality!

Is it repetitive yet?

[sick dock]

never

we will make the same dumb puns until my friends want ot punch me

[kindred crater]

you're just upcycling it, all good

[north yew]

yeah each time you use it is another roll for legendary

[sick dock]

i can't roll with normal jokes. my puns are just too epic 😎

[wary seal]

Oh I so wish someone who had the access could make a bot that just randomly assigns a reaction of the quality emoji anytime someone uses the word "quality" now

Weighted like the game and everything

[median verge]

That would be amazinf

You have implicit quality bonus based on what tags you have

[quasi cedar]

What's everyone's plan before getting the recycler? I think putting Q1 modules in intermediates could be good and diverting the higher quality to a mall for space platform parts. Use the standard quality majority for regular science

[kindred crater]

another easy one is putting them to ammo production, even within production, so you can siphon off the better ammo for expansion/killing nearby nests

[sick dock]

i think early you just want in end products...but not sure about. i guess if you make normal quality intermediates, you can just chuck them backwherever you want them

certainly should consider making rare t2 prod mods for your labs and rocket silo

not having a recycler puts a damper on early quality, perhaps intentionally so. gotta wait till Fulgora to make the most of it

[jaunty citrus]

My current plan is to stick a couple of low tier quality modules into regular production for things like circuits and plates, then siphon the quality ones off for personal equipment, space platform items and

[jaunty citrus]

Well I'm not using prods that early on anyway, might as well use quality and get a small boost to personal equipment

Definitely not using them for intermediaries later though, prod is just too good

[quasi cedar]

Yeah I think there will be a definite shift before/after getting the recycler

[jaunty citrus]

Except it isn't because it decreases the speed of the machines and increases the power consumption

Once your past 'early on' then yeah modules become viable

You don't need a lot of quality items, you just need enough for some power armour and equipment

[north yew]

early on establishing new ore outposts is costly both in time and resources, so stretching them further by getting in all possible slots ASAP is very worth imho

[quasi cedar]

And an 8x8 armour grid is actually 14% more space than 7x7. I'm assuming MK2 won't be unlocked until later

[jaunty citrus]

I don't really consider setting up multiple outposts 'early on' but that might just be me

[quasi cedar]

I guess we'll want a few more outposts earlier in SA though as we'll be away from the planet for a period of time setting up bases on new planets

[jaunty citrus]

Managing outposts will be much easier with the new QoL tools including improved remote view

[north yew]

I can definitely see myself overengineering my Nauvis base to function for a very long time without me b/c I'm scared to leave it

[jaunty citrus]

100%. Good thing it'll be easier to do so

Logistics groups actually making resupply trains nice to use

[north yew]

I'm starting to suspect a roboport wagon as a final way to really juice remote view

[jaunty citrus]

Nah, that's what spidertrons are for

Not really? I suspect we get them on one of the middle planets, my bets on Bacchus

You can use logistics groups to more easily set up adjustable resupply train stops

[tall sandal]

putting quality modules in my mall and using filter inserters to split the output into different chests by quality. I'll limit each chest to 1 stack and I'll get 1 or 2 uncommon items if I'm lucky

[tall sandal]

But we cant set up the train stops themselves remotely unless we have robotport coverage. We cant get roboport coverage on Ful

[jaunty citrus]

well, no, but I never said you could.
It's not really that much of an issue, that applies to all outposts, it's what spidertrons there to do

[tall sandal]

On other planets you can chain robotports together to get construction coverage. My hope is we will get logistic bot train cars though

[jaunty citrus]

Or maybe it's a deliberate limitation of fulgora that would be aleviated by another tool you get on, say, Bacchus

I'm not so sure this will be that much of an issue given you will be there to set up the planet's base initially, you won't need to set up outposts remotely

[opal haven]

How long handcrafting res sci for automation took?

[jaunty citrus]

Automation is always 10 science packs iirc

it's not changed by the multiplier

[opal haven]

oh. that's slightly less of a torture

[daring siren]

[wary seal]

Like Sticks said, automation is always 10 packs... that 16 hours is with 5 or 6 labs placed (I can't remember if I did the last one yet) and reds automated

[opal haven]

yeah, that's just crazy)

[wary seal]

That's what I do 🙂

[opal haven]

what was the order of stuff? like automation -> small automatic red -> splitters -> proper automation?

[wary seal]

Basically, yeah. Ultimately it's probably gonna be designing a compact build that takes raw ore and makes 100spm, and then just stamp it down everywhere

But I've got hundreds of hours before that's a thing

[old lake]

What are you playing? Pyanodon?

[north yew]

x1000 science

[fallow flint]

so you gotta handcraft 10000 to get automation? that's rough

dang good lol, that'd be 14 hours almost

[daring siren]

[neat shard]

I still maintain that it'd feel a lot more satisfying if gave a nice round +100% boost instead of +90%.

[jaunty citrus]

Nah, I maintain that having each level giving the same multiple is more satisfying

[neat shard]

I mean, I can appreciate that too, but that's not how they've stated it's going to work.

Legendary is a +60% jump compared to +30% for the others.

[limpid gorge]

woah, secret quality thread

[jaunty citrus]

So the same 30% multiple.
+0 +1 +2 +3 +5 for each quality level times the boosted effect

[sacred totem]

when has factorio ever gave you that kind of satisfaction?

pretty sure the dev team actively avoids nice round numbers

[north yew]

if you don't use prod most of the crafting recipes have fairly clean ratios

but yeah prod throws that out the window and I think that's part of the tradeoff

[final nimbus]

None of the assemblers crafting at speed 1 was always odd to me

[neat shard]

gives you a clean round number

[sacred totem]

150 = 5 * 30 btw

[neat shard]

That paired with the +60% jump means players will feel forced to go towards legendary when they wouldn't really have fun playing that way

[jaunty citrus]

Then don't play that way (idk from what we know you don't even get legendary items until quite late in the game so you probably can't be 'forced' at all tbh)

[neat shard]

That would be an easier pill to swallow if gave +100%, because the round number would make it a satisfying stopping point

[final nimbus]

A single is still several in cost.

[jaunty citrus]

Going +30, +60, +100, +150 is nonsensical though and doesn't work at all with the +1, +2, +3, +5 for power pole ranges
Nor does it work for the +10% items like robots, should get +33.3% battery? get +33.3% range? cause currently it's +10, +20, +30, +50

[final nimbus]

[Citation Needed]

[jaunty citrus]

#friday-facts message
https://cdn.factorio.com/assets/blog-sync/fff-397-quality.png

[final nimbus]

Wrong citation

[jaunty citrus]

I got robot battery and turret range mixed up

The point was that not every bonus is +30% per level

and the levels currently go 0, 1, 2, 3, 5

[jaunty citrus]

For a good summary: https://wiki.factorio.com/Quality

[sacred totem]

kwalitee

[median verge]

That optimal section seems arbitrary. Anyone care to explain?

[sacred totem]

basically some people ran the numbers and those were the results

#friday-facts message

[north yew]

smh my head that optimal section advises very suboptimal strategies

[sacred totem]

this optimal quality stuff goes over my head

im gonna ignore it all until FSA comes out, then ill take a look when the boffins do numbers with the actual game

[north yew]

well the nice thing is that Wube are good at game design and you don't have to have a math PhD to do the best thing: recycle loop at the end

[sacred totem]

its weirdly not the best always, though, is it?

isnt that the point of that spreadsheet?

though its definitely easier to just do the loop at the end

[north yew]

#1215078107334057984 message

the tl;dr is that introducing quality modules in your intermediates makes it orders of magnitude more complicated to set up, more costly, and nets you little to no benefit

I should note though that I don't hate the idea of putting quality modules in drills then sending normal ore to science production and quality ore to a special quality section

but then you just use your best prod modules on all the intermediates in the quality section

[sacred totem]

i think it could work with bots, as there is essnetially zero added complexity besides pasting down another 4 sets of machiens

[north yew]

sure, but that's a fairly expensive way to solve one out of the three major problems

[sacred totem]

bots? expensive?

[north yew]

yeah, it'll take thousands of them constantly active

[sacred totem]

though, yeah, theres nothing to be done about plastic and steel being plastic and steel

[sacred totem]

thats not very expensive...

[north yew]

it's a lot more expensive than belts or trains

[sacred totem]

whats the cost metric here? ups?

[north yew]

electricity

[sacred totem]

[north yew]

power is cheap on Nauvis

when you have lategame tech

[sacred totem]

nuclear is not lategame

neither is solar

having holes in your solar array is kinda ugly, but what can you do without cliff explossives or lavafill

fulgora doesnt really have that though

and its the planet of quality...

[north yew]

and even if you assume the cost of thousands of constantly operating bots is negligible, it's still one out of three major problems solved

[jaunty citrus]

Putting prod into everything drastically scales your spm and scaling potential, putting quality into every drastically scales logistical complexity

I don't really disagree with putting quality into miners. Get quality ores and put them into a small quality factory.
Probably wouldn't do it on nauvis. Fulgora seems like a decent option, directly getting from scrap seems pretty good

[sacred totem]

fulgora is already the land of priority splitters

adding more cra pto the outputs of drills wont hurt anyone

[north yew]

yeah quality miners are fine, since the proportion of not-normal quality stuff will always be the same relative to eachother the ratio problem doesn't exist if you only do quality in miners and final products

still probably not something you do in the early-midgame b/c that's a lot of extra pollution compared to

also having it so that you use your best prod modules in your quality section and roll quality on stuff that's already uncommon or better is a pretty big benefit

[jaunty citrus]

Miners already get prod bonus from research, so there isn't as much opportunity cost for not using in them as opposed to assemblers.
Skimming some ores off the top means you need ~10% (depends on quality modules) of the quality loops for the same quality output, you're starting at at least

[sacred totem]

kwiletay

[jaunty citrus]

Would probably make the most sense to funnel the raw resources into a modular smart bot mall

mid game quality strat probably

[daring siren]

Or even a beltless cargo wagon auto-switching mall

[terse lagoon]

So, will repost it there. For anyone who interested how quality looping is solved (using matrices). Watch on pc, or it won't format properly.
First, matrixes themself and their labels:

[1000000, 0, 0, 0, 0] <-> V0 - input vector. Each cell represent quality, as u can see we start with 1000000 common batches on ingredients. 1 batch is what is needed to make 1 output thingy (not including prod bonuses)

[1.5, 0,   0,   0,   0]
[0,   1.5, 0,   0,   0]
[0,   0,   1.5, 0,   0] <-> P - prod bonuses from assemblers, which have 2 prod T3Q5 modules 
[0,   0,   0,   1.5, 0]         (prod, that's why "P")
[0,   0,   0,   0,   2]         (last one is x2 cause it uses all legendary ingridients, so no need to try to up quality)

[0.876,  0.1116, 0.01116, 0.001116, 0.000124]
[0,      0.876,  0.1116,  0.01116,  0.00124 ]
[0,      0,      0.876,   0.1116,   0.0124  ] <-> A - quality matrix if u use 2 quality T3Q5 modules 
[0,      0,      0,       0.876,    0.124   ]         (we use it in assemblers, that's why "A")
[0,      0,      0,       0,        1       ]         (if we used 4 quality modules it would be equal to R matrix)

[0.25, 0,    0,    0,    0   ]
[0,    0.25, 0,    0,    0   ]
[0,    0,    0.25, 0,    0   ] <-> S - recycler eats 3/4 resourses so we left with only 1/4 of them 
[0,    0,    0,    0.25, 0   ]         ("S" cause "Scam")
[0,    0,    0,    0,    0.25]

[0.752,  0.2232, 0.02232, 0.002232, 0.000248]
[0,      0.752,  0.2232,  0.02232,  0.00248 ]
[0,      0,      0.752,   0.2232,   0.0248  ] <-> R - quality matrix if u use 4 quality T3Q5 modules 
[0,      0,      0,       0.752,    0.248   ]         (we do it in recycler, that's why "R")
[0,      0,      0,       0,        1       ]

Now what we do with all the staff. We basically cycle it until there is nothing left beside legendary stuff. Steps:

1. We produce stuff from those butches of materials, getting products. We calculate V0 * (P * A) = V0p ("p" - processed). Result with given example values:

V0p = [1314000, 167400, 16740, 1674, 186]

2. We take our legendary stuff out of that process and put it in some chest, cause we don't want it to get recycled: c = 186, so our vector is:

V0c = [1314000, 167400, 16740, 1674, 0] and c = 186

3. Now we put all there is into recycle getting out batches of ingredients (losing 75% in the process obviously). We calculate V0c * (S * R) = V1. Result with given example values:

V1 = [247032, 104792.4, 19820.16, 2916.108, 392.832] and c = 186

4. Now we completed the full circle. We don't take legendary ingredients out (even tho we technically can) cause what we going after is legendary products. Now we repeat steps 1-3, but using V1 instead of V0 until we V2 and new increased "c". And then repeat again and again.
5. We repeat that cycle until sum of V becomes barely noticeable, practically zero (i used 0.5 in my program, but i also start with 1 trillion batches). At that point we stop calculation and look at what "c" we get. Ratio of 1000000/c should be just shy of 80 in that particular example.
   Conclusion: good luck turning that thing into a program and playing around with it. Or smt.
   Anyway. That's all I've got for today... go away now

[pure spire]

can we have quality splitters

[daring siren]

Yes. But they only have more health

[north yew]

alternatively, if you meant filters for specific qualities and combinations of qualities, the answer is yes

[daring siren]

And relation conditions (<, >, <=, >=, !=) for specific qualities or specific items

[north yew]

although sometime I wonder if since there's only five options anyway if it would be easier to just do five checkboxes instead of relational ones like that

[daring siren]

There can be 254 qualities

[north yew]

why 254 instead of 255 or 256

[tall sandal]

secret final quality

[terse lagoon]

0 is reserved pretty sure. So it leaves 255 i think?

a, right, i found why 254

but on the serious note, this:

[sacred totem]

maybe in the future the migration will be removed, and we will get 255 levels

perfectly uint8 max, as all things sohuld be

[neat shard]

I've got a vision of having several parallel factories for differing quality levels

Whenever a resource of any kind wins the diceroll for increased quality, it gets passed to the next factory over

[tall sandal]

thats the way! build it and quality shall be yours

[terse lagoon]

The problem is that it almost never worth it to insert quality modules over prod, cause of how much u lose. Even if thing doesn't accept prod modules, it rarely worth it, cause down the production chain some components where the ones that accepted prod, so it breaks quality. There are only 2 exceptions from that:

• Perfect looping (aka 13 lvl of prod research in new building with ). In that case u can for free raise the quality and use full prod bonus anyway.
• Mining. Cause of mining prod research u get diminishing returns from putting prod there, so u more incentivised to put quality modules there. And then do whatever with them. But u won't get that much quality ores until super lategame that it won't be worth it to send it anywhere other then mall

[daring siren]

is the planet. 4 modules in big miners, 4 modules in recycler. Even with lower tier modules, you can get a nice spread of qualities from that

[little orchid]

And it most certainly helps that the quality is on ready-made intermediates, even really high-tier ones sometimes

[daring siren]

Kinda... longer chains allow for higher qualities

[little orchid]

Well, now you can get quality without the long chain

Just dig it up out of the ground

Ah, wait "higher" as in higher compared to this method, I see it now

[daring siren]

With the standard production chain, you have issues making quality because you have a lot of quality but not enough quality etc

In it flips the chain, allowing for more high quality lower level items, and less high quality high level items.

[north yew]

maybe lategame I'll think about it but the supply chain on is already so complex without quality. gonna be a real challenge to throw quality in the mix.

notable, though, is that since all scrap is "smelted" in the recycler, you can avoid the ratio problem w/ quality modules in drills and recyclers

the first stage of recyclers, at least

if you throw quality modules in to any further stages then you're gonna throw off your ratios

[neat shard]

It's not optimal, no. But it sounds like a fun build.

[daring siren]

You get quality "for free" on the recyclers, because they can't accept prod, and are super fast already

We don't know their power usage, btw

So then, do you recycle the quality items down to plates or not?

We'll have to see the various recipes

Belts will be made from gears and recycled gears

Inserters will require recycling and/or

[north yew]

power cost and module construction cost would be the two prohibitive factors when you're first getting set up

actually I think the ratio problem might be worse on even if you try to standardize quality exposure

I'll have to ponder it a bit

[rare thistle]

you can use some surprisingly simple algebra to solve for what that balance point will be. that way you can just use that number in the input vector and calculate the final answer in one iteration, (and zero out slash ignore the common output from the R matrix).
in english, the formula reads "the total input equals the new input plus the leftovers from the previous cycle" and in the balanced case the leftovers will also be related to "total input". anyways, the algebra is just solving for x in x = 1 + kx where k is expected percentage of inputs leftover from the previous step, interestingly, it'd be the product of the first elements of all your matrices k = 1.5 * 0.876 * 0.25 * 0.752. x = 1 + 0.247x -> x(1-0.247) = 1 -> x = 1.328 which in english means "one item of input in a system where output is fed back into the input is worth 1.328 items in an equivalent system without that feedback"

[fast gull]

Slap a priority splitter on that 33% and you have yourself a nice assembly line :3

[north yew]

had a conversation yesterday that made me want to do the math, so I did. here's an idea of how quality intermediates would massively expand the scale of the factory. let's first imagine a simple mid-lategame setup:

• four em plants making
• 5x and two
• all normal quality
  this would put out about 3.3 per second. to make an equivalent number of using em plants with quality modules instead, you would need nine of them. consumption would be 33% higher, so you'd need to be building more machines for those outputs too.

over double the number of machines needed as well as higher resource demands along every single step of the chain will balloon the scale of a factory very fast

that effect only becomes more pronounced when you start to factor in higher quality modules and em plants, since higher quality prod/speed modules will increase output while higher quality quality modules will not

[daring siren]

Quality?

[tall sandal]

is quality ... terrible?

[jaunty citrus]

what is... Quality?

[daring siren]

How quality... is?

[sage vapor]

why is quality?

[heavy ember]

But most importantly:

When is quality?

[burnt mist]

It will release in 839 days... I think.

[burnt mist]

It will release in 554 days... I think.

[ember kernel]

I'm not following that one. Higher tier quality modules will for the same input improve the average quality of the output.

It's also slightly apples vs oranges because and are not the same.

That aside, I think quality will matter most on the space platforms and malls.

They're limited by their weight to thrust ratio. Higher quality is the only way to improve that categorically.

[daring siren]

Also in personal stuff

Personal and space are the big ones. Also quality in the quality module makers 😛

[ember kernel]

The other interesting thing about Fulgora is that it brings a new design constraint.

Normally in Factorio, the easiest design are either output or throughput constraint. So you let your belts back up.

With the recycler loop, it looks like they must be input constrained. Otherwise the factory can jam.

And I guess quality will have the same constraint. If recyclers are used.

[daring siren]

There's some logic for when to double recycle things that we worked out in the main SA channel, unrelated to quality

[tall sandal]

whats the gist?

[daring siren]

In vanilla, cracking is easy and can be done with e.g. " > 20000", considering is used more than everything.
Here we don't know if there are chains like this which are obvious, so instead we would go more like " > + some_bias", but instead it would be e.g. " > + + some_bias"

So it will kinda balance everything

[ember kernel]

Ah. Just recycle any overflow down to the most basic ingredients? And consume ingredients produced by recyclers first?

Guess that explains the train priory system.

[daring siren]

Not just recycle. Only recycle if the output is low, because otherwise it's more likely to stuck or get wasted

[ember kernel]

That's not quite like oil though. With oil you can always crack down any overflow. Unless I misunderstood something.

Ah. I guess the problem is that we can also produce the thing we just recycled.

[daring siren]

It'll be an interesting challenge

[ember kernel]

So for a given inventory T, we have a L and U bound. If below the L bound we produce and above the U bound we recycle.

Then if we have 5 assemblers for T, one for each quality level. Each takes the minimum level ingredients for its quality target.

Repeat for every item in the production chain and that should be a minimal waste production line.

That works for a mall anyway.

Yeah. And being input constraint will help.

In fact, being input constraint is the only resource efficient way. Everything else will have a recycler-assembler loop somewhere or deadlock eventually.

[daring siren]

Somewhere yes, but eventually if everything is stuck on one thing, we will quality loop it. Of course, only if it's stuck while science is running.

[rare thistle]

worst case you can just recycle the lowest tier ingredients to 75% delete them

[daring siren]

In general, we'll have to see the reasonable setup to get sciences (+ rocket launches), and then see which items are in excess

We'll then have to find a useful way to use them, or just quality loop

(assuming items of quality are always needed somewhere, and are relatively low throughput)

[tall sandal]

I'm a bit confused. Why would we be recycling in the first place in this scenario?

[daring siren]

It was just an example... It's more like or whatever

If your is full, and you're low on and , then recycling it makes sense

[rare thistle]

concrete might be a good example to focus on?

[daring siren]

CONCRETE THE PLANET

[rare thistle]

(assuming that we don't end up exporting that to vulcanus)

[daring siren]

I don't see why we would need to export it to a planet that has foundries 😛

Reminder that foundries make concrete with +50% prod, but we don't know the exact recipe

[rare thistle]

true but if its used in this it might be worth it

[daring siren]

Unlikely. That's just a placeholder image.

It's tungsten + coal or whatever

[rare thistle]

I agree

but you never know

[rare thistle]

ok sorry way off on a tangent but.... my theory is that that is what you get when you cook tungsten in a regular furnace. then probably you cook it again in a foundry to make the plates

[tall sandal]

I guess I dont fully get it. Why not have the recycling condition be > 10000?

[rare thistle]

you would def do that too, but you'd also want to pull from chips if either iron or coils get low

[daring siren]

Because that will burn when we don't need the outputs

It will get stuck recycling the excess regardless if the factory is working or not

It's like turning to

[rare thistle]

ok but serious question: what are we doing with all the batteries?

[tall sandal]

I mean, the point of recycling is to get rid of them, right?

[daring siren]

Accumulators...

Only get rid of them when you need to

Also, recycle batteries to better accumulators.

[tall sandal]

Is this in the context of

[daring siren]

Yes

[tall sandal]

Okay. So in this case, you are recycling because you want the wires and plates. Not to get rid of

[daring siren]

Yes

[rare thistle]

true!

[daring siren]

But there are other ways to get those, only recycle if they are low and you have a lot to recycle + bias for priorities

needs a lot of accumulators

[tall sandal]

So for example, if you were low on , you could get it from or and you should decide which is better to recycle

[daring siren]

Yes.

It also comes from scrap directly

[rare thistle]

have we talked about how they confirmed the exact speed drop from quality modules here?

5% per module

[kindred crater]

so its the same level 1-3?

[rare thistle]

seems like it

[north yew]

It's just x^2 vs x^3. For prod and speed beacons, it's machine base speed * speed bonus * prod bonus, all three of which go up as quality increases. (Also notable, fewer machines and modules required so it's drastically cheaper to upgrade quality.) For machines with quality modules in them, increasing the quality of the machine itself increases the base speed, yes, but it also means the bite that quality modules take out of it is that much larger.

[rare thistle]

oh no, you're making a pretty good case for experimenting with a speed beacon on 5 quality EMPs

[north yew]

we did the math at one point on how many comes out of a fully juiced em plant

iirc it was like 120-140/s

[kindred crater]

so much horsepower out of such a small footprint

[rare thistle]

is there space to fit 16 beacons around it? my visualization is struggling rn

[north yew]

Also it's dubious how relevant how much upgrading the quality modules would really help. Hard to say it's even accurate to describe as x^2 with the ratio problem and all that

[rare thistle]

cuz 4x4

[north yew]

nah 12 is the limit for 4x4

[rare thistle]

:/ shucks

what's the math on the "8 beacon" setup with foundries direct inserting coils into emps :P

[north yew]

we can't know until we get the recipes

built-in prod and quality on modules makes any dreams of clean direct insert ratios fly out the window, though

[rare thistle]

we did the math on the crafting times a while back. based on rough estimates the foundry has a crafting speed of 8 and the coil recipe has a crafting time of ~4 (I theorize 3.7)

based on the animation

[ember kernel]

But what are you comparing?

Suppose we are starting from ores and oil and the production target is . Then there must be quality modules involved at some point.

So to maximize the output for a certain input, would it be better to start quality early in the chain or late?

For example if we start quality early and we have legendary inputs then the output will be legendary and can be prod modded.

Or we can try using basic inputs that were produced with compounding compounding productivity. And recycle the final product that doesn't meet the quality standard.

[north yew]

the original example was a specific case to demonstrate point two in this #1215078107334057984 message

the core of the argument is that rolling for quality on intermediates is a huge noob trap and is generally not worth it

if you only want best-possible-quality outputs, the benefit of rolling for quality in the intermediate chain is slim to none (mostly because of the massive opportunity cost of not using prod and the ratio problem making most of the quality rolling worthless). if you are willing to use a mix of qualities based on where it's needed, prod in the intermediate chain is massively better.\

[rare thistle]

maybe this is what the prod researches are meant to solve. they are the incentive to roll for quality at those items because otherwise there'd be no reason not to just roll the final item

(also if there's a significant enough time gap between getting modules and getting beacons I'm still going to quality module everything :P )

[charred gyro]

I imagine quality modules won't be useable in beacons anyway.

[rare thistle]

correct. but speed modules don't play nice with quality

[charred gyro]

They don't?

[north yew]

yep, speed modules have a built-in quality penalty

[charred gyro]

Oh. Ouch.

[north yew]

although it's hard capped at +0% so you don't have to worry about negative quality

[rare thistle]

do we know the number?

[north yew]

it's not directly confirmed but +0% is the default with no modules and it would be wild for them to let it go below zero imho

[rare thistle]

oh yeah for sure but do we know the speed module's reduction number?

[north yew]

oh, that hasn't been directly stated

[rare thistle]

kk

[north yew]

but even if it's small it still makes mixing speed and quality a very bad idea

[charred gyro]

Eh. We'll see. Seems like it should be balanced in a way that allows it to work fine. I imagine it's something like 1/5 of a quality module's bonus, which would be pretty easy to offset.

[north yew]

you're directly offsetting the bonus of one module by another module. it's using more module to get less module bonus

[charred gyro]

Until you have a lot of beacons for everything, anyway. With a 12-beacon setup (or maybe even 8-beacon setups) you'd be losing a lot more.

[rare thistle]

even if its just 1% you'd only make that tradeoff if space is constrained

or power 🤷

or if you don't care at all about wasting inputs

[charred gyro]

That may make it more of a matter of preference for you if there's no value that would prevent you from avoiding quality+speed together. If someone wants high quality stuff and they want it fast, I imagine both quality and speed modules would be part of their design.

[north yew]

well this goes back to the argument of why speed modules on their own are a waste: you can just build two assemblers

[charred gyro]

I see the similarities

[rare thistle]

I'm sure there's a niche scenario I'm not considering but I really can't see a reason

if you're far enough into the game to have beacons there's not much that'd force you to make the tradeoff

[heavy ember]

Why do we use blue belts then?
Might as well have two red belts and it would be better

[ember kernel]

Okay, that's the point I'm not following because once you hit legendary, you can have a production chain that uses productivity modules again.

If you recycle at the end you are recycling very expensive components.

[tall sandal]

Argument 2 and 3 isnt very convincing to me. I dont think people will be doing beaconed builds until mega endgame. And the ratio problem is just a minor issue imo. But argument 1 is the sticking point.

[north yew]

ratio problem is the big one

defeats the purpose if most of your rolling for quality is wasted

[north yew]

b/c sometimes space is constrained enough to ask for it. there's a limit to how many belts/inserters can fit near a machine

there isn't really a limit on how many machines you can build, at least not one that justifies the additional cost of speed modules (by themselves)

[north yew]

if you're gonna recycle loop, it doesn't matter where in the chain you do it. the cost increase of the whole chain will multiply the same amount

[tall sandal]

ratio problem is solvable by correclty splitting between producitivity and quality modules

[heavy ember]

I have low beacon builds before endgame occasionally.
For example this gc build

[rare thistle]

are you sure? I don't think that's true

[tall sandal]

thats's cool, I like it

[north yew]

it is kind of true, but solving it that way is very complicated and amplifies points 1 and 2 by 100 fold

[charred gyro]

I hope beacons are still available starting on Nauvis. They're such an impactful part of how anything can be produced that it'd suck to unlock them only later on another planet.

[tall sandal]

It must be, because one valid solution to the ratio problem is 100% prod

[heavy ember]

Why swap from stone to steel furnace?
Why t2/t3 assemblers

[north yew]

stone furnaces pollute less and consume less coal, t2/t3 assemblers have more module slots

wait they don't pollute less

they do consume 1/2 the coal though and that's relevant earlygame

[jaunty citrus]

Because t3 assemblers have both more modules slots and a higher base speed to better utilise those module slots

[north yew]

also t2 and t3 assemblers each represent a massive decrease in pollution per item produced b/c pollution goes down while speed goes up

[heavy ember]

I would argue that most t2/t3 assemblers never have their module slots filled before rocket is launched...

[north yew]

the way you're proposing by carefully coordinating quality and prod modules does work, but it's a very marginal benefit for a massive increase in complexity and difficulty of setting up, and on top of that it's gonna be full prod in most machines anyway

[tall sandal]

I mean exactly. Its the complexity that's convincing

[ember kernel]

There is one big loop at the end if you do quality there. But if you start early, that loop will be much smaller.

And depending on recipes and chain length the two might be competitive.

[jaunty citrus]

Also it's a method of scaling up vs out, why wouldn't you take +66% production by just upgrading the assemblers from t2 to t3

[rare thistle]

the solution to the module problem is going to be different for each target item and involves forgoing quality in the majority of machines

[north yew]

for example: with T3 modules, there's no point in rolling quality on any step inbetween and T3 modules because are a component of T3 modules

[heavy ember]

Can you loop ore?

[rare thistle]

yes but its obviously very inefficient

[north yew]

the loop at the end will waste less but the early loops will waste more

[jaunty citrus]

It will always be more efficient to loop something you craft, as you get two rolls per 3/4 loss instead of one

[tall sandal]

because you dont have enough inputs

[north yew]

@ember kernel keep in mind that there's effective loss in product at every single step that has a module slot that isn't being filled by prod

[ember kernel]

I thought the train system and parametrized blue prints solved that complexity. Throw in a few city blocks for each quality tier and it's done.

[jaunty citrus]

why would you upgrade them if you were starved of inputs, you would upgrade the input production first?

[tall sandal]

to be clear recycling ore just destroys 75% of the ore and gives you ore back. same with plates

[tall sandal]

yes, thats my point

[rare thistle]

mmm but mathematically you can count a higher quality item as multiple lower quality items compressed together

[jaunty citrus]

"Just" have a block for each quality, maybe late game I guess?

[ember kernel]

The loss from the recycler also compounds.

[jaunty citrus]

Okay? So build another outpost so you aren't starved then upgrade the assemblers?

[north yew]

chained prod is just a much steeper exponential curve

[rare thistle]

from what perspective?

[north yew]

going by 1.1 example, it's 1.4^n, where n is the number of assemblers in the chain. that's real steep. with quality, it's much more complicated to derive a formula, but it's certainly not as steep as 1.4^n

[north yew]

if you only care about best-possible-quality, you're going to need to recycle loop heavily even in the case where is in every single module slot in the chain

it's easy to assume that quality modules on intermediates is worthwhile if you only look at a single step. when you consider how it flows along a supply line, it becomes obvious it's a bad idea

[tall sandal]

I think we agree?, so lets drop it here

[north yew]

another easy example: roboframes take steel, regular engines take steel. they're both gonna be bottlenecked by the steel quality ratio. any quality rolling on regular engines and electric engines is wasted

[rare thistle]

this part I'm 100% on board with. steel here is causing the worst possible case of the ratio problem

[north yew]

the thing is this happens for every single recipe we know of basically

I've genuinely searched for a recipe that isn't bottlenecked by a short chain

there are few if any

exactly

almost all recipes in 1.1 have some combination of advanced stuff with long chains and simple stuff with short chains

[ember kernel]

So it is the recipe chain that matters. That's a much clearer argument.

[rare thistle]

ok I think you've convinced me so far. what if we consider pairs of recipes that share the long chains and have different short chains

[north yew]

yeah maybe I need to write up a post that more clearly explains the ratio problem

b/c in all cases when people disagree with me about it it's because they're not understanding it, not that I'm wrong about it

it genuinely is the big thing that undermines all theories about quality intermediates

[rare thistle]

mathematically, yes. but doesn't the stage of the game have a huge effect on it?

[north yew]

well the only theoretical workaround for the ratio problem involves building at a scale and complexity that would only be accessible lategame

[rare thistle]

like, the "wasted" items can be used for handcrafting or whatever

[ember kernel]

Nah. You were making me the wrong argument. The compounding prod isn't convincing because the recycler losses also compound.

It's the bottleneck on short chained simple ingredients that always push the compounding effects of the recycler to the end.

[north yew]

the message you replied to that started this current discussion literally used the ratio problem as its core reasoning

and then my next reply mentioned that the ratio problem is the thing that undermines the attempt for quality intermediates

[ember kernel]

The ratio problem sounds like something else.

[jaunty citrus]

The real strat is to use recycling loops on whatever recipe you can get closest to 300% productivity bonus

[ember kernel]

Oh and then sticks throws more fire on the fuel. 🤯

[north yew]

that's another strategy that is really only accessible in the very lategame

[jaunty citrus]

Perhaps, but with the EMP it's more achievable, you get 5 module slots, 50% free prod and processing units research

if you're using (still late game yes) in the EMP you only need 13 levels (~500k science packs) of prod research to reach 300%

[north yew]

ah just get many copies of one of the most expensive items in the game and 500k science packs ez

[jaunty citrus]

Plus it's something you can set up mid-game, and have it progressively get better over time

[jaunty citrus]

Nauvis will have lots of time to chug away at the research, and you don't need 300% for it to be optimal, getting to ~250% is way easier and gets your recycle loss ratio down to 1/8th per loop

The main benefit is that while expensive, it is the optimal late game strategy, but it is also something you can set up earlier and improve gradually

[north yew]

it only really helps with things that are downstream of recipes with infinite researches anyway though

and even then all of the ingredients that don't have infinite researches are just as expensive

[jaunty citrus]

No, because you can recycle backwards to keep the quality and get lower level ingredients. Using processing units as an example it means your is 16x as expensive and not ~80x as expensive

Assuming we do end up getting prod research that gets you for 4x the cost

[north yew]

yeah, that's fair

[rare thistle]

at the very least, blue chips and steel are both "confirmed" so the same ingredients are covered

[north yew]

LDS was listed in 's post too

[jaunty citrus]

afaik steel isn't a candidate for this strategy because you can't 'uncraft' it

So is the biggest sticking point, maybe we get prod?

[ember kernel]

Is prod research still a thing, I thought it went out of the window with RCUs and the foundry.

[north yew]

it's still only gonna be accessible in the lategame. deriving all of your production from recycled is gonna require an insane number of machines running and therefore an insane number of

[rare thistle]

but would recycling into steel get the prod bonus??

[north yew]

and I'm pretty sure that the ratio tanks heavily as you move away from the +300% prod

like very heavily

[jaunty citrus]

If your goal is to get everything then yes, but it it's to get some ingredients for equipment and modules, it's more doable

[jaunty citrus]

It does, yes. let me see if I can find the graph

[north yew]

it's still a thing, we saw a screenshot of prod after RCU was obliterated

[ember kernel]

Ah cool!

[north yew]

notable, though: the screenshot looks like a very lategame base and it's still only on level 7 of the research

[jaunty citrus]

https://media.discordapp.net/attachments/603392474458882065/1199258603236904960/image.png?ex=65f9429c&is=65e6cd9c&hm=8927eaef0d73a9493e8e5d61284118e9fd1befac946880e3c4b70a6168752543&=&format=webp&quality=lossless&width=639&height=482

From Dylan

[ember kernel]

Thanks for the talk guys. It's 3am, I either bust out the spread sheets or I got sleep. And I really can't afford not to sleep. 😄

[north yew]

listen there is infinite capacity for quality to keep you up at night you have to give up eventually

trust me

[north yew]

yeah that's very steep. even just 300 -> 250 is a huge jump

[rare thistle]

what are the units on the vertical axis?

[jaunty citrus]

the number of loops you get for your resources on average

[rare thistle]

lol

aight

[jaunty citrus]

so as you get close to 400% it approaches infinity

[north yew]

so going by this, even a legendary em plant with legendary prod modules is still only at +245%

235* b/c they haven't completed level 7

[rare thistle]

but like you only need to hit 16 to get the majority of items to legendary, right

[north yew]

so for that base deriving stuff from looped blue chips is still very expensive

[north yew]

not quite sure what you mean by this

[rare thistle]

at 25% upgrade chance and 4 tiers of quality to upgrade thru, you'd expect an average of 16 loops to get each item to legendary

roughly

[north yew]

☝️ 🤓 um actually it would be 24.8% chance

[rare thistle]

never gets old :P

[north yew]

I don't think that's quite right

b/c only 1/4 of the output is coming out of the recylers, and then only 1/4th of those outputs are getting upgraded

so you "miss" 75% of the time in the assemblers then you "miss" 93.75% of the time in the recyclers

[rare thistle]

but you have to account for this being at very high prod

[charred gyro]

The prod would have to be insane to offset -75% productivity per loop

[rare thistle]

you have a point tho, the falloff gets double counted in a way that really holds it back

[jaunty citrus]

In the background all module bonuses are integer values.

[north yew]

yeah like you absolutely need to hit +300% prod for it to not be lossy

and I'm pretty sure even jumping down to +290% starts to represent a substantial loss

4x output -> 1/4 recycler = 1

yeah it would

[jaunty citrus]

Technically they are because you can research beyond level 30

[north yew]

iirc there was discussion with the red names about that when this was revealed and it's basically not feasible to ever hit level 30 on any of the researches

[jaunty citrus]

Will it do anything? no. But you can do it

[north yew]

I think the numbers were that it would take a 1k spm base decades to get there. might be misremembering

[jaunty citrus]

Definitely misremembering

[north yew]

it was years or decades, either way an impractically long time

[rare thistle]

I'm still on team 375% will cut it. plus you'll have an extra 100% to workwith from modules and buildings, so you only need the research bonus to get to about 275

[north yew]

The most powerful force in the Universe is compound interest.

a smart man once said that

that man's name? Albert Einstein

with infinite science you're doing battle with the most powerful force in the universe

[jaunty citrus]

You'd never need to get to level 30 anyway, but it would be a fun challenge I think
You'd need to do ~575 million science of research

[rare thistle]

I love yall but I think I might be too irritable rn to get any deeper into the weeds. your math is very clever and i look forward to being utterly wrong :P

[jaunty citrus]

Which would take about 40 days of continuous 10k spm research

Given that the prod researches won't need every science pack and the crazy bonuses we get from quality and potential research productivity research I'm 100% convinced it could be done in under 2 weeks

[north yew]

that's like ~29k spm

even with crazy Space Age tech that's a lot

[jaunty citrus]

Bet. give me +50% in labs just with that cuts it down to 20k spm production
Using the scaling from EMP, Foundries, not needing to make all of the sciences and quality broadly? 100% doable.

[jaunty citrus]

Just one example. #friday-facts message

Base speed for EMP is 2 * 2.5 = 5
Each beacon gives 2 * 0.5 * 125% speed, each module in machine gives 125%
Total speed bonus for EMP is 1.25 * (12 beacon + 5 module) = 21.25 or 2125%
Total speed is (5 base + 5 * 21.25) = 111.25
Crafting time for is 10 seconds, with 111.25 speed 1 machine will complete 11.125 crafts /s, consuming 222.5 22.25 and 55.625 /s.
The total output will depend on your level of Productivity, but at base it will output 16.6875 /s and at most 44.5 /sec

The primary concern on how quickly you can do research is UPS, and Quality blows it out of the water compared to 1.1

[north yew]

a thought I just had: you'd be sacrificing two beacon slots, but you could still 10-beacon an em plant with direct to train

[rare thistle]

surely belt stacks are massive ups savings per item

[jaunty citrus]

A little under 4x savings I would assume? given not all of them would always be stacked 4x

[rare thistle]

if you're only using stack inserters they'd always stay full

[north yew]

em plant being 4x4 actually is perfect for direct to and from trains

[jaunty citrus]

True, and stack inserters have a small UPS bonus themselves in that they only move when full

[north yew]

load one train with the right ratio of green and red chips, load the other train with blue chips

have them both switch out at the same time

could still sneak 10 beacons in there I think

[jaunty citrus]

hmmm, I'm not so sure you could fit more than 2 beacons

[north yew]

train track is two tiles is it not

so you can beacon the other side of the track

[jaunty citrus]

• the inserter

[north yew]

oh yeah

damn I thought I was on to something

[jaunty citrus]

But you could probably get it to work well if you did filtered slots with both ingredients and products on the same train

Take the ingredients off the train then load the product directly back on

could still get 7(?) beacons

but then again you could just do that with normal assemblers

[north yew]

it could still be four beaconed right

the train tracks would be six tiles apart

and 4x is still +500%

[jaunty citrus]

Yeah 4 beacons true

With prod research you could get away with speed mods in the machines as well

[rare thistle]

hows this lol

[north yew]

not a bad suggestion tbh

in fact kinda brilliant

[jaunty citrus]

I like it

[rare thistle]

prob not enough speed with just 2 inserters but y'know it just might work

[north yew]

is pretty gosh darn fast

I've never needed to optimize around UPS before but generally direct to/from trains is very efficient right?

[rare thistle]

that's the theory

I've also never done it but I've enjoyed reading everone else's reports

[jaunty citrus]

Would it be ~122.375 i/s fast?
(22.25+222.5)/2

[north yew]

also bonus points for symmetry btw

[rare thistle]

thank you I try

[jaunty citrus]

You might need logic to make sure that you don't clog your chest with only one type of circuit

maybe you could avoid that by mixing together the filtered slots on the wagons

[north yew]

steel chest holds more than cargo wagon so if the input train is balanced and waits to be emptied it's fine

although there are potential throughput issues there that you might want logic for anyway

[jaunty citrus]

I would expect so, but I couldn't be too sure

[rare thistle]

lol. the system is fragile? that's upstream's problem

as a programmer I shudder to endorse this but it does technically solve the problem

[north yew]

perhaps it just makes more sense for the train to be output and belt(s) to be input

[rare thistle]

how much prod are we talking

[north yew]

all of it

each, any

[rare thistle]

if the number of inserters becomes a bottleneck then it matter which way round

but yeah lets not worry about that

[jaunty citrus]

^ this much

[rare thistle]

oh ok no big deal then

for an item as slow to craft as blue chips 👌

[jaunty citrus]

This is pushing the limits using only speed modules. A more reasonable setup would be much less and use prod modules.

But hey, when you can get to 300% without needing to use the modules, you run the calcs for it

[north yew]

you could even theoretically have 3 belts of input for very fast machines if we flipped it so it's outputting to train

very fast recipes, I mean

[rare thistle]

after a certain amount of speed isn't it higher items/second to use prod anyway?

[jaunty citrus]

The idea is you're using the processing units prod research to hit the prod cap anyway

Theoretically you can make 44.5 /s using a single max EM plant with 5x

with +300% prod from research

well, 250% from research and 50% from the machine

[north yew]

oh wait I've spotted a very small issue

where does the train stop and rail signal go

[jaunty citrus]

outside the beacon setup

[north yew]

yeah the stop would be outside but there's gonna be like a two tile gap between trains

potentially even THREE tiles

[jaunty citrus]

Two wagons go there, the loco and stop go further over

[north yew]

oh yeah that's much smarter than using a single wagon

[jaunty citrus]

Not sure how many times you could repeat it before the setups don't line up with the wagons

or if you could put loco's in the middle of the train to get fit more

[rare thistle]

not many

[jaunty citrus]

4 wagons isn't bad, you could stick another loco afterwards and repeat. (1-4-1-4... etc)

or just seperate the beacon setups a little every two

that would be more reasonable

[rare thistle]

ok new terrible idea

doesn't tile BUT very direct to train

[north yew]

the big bottleneck there is getting the old trains out to get a new one in

[rare thistle]

ramps do a pretty good job at that don't they

[jaunty citrus]

I think the bottleneck there is definitely the inserters 🤣

Does an inserter reach the back of the wagon if you put them in the other corners?

never tried it

[north yew]

pretty fast. 16 items per swing, presumably 5-6 swings per second(?)

[rare thistle]

they don't. its the worst possible length

[jaunty citrus]

I refer back to ~245 i/s for processing units (on the the most overkill setup)

[rare thistle]

I cry every time

[north yew]

what happened to the loco

[rare thistle]

1-1-1 train but with a loco in the middle so it can pull in all sneaky like

[jaunty citrus]

Sneakybeaky

[rare thistle]

you meant the bounding box didn't you. I was trying to put in some curves to see how quick the reload could be

seems alright

[north yew]

oh my

that's mildly cursed

also TIL you can have a non-loco be the wagon that stops at a train stop

[rare thistle]

glad I could TIL you

you lose acceleration doing it this way. something about aerodynamics blah blah blah, but its pretty cool

slightly improved

[north yew]

what if all three

[rare thistle]

I have created a monster

[jaunty citrus]

Oh my 😳

[heavy ember]

I wonder if there's some prod/quality/speed beacon mix that would minimize the amount of high quality prod/quality modules needed for given throughput

Sure more raw resources would be needed.

[jaunty citrus]

The best combination of quality modules and speed beacons is to not use speed beacons

[heavy ember]

Do we know the penalty amount?

Also I'm not talking about resource cost of making high quality item.

I'm talking about maximizing output, for a given quality/prod module count. Input be damned

[charred gyro]

Nah, it's illegal to use speed modules alongside quality modules. The biters will arrest you.

We don't know the penalty amount as far as I know.

[jaunty citrus]

There will probably be a number of speed modules you could use that would increase overal quality item production, but in practice I can hardly see it being worth it overall

[kindred crater]

they might just make speed modules -9999% quality (down to 0%)

[heavy ember]

+125% quality.
50% prod.
70% speed.

Add a beacon with speed modules in it and the cycle rate of that assembler triples

Say those two speed modules give a total penalty of 25% to quality.

That would mean around 2.5 more higher quality items out from the assembler per minute

[jaunty citrus]

Where are you getting +125% quality from? The best we can get is 5x = 31% in the EMP

[heavy ember]

From the fff.

Chance to get higher tier stuff would be 12,5%

[charred gyro]

The modules themselves so far seem to say they add +25% quality. But that is not the chance for them to add quality, but rather their quality strength.

[heavy ember]

The emp would read as 310% quality or so

[jaunty citrus]

Right, ngl that's pretty strange

https://cdn.factorio.com/assets/blog-sync/fff-399-quality-stacking.png

[charred gyro]

I imagine it won't be like that on release.

[heavy ember]

I do hope the speed caused penalty is low enough to give consideration.

[charred gyro]

That combination (speed+quality modules) is the only one where it would have any effect so they can balance it without affecting anything else.

[heavy ember]

And it would only really affect the q1-q2 cycle since that's where nearly all of the throughput is needed

Shotgun that with speed, then do proper high quality setups for getting q3 and better

[jaunty citrus]

The idea might have merit? But I don't personally see it being worth rolling more resources to get quality quicker when you could use a more efficient setup and use those extra resources to build out your factory

Unless the quality penalty on speed mods is very very low, which imo would be antithetical to the statements they made in the quality fff

It all gets very complicated when you want to get to the specific numbers on various different quality levels of different tiers of modules as well...

[charred gyro]

I'm wondering if they might introduce some other kind of module. Or even just module effects so that mods can add others that aren't just slightly adjusted rehashes of the current modules.

[jaunty citrus]

What other effect would you suggest?

Maybe a "quality effectiveness" module that increases the quality bonus an entity has

Get even longer range power poles

[charred gyro]

None of my ideas seem like they'd fit. Just hoping for more options. I'm wondering if we interact with enemies on other planets the same way we do on nauvis - via pollution. If not, perhaps whatever new mechanics are used there could contribute to a new module.

Or there could be modules which add other things which are less related to the machine's effects. Like... a module which adds a "power pole" electric field to the machine.

[jaunty citrus]

A module that boosts the effectiveness of other modules would be nutty

[north yew]

scrap module: each craft has a chance to produce scrap of one (or more) of its ingredients. the scrap can be reprocessed in a recycler

[charred gyro]

Heh, that sounds kinda fun. A different style of productivity module

[daring siren]

Productivity with more steps

[north yew]

idk there just isn't a whole lot of design space for modules outside of very wacky ideas

quality being one of those very wacky ideas

with prod, eff, and speed modules you have all of the bases covered in terms of numbers on the machine you can directly modify, so you're only left with modifying the output somehow

[jaunty citrus]

There are a couple of things you could do that would just be productivity but different

Like adding a chance not to consume resources

Same way they've added a chance not to consume an ore patch

[daring siren]

True. And that's a multiplicative effect

[jaunty citrus]

It's going to be pretty wild just how much we'll be able to stretch ore patches when the big miners only consume ore 17% of the time

[little orchid]

You can counteract the speed penalty of q modules without nerfing them, by using a quality machine

Four Q3 modules will give a 20% penalty, so even a machine would be enough to completely negate it

But then again even can't match the speed of 8 beacons, even beacons with only

So I guess it's more of an opportunity cost thing

[daring siren]

Also, we don't know the amount of qual reduction from speed. If it's a minor reduction, putting a single speed beacon can be still useful

[little orchid]

Right, that is important

And we don't know how it scales either

Is it a reduction based on the applied speed, or does every tier of speed module have a set quality reduction stat? And in the case of the latter, does quality increase the speed but not the debuff?

[daring siren]

Like how a single speed beacon makes a 4 prod machine 2.25x faster

It's simply a -qual parameter like -speed on prod

[little orchid]

Then it doesn't sound unfeasible to eventually reach a point where the larger number of rolls outweighs the reduced chances of each roll

With beacon setups or something

[daring siren]

With the emp, we get -25% speed. A single speed beacon will negate that. Depends also if the qual reduction from is different between speed tiers

[little orchid]

I would assume it is, continuing your prod comparison

[daring siren]

It might be better to have a high quality rather than etc

[little orchid]

This isn't so much about the speed penalty of the modules, but about losing the speed increase of beacons

So you would have a machine running at 1.05x speed instead of 5x or 7x speed

[daring siren]

Yes, we will surely not have rows of beacons

But negating the speed reduction is very useful still

[little orchid]

Well I already said this

[daring siren]

Not exactly

[little orchid]

Even better than a single beacon because there is 0 quality reduction

[daring siren]

Machine speed is multiplicative with speed effects

[little orchid]

I was just about to type that out

[daring siren]

So negating it with a beacon is worth a lot

Faster machines = less expensive modules.
Very important before endgame

[sick dock]

Speed vs prod mod are always going to depend on what you want. Speed saves space, and prod extends your resources. Obv, most players will end up mixing the two when they have the infrastructure ready. However, prod modules don't get less productive when used with speed modules, whereas quality does. You'll have to run simulations, but I'm dubious you'd want to mix speed & quality

but, then again, if you have massive piles of resources, and could not be bothered extending them, then maybe the space savings of speed will be worth it.

Obv before any real mathematical analysis, we're gonna need the raw numbers 😛

[daring siren]

Prod also increases speed of outputs

[sick dock]

yes

[daring siren]

Speed also saves on machines/expensive modules

[sick dock]

that it does. it does save you soem on the fixed cost of infrasturcture

and while EMPs are pog, those Q5T3 modules will never not be spendy

[heavy ember]

I'm not sure why we're so hellbent on 1% savings on ore.
A single mining prod level fixes that.
The maps are practically infinite with ore.
And then there's the buy in cost to the setup. Sure I might be getting less Quality stuff per cycle, but do I really care about that if the setup cost my third in resources? Especially when the high quality modules are stupid expensive already

Finally UPS reductions are worth considerations as well when scale gets large enough

[sick dock]

Speed directly negates quality, it is not like prod & speed

[north yew]

It is also notable that we created a quality penalty on speed modules, because haste makes waste, and we wanted to reduce the number of places where beacons full of speed modules is the best way to go.

[sick dock]

it's a question of how much quality power you're losing

[north yew]

phrasing suggests that speed + quality is simply not intended to me

[heavy ember]

Productivity directly negates speed

[sick dock]

yes, but productibity is the thing most people actually care about

it's the coveted asset

it's a question of your goals

[heavy ember]

beacons full of speed modules is the best way to go
This seems like a low beacon setup with highi quality speed modules is still viable

[north yew]

productivity stacks multiplicatively with speed, it just comes with a speed penalty. counteracting that penalty with speed modules increases the amount you get from the prod bonus.
quality + speed is the opposite. the more speed modules, the less you're getting out of the quality modules.

[heavy ember]

Maybe the DSP mentality of minimizing machine count has infected me from the max prod mentality factorio seems to have

[north yew]

in one case you have two different effects that boost eachother, in the other case you have two different effects that directly counteract eachother

[sick dock]

look, i don't think i've ever put a t3pm in a gear assembler. i believe its overrated i'm more aruging what is efficient

man, if Kovarex is right that there's more mechanics that are just as controversial, i'm freaking hyped :^)

[heavy ember]

https://cdn.discordapp.com/attachments/1215078107334057984/1216918592227119175/image0.png?ex=660222c1&is=65efadc1&hm=a50b585bbdf2cbd8c7b7ed8b7ab39da4e47248e43bb301c2cf3ca9bd152ceb5e&
This setup for mid game green chips is a good example, I think. Mixed speed/prod in GC assemblers is viable for me since it reduces the amount of them I need

[north yew]

yeah people are definitely too all or nothing w/ speed beacons + prod

is very good, limited beacon setups are very good

[sick dock]

i really ought to run the maths at some point. the payback period of a gear assembler that is fully decked is like..forever

[north yew]

it's one of the faster prod module payoffs

https://factoriocheatsheet.com/#productivity-module-payoffs

[sick dock]

you're getting 40% more (or 24% if we're counting in opportunity costs) off like 10 iron/s

jsut cause it makes so fast?

[north yew]

yeah payoff time has more to do with how many resources are flowing through the machine per second and less to do with the cost of the recipe

[daring siren]

That said, T3 modules are cheaper in SA because of EMP

[north yew]

that's why green chips are one of the best early targets

[heavy ember]

https://anntauri.shinyapps.io/Factorio/I prefer this one

[sick dock]

you right, i make simple falacy

the resource per craft will always be better off, ofc, but its the flowrate we're generally concerned with

[north yew]

so for example, is actually a pretty long payoff time b/c while each individual one is expensive, the crafting time is long so the flow through the machine is in the middle of the pack

[heavy ember]

1h 5m 38s assuming 8 beacon layout, infinite length columns and infinite amount of columns(average 8 machines affected by each beacon)

[sick dock]

this feels very counterintuitive, but you're right, lol

i play this game so long and yet still things to learn

[heavy ember]

17 minutes 40s if you only count iron

[neat shard]

One thing that's not too expensive, but likes productivity because resources flow through it rapidly, is sulfuric acid.

[cerulean harbor]

e (just joining)

[rare thistle]

you found the secret chat!

[tall sandal]

chat

[sacred totem]

Kwalutee

[iron root]

higher quality mods probably will mean you can get more items/s out of a machine with prod+speed than speed alone

cause of the multiplication

but yeah, prod being on the end of the chain is often just a quick heuristic for where it's best to put them
the exact solution is the machines that use the largest # of raw resources/s

[rare thistle]

that happens in vanilla too. at +550% speed the i/s increase of either module3 is the same. With 12 beacons its actually optimal i/s to have two prod mods and two speeds in the machine.

with legendary modules 4 prods is optimal i/s even with as few as 8 beacons

in an EMP with 12 beacons, 5 prods is optimal, with 8 beacons only 3 prods
in a foundry with 16 beacons obviously 4 prods is the play, same at 12 beacons, and again 3 prods at 8 beacons

(oh and foundries with 10 beacons is 4 prods)

[north yew]

are we talking purely in terms of items per second here?

[rare thistle]

Yeah

[unborn flax]

I've rolled some crude numbers. Suppose you have unlocked and but not . Then I think you only need to burn about 607 sets of ingredients on average in order to bootstrap your first 9 modules (5 for an EMP, 4 for a recycler).

This is a conservative upper-bound: to keep the state space small I've assumed we only get one quality bump at a time. In practice there will be skips, which will bring the number down even further.

The strategy is just going up through the tiers one at a time, repeatedly swapping in your higher-tier modules into the EMP first, then the recycler.

Once you have bootstrapped, producing a legendary in the EMP takes a mere 25.7 normal inputs.

[daring siren]

Remember that while working on for the quality loop, any mid level modules are still used, so it's not a waste

[unborn flax]

Sure - though to be clear, by 'burn' I mean lose most in a recycling loop.

The takeaway for me is that producing max-available-quality-tier kit with an EMP looks pretty doable, even pre-endgame.

[north yew]

(the one issue with that is that you don't unlock until endgame)

another potential factor: while the emp makes modules cheaper, they also have an additional ingredient now and we don't know how expensive superconductors are

[daring siren]

We kinda do know from the clip

[north yew]

fair point

superconductors seem to be , , and holmium plate

light oil's pricey b/c it cuts in to your precious ice supply, copper plate is pricey b/c it means recycling or a whole bunch of

potentially too, it's hard to say since none of the inserters actually swing on the superconductor plant

the more I think about it the more I start to wonder if maybe 's tech tree was designed around superconductor and supercapacitor manufacturing

[daring siren]

On a planet focused on electromagnetics? No way

The pack most definitely will need supercaps, and possibly

[north yew]

also electrolyte takes so wowee ice is a precious resource

[daring siren]

As precious as other resources

[unborn flax]

I chose and lack of to make things hard and show the worst case - so everything before that (mid-game) is even cheaper and more accessible by definition

[sacred totem]

not if shipping from space is free, yeah?

[north yew]

you'd still have to get it from the drop platform to your cracking in sufficient quantities

doable but not trivial

[unborn flax]

I'm moving from "we don't know for certain that EMP takes prod modules" to "I suspect EMP may not take prod modules". I know I'm on my own here and could well be completely wrong. Here is my reasoning:

1. A prod-themed machine (Foundry) fits with the character of Vulcanus, and a quality-themed machine (EMP) fits with the character of Fulgora.
2. I don't think it's correct to say the EMP will become obsolete in the late game if it doesn't take prod modules, for two reasons: i) like the Foundry it has unique recipes (and I guess the intersection of the EMP/Foundry's recipes is almost empty), ii) even without prod modules, the EMP is almost as good at producing legendaries as the Foundry due to the extra module slot (compare red line to the green/cyan lines).
3. Suppose you don't care about quality but you do care about productivity bonuses. If the EMP supports prod modules then once you have an EMP there's no reason to use an assembler3 for producing say green circuits. Whereas if the EMP doesn't support prod modules then the assembler3 still has a role, which feels like good design to me.

4. In JG's deleted video depicting an EMP spewing copper wire, the EMP had speed modules. However prod modules would have resulted in even more output/s. (ok this point is weak and circumstantial, but the video does show that EMP can hold its own in the endgame without prod modules)
5. If the EMP takes prod modules then the optimal strat for legendaries would be only one quality module, which means that even on Fulgora, quality modules won't get much of an outing in the endgame. Whereas if the EMP doesn't take prod modules then the optimal (and only) strat would be filling it with your best five quality modules - meaning they'd get some proper airtime.
6. If the EMP does support prod modules then it would make the number of inputs required for a legendary extremely small in the late game (see the first point of the blue line), so producing legendary modules may become a formality earlier than it should.

Here's a picture of how infinite productivity research affects the number of inputs required per legendary.

The leftmost yellow point is the 79.9 we know and love. The line "EMP without prod modules + asm3" refers to the fact that if the EMP doesn't support prod modules but a recipe does (e.g. copper wire) then for legendary inputs, you're slightly better off using an assembler3+ filled with prod3s than an EMP with no modules.

[heavy ember]

On 3. Why would I use electric furnace when I have the foundry on the recipes it supports?

[unborn flax]

Well there's no real overlap of recipes right? https://cdn.factorio.com/assets/blog-sync/fff-397-foundry.png though in any case I'm not saying that you should, I'm saying it's satisfying that 'older' buildings still have a role late game

I think 2) and 6) are the factual points not widely known; the rest is me saying "one way feels like better game design than the other", which is a lot more subjective

[north yew]

1. it's blue so if anything it's themed
2. there's still a huge range of recipes that don't fit the foundry, em plant, or speculated use of the building, leaving space for the assembler
3. that was clearly just a quick and sloppy demo to show it going fast
   2 + 5 + 6. you don't want to put quality modules in intermediate production trust me

we know a lot more about the foundry and its recipes, and it's similarly powerful. it's a huge force multiplier for your ore

[unborn flax]

Could you put some argument behind 'trust me'? Just because you don't want to keep 5 stacks of plates spinning it doesn't mean others don't fancy the challenge

[north yew]

#1215078107334057984 message

it's a dead horse I've thoroughly beaten tl;dr it's just not gonna work out as well as you want it to for a lot of reasons

[unborn flax]

I don't think it's dead - I think it's one of those places where perfect is the enemy of good. I wrote this down when first thinking about this:

Thought experiment: Suppose there is 1000-step process that produces "polished steel" from "steel", "polished polished steel" from "polished steel", ..., "polished^{1000} steel from polished^{999} steel", everything with ratio 1:1. At some point in the line your ingredients will probably get bumped to legendary and then stay there for the rest of the process. So you should have to do essentially no recycling in order to obtain 100% legendary polished^{1000} steel.

point being that to any extent you can get quality intermediates to work, it is useful

[final nimbus]

Legendaries becoming less prohibitively expensive in the post-endgame isn't unreasonable.

Making it not accept prod would put the Electromagnetic Plant in a weird position where it's temporarily better than for circuit manufacture, so you remodel half of your factory to incorporate it -
And then later you remodel the same part of your factory to use AM3 again.

[charred gyro]

If it's balanced so that one way is the best way, it probably isn't balanced as well as it could be. So regardless of how it is now, I would expect either option to be viable upon release.

[unborn flax]

Maybe I'm masochistic but I like the choice in this - do you choose the 'better for now but will have to tear up later' solution or the 'better for later' solution?

[final nimbus]

It wouldn't take that long to make AM3 better, 4 x is enough.

[north yew]

the problem is that Factorio isn't a game where it's just item 1 -> item 2 -> item 3 and so on and so forth. easy example:

• presumably, and are coming out of the furnace at the same quality ratio (ie the same proportion of common to uncommon to rare etc)
• we roll quality again while turning in to
• now has a different proportion of quality to
• when making , its constituent ingredients have different quality ratios
• we can only use the worst ratio of its constituents, so the roll for quality on was completely wasted

this is one specific example, but understand that this problem repeats itself over and over and over across the whole supply chain

if you throw a bunch of quality modules in your intermediate chain, a good 80-90% of that quality rolling goes completely wasted

if the immediate thought for a solution to this is "ah, well I'll just avoid putting quality modules in specific places where it gets wasted like that" then what you're going to end up doing is just putting no quality modules in the intermediate chain