Quality

[burnt reef]

it should just be a mod then probably

[grand igloo]

Absolutely the fuck not. Colorblind people exist and should not be ignored when designing graphics.

and are all very clear for the colorblind while being hardly more intrusive than a single dot wherever they are used.

[half raptor]

Colors and patterns are easy to translate at a glance how good they are. Thats one reason games like Diablo put fancy borders around their higher quality items because sometimes an orange colored name and a red colored name might look similar. Granted Blizzard does a deeper shade of red for ancients so even color blind it's not so bad

Game developers who take color blindness into account usually have fewer customer complaints

[grand igloo]

Mhm

If you make an effort to support the disabled, then it turns out everyone actually benefits

For example, I’ve been playing Rabbit and Steel lately, and it has color palette (for bosses and such) and visibility settings (for general transparency) for accessibility. I use the hell out of them even though I’m not colorblind because it makes the screen easier to parse

[uncut helm]

Too bad inserters belts and logi chests doesnt

[grand igloo]

Yeah. Luckily (or unluckily?) my father has never gotten far enough in to use bots

[frank warren]

How would you redesign them if you were going to?

[uncut helm]

For belts only the icons would need to change as the entities has different animation speed. I already made icons differ in my colourblind cure mod

As for inserters I have no idea

They would need probably complete redesign so their shape is different

[frank warren]

IE every inserter having a bit of a different shape?

To be fair, doesn't have an issue :)

[uncut helm]

This one is a little bit different yeah

[obsidian crescent]

You probably wouldn't need to go that far. Regular/Fast can probably use one shape, with contrasting colors to tell the difference. And Bulk/Stack can use a different shape.

The logistics chests are a bit more complicated. Even without any color vision deficiency, the difference between requesters/buffers, and active/passive providers is hard to see at a glance. Giving one of the two pairs distinctive shapes would really help. Requesters and active providers could look one way while buffers/passive/storage would look another.

[uncut helm]

The whole point of this discussion is that color is not enough difference for colourblind people.

[fallen urchin]

I think they can have arrows painted on the chest

[obsidian crescent]

Even people who have complete color vision deficiency and see entirely in black-and-white can distinguish between darker and lighter objects. The key is to find colors that are high contrast even if you can't see the colors.

[obsidian crescent]

Is that going to be noticeable enough at a distance though?

[unborn flax]

re: #friday-facts message , for machines you could make the status/bottleneck indicator have the shape of the quality level. Also colourblind-friendly.

[fallen urchin]

I'm not sure whether it's needed to have noticeable difference at a distance

[frank warren]

that's crazy

[daring siren]

Different meanings for the same color. I'm against

works because it's color and shape

Speaking of graphical changes, I kinda wish pipes showed fewer fluid symbols

This is super cluttered

Fluids show up on every intersection and every 2 segments when it's a straight

Would be nice if it showed between segments... like between each 2 , and in the middle of the 4 pipes for

[gusty trench]

Definitely want a way to remove quality indicators without removing all alt mode information

It's not even that bad here, since there are no inserters

[chrome mauve]

meantioned that it would be nice if there was an option for showing quality before but it was more of an map option rather than part of ALT mode

[rotund egret]

it depends on what colour pairs there are.

for inserters, the only ones that give trouble are regular vs bulk

strictly colour-wise, making one canary yellow and the other dark orange would make differentiating between them a non-issue

telling colours apart is much easier when one is lighter and the other is darker

[full coyote]

i think the trims would be cool. colourblind accessibility, sure, but the alternative (not doing that) is strictly equally or less readable than trims no matter how good your colour vision is, surely?

[frank warren]

I think trims look really bad

[kindred crater]

if someone wants to mod it by all means, but yeah the icons look way better IMO

[iron root]

im an absolute sucker for trims but the icons are probably a bit less ambiguous

and having both.... ehhh?

[uncut helm]

I'm not colorblind and I don't trust people saying those 2 colors are causing problems and other set of colors doesn't. It may sound ignorant of me but I would rather make sure each entity looks distinguishable when i preview them grayscale then rather do scientific research about how to get away with it, getting a testing group of colorblind people to give me feedback that i may or may not understand because im not colorblind. Its kind of like telling a blind person on how to improve their drawing

[frank warren]

would rather make sure each entity looks distinguishable when i preview them grayscale
I would like to at least say here that I believe their point is that the two would look different grayscale, because the colors are so differently bright.

[uncut helm]

thats true but im not sure if thats good

[jaunty citrus]

Yeah it's the contrast that matters, which grayscale gives you a decent enough approximation of.

[frank warren]

I think I see what you mean though, you'd prefer they have different 'shapes'

[jaunty citrus]

I still want to make the bulk inserter graphics just an incredibly muscular arm.

[frank warren]

I <3 the technology icons

[jaunty citrus]

They're fantastic!

[daring siren]

Not bulk anymore. They switched the names, unfortunately

[jaunty citrus]

Yeah the muscular arm belongs to the hulk 💪

[rotund egret]

about 99.7% of all colourblind people have red-green colourblindness (protanopia et co + deuteranopia et co).

due to how the cone cells work, there's actually a lot of common ground between the two. whether you don't see red or you don't see green, you have trouble with the same type of pairs, just for different reasons.

here's a list of the confusion colours for red-green colourblind folk:

• cyan and grey
• rose-pink and grey
• blue and purple
• yellow and neon green
• red, green, orange, brown

while making everything distinct on greyscale is very sweet, it's also much more difficult than working with what colour pairs are distinct.

for the two troublesome inserters what could help is adding a metallic line (kinda like hydraulics) down the main arm of the bulk inserter, similar to how its claw is metallic

behold, my l33t paint skills

[daring siren]

have bigger hands

[uncut helm]

sure you have a list of confusion colours for red-green colourblind folk. I still have no way to test how it looks for colorblind people and I never will

i just know some colors that we call some names without any strict definition of rgb values causes colorblind people to see differently resulting in them not being able to distinguish those

[dense kraken]

even if they have the same rgb value, different monitors may do different things

[uncut helm]

i wonder how many colorblind people calibrate their monitors

[dense kraken]

wouldn't surprise me a lot of them decalibrate them

[opal haven]

is taking the screenshots and removing all red color component, while adding like 1/3 of red to green (for example) gonna emulate color blindness to some degree?

[rotund egret]

it depends:

99.7% of all colourblind people fall under these categories:

• protanopia (you don't see red at all)

• protanomaly (you see some shades of red)

• deuteranopia (you don't see green at all)

• deuteranomaly (you see some shades of green) --> this is by far the most common type

the issue is, most people with colourblindness can still see most things in colour. simulations usually account for "worst-case" scenarios, aka the two -opias

[opal haven]

thanks, I think I got it
if majority of colorblind people can see blue at least ok - maybe the ||crude and simple|| way to test will be to remove all colors but blue and see if all similarly shaped entities are different in their blue channel

[dense kraken]

what about the 0.3% that can't see shit? :/

[rotund egret]

nah, those 0.3% mostly have blue-yellow colourblindness (tritanopia), which is really weird and rare

[dense kraken]

well there is achromatopsia

[rotund egret]

complete colour blindness (black-white) is really really really rare, and generally comes with other issues (albinism, photophobia etc)

[opal haven]

well, you can run removing colors trick 2 more times to check other combinations - it's not gonna cause all inserters to be only on black and white spectrum, but it's totally gonna complicate stuff quite a bit

[opal haven]

I tried to test blue channel trick
it's noticeably messing up combination of 3 colors ||due to me being bad at drawing ||
but differences in blue channel between inserters might just do the trick

blue channels only for comparison

[winter saffron]

diffent color brightness converted to grayscale is a trap. It heavily depends on the background noise and average brightness on how much we (humans) perceive brightness

and donÄt get me started on different monitors (settings) too

[grand igloo]

at more distant zoom levels this is not going to be easily visible

[heavy ember]

Could have it always visible and the current quality icon be visible under alt mode.

Similar to beacons and modules

[grand igloo]

Well:
Bulk inserters already look different, and I think stack inserters also have a different hand type.
Burner inserters could maybe get a little chimney or exhaust pipe.
Long inserters are obviously long.
I think it’s just the fast vs. regular that could be an issue, and I think that could be fixed by adding dark stripes… or maybe additional feet…? to fast inserters

and filter inserters don’t exist anymore

[grand igloo]

tbh how I’d fix it would just be to add color settings for certain things tbh

[uncut helm]

there are settings like that for colorblind people and all of them i talked say it doesn't help at all or make it even worse

[grand igloo]

oh I don’t mean the fixed palette thingy

I mean a straight up color picker

[jaunty citrus]

The only good colourblind setting is one that just lets you pick the colours.

[shadow vine]

would be nice if there was an option to color the pipes themselves according to whats in it

colored pipes and/or the symbols

[daring siren]

That's what the window is for, but it's limited

[shadow vine]

yeah it gets a little hard for me to see whats in the pipes at times

maybe theres a mod for this, to color pipes according to whats in them, ill check tonight

[half raptor]

Wouldn't be too hard considering they already animate (windows) based on what's in it so when rendering the pipe it is checking what is inside

[grizzled lagoon]

It might already have been answered somewhere, but I'm a bit confused and can't find it immediately. What is the relation between the probability to get 2 quality tiers higher p(q+2) than the current tier q relative to the probability to get 1 tier higher (p(q+1))? Is it p(q+2) = p(q+1)/10, or p(q+2) = p(q+1)^2? Since I'm trying to make a generic spreadsheet calculator that computes how many of a certain tier you'll get given certain modules and which tiers you recycle. I'm currently using the p(q+2) = p(q+1)/10 formula (as that's how I understood it), but can't reproduce the 56x more expensive number from the FFF. I have 94,4x more expensive (using 2 legendary prod3 and 2 legendary quality3 in assembler and 4 legendary quality 3 in recycler). I added my current spreadsheet in attachment. Feel free to play around with it.

[daring siren]

The 56x is a myth

Unless you consider stuff like the EMP with 5 modules slots and built-in +50% prod

The probability to get 1 level higher is the % of the modules i.e. 4 giving 25% then you'll get 75% of the input level. Then roll a 10 sided die. On 1-9 you keep the level, on 10 you get one higher. Roll the die again until you lose.

It's a bit more complex than that, because with quality over > 100% you can get some nice bonuses

You can look at it as an exponential function with k = 0.1, and the starting point at "qual%"

[grizzled lagoon]

Ok then I think my calculator should be correct. Have there been any other ones that I can use to crossreference?

[daring siren]

Yes, you can test how it looks in the Janky Quality mod. I've calculated stuff there with Factory Planner and it seems to fit the calculations we had in these threads.

This is the function for the array of probabilities

function libq.make_probabilities(effective_quality, max_quality)
    if max_quality <= 1 then
        return { 1.0 }
    end
    local factor = 10
    local probabilities = { 1.0, effective_quality }
    for i = 3, max_quality do
        probabilities[i] = probabilities[i - 1] / factor
    end
    for i = 1, max_quality do
        probabilities[i] = math.min(probabilities[i], 1.0)
    end
    for i = 1, max_quality - 1 do
        probabilities[i] = probabilities[i] - probabilities[i + 1]
    end
    return probabilities
end

[iron root]

even the devs couldnt reproduce the 56x under any reasonable configuration

[daring siren]

With EMP we got 30x~ IIRC?

[iron root]

we've found configurations with much less, and also much more iirc, but never anything right around 56

[daring siren]

There was some crazy I don't fully remember

[iron root]

but quality cycling interacts in very interesting ways with productivity that will make the number go around like crazy

and eventually down to just 1x at 300% prod (which actually appears reasonably achievable in some scenarios, notably EMP with maxed out modules)

[grizzled lagoon]

If I use an assembling machine, with legendary T3 modules (2 quality, 2 prod in assembler, 4 in recycler), I get about 10.6 legendary outputs out of 1000 ingredients for a 1 ingredient --> 1 output recipe. so 95x more expensive than normal. Does this sound roughly correct? Same thing in foundry gives me 34.6 outputs (28.9x)

[daring siren]

Yes

Sec I'll try to find pictures I did

[grizzled lagoon]

Great. Then I think I'm ready to publish the calculator. Let me know if there is anything you'd like to change (or feel free to change it yourself)

Edit: forgot to put default values back to something reasonable

[iron root]

95x isnt the farthest thing from what i found from a quick spreadsheet run

[grizzled lagoon]

What do you mean?

[iron root]

i ran some calcs in excel a while back to figure out how 'expensive' legendary stuff was in a given module setup

and it was somewhat close to 95
was definitely closer to 95x than it was to 56x

[daring siren]

#friday-facts message
#friday-facts message

[grizzled lagoon]

Yeah I rounded. It is 94.4x after 15 cycles

[daring siren]

Check these two posts

(it's doubled because it's to , so just divide by 2)

[grizzled lagoon]

Hmmm you get different numbers...

I sadly almost got to go. There might be some mistakes somewhere.

[daring siren]

It's 3 different combinations, and again, it's doubled

[iron root]

none of them map to 95x

[daring siren]

First just just , second is with in assemblers. Third is 2:2

They are all vetted by others

The smallest is 79.5x which is nice, but not 56x 😄

[iron root]

yeah, and then when other sources of prod get thrown in, we wind up skipping cleanly past 56x lol

[keen igloo]

in janky quality sometimes i've got a uncommon and a rare at the same time, which i think is just a consequence of 1.1

[grizzled lagoon]

Yeah I'm comparing to the last one. Get very different numbers. I'll look into it. Ping me if you find a mistake (no obligation to look at it)

[daring siren]

Not a problem for statistics

[grizzled lagoon]

Does janky quality take into account that you only get 0.9*<quality percentage> of the next tier instead of just the quality percentage?

Except for legendary tier

[daring siren]

Yes

Statistically, if you do 1000 crafts, you should get a result very similar to what 2.0 will give

[grizzled lagoon]

Not sure where it’s deviating. I get the same trivial answers (no recycling). I’ll check closely for errors in the recycling code.

[daring siren]

Did you include in your recyclers?

[grizzled lagoon]

Naturally

But I noticed some minor material loss while recycling. Don’t get exactly 25% of the inputs back. So there must be a mistake somewhere

(Used Excel for accessibility for non-programmers. Remember why I never use it)

[daring siren]

Don't you mean Google Sheets? So it's online and easily accessible?

[grizzled lagoon]

Good point. I’ll copy paste it to sheets once I’m done

Maybe I also should replace the percentages with the type, quality and amount. More user friendly

[half raptor]

I'm not going to math quality too hard, I like trains grid bases so I'll probably have a grid block dedicated to each quality of each intermediate and probably build multiple blocks for each normal quality. With train interrupts it'll deliver a mixed output to each block and sort it out there. Any excesses (destination full and 5 sec inactivity) the train will go to a recycle block and unload that material for upgrades. Then if I ever overload on legendary intermediates you can substitute a higher quality intermediate as a lower so each of my normal to epic blocks will have an overload station to accept excess legendaries. If that fills, then that means I wasn't paying attention to my base

Tell my bots to spread to all machines and then upgrade it to then to as available so the base will change its math as we go.

[grizzled lagoon]

Do you recycle everything? Since I have redone everything and get exactly the same numbers as before, which are different than yours (the slight error I knew was there, did not have an effect on the final result). How confident are you about your numbers?

[daring siren]

Very. It's been calculated several times by different people in different methods.

[frank warren]

I can corroborate soul-burn's numbers.

Both of my programs come to the same numbers as everyone else here.

[half raptor]

a lot of math relies on going from normal to legendary in one craft step. if you try for quality on every step of production and recycle excesses, you can have 6-8 crafting steps to reliably try for quality on. You only have to get lucky 4 times and some of those times you might go up 2 steps

[obsidian crescent]

Production in those circumstances becomes increasingly more complicated. Getting more bites at the apple does reduce resource consumption, but quality cycling a 3-step process is hard.

Now, there are times when it could be worthwhile. Once Fulgora is running at full tilt, you're going to be awash in red and blue circuits. Using them to do quality cycle module production across multiple stages will net you higher quality modules and higher quality circuits (from lower quality recycled modules).

[half raptor]

By setting up a production block for each quality for each intermediate, upgrades will accumulate through the system which will flow much more freely as modules get upgraded and we increase production of normal blocks. When Destination is full for 5 seconds, it'll recycle the excess and hopefully return a few upgrades as well as another chance to upgrade.

As for wasted materials, I am going to set up alert speakers that alert when 1) I have more than half full ingredient buffers 2) I have less than half buffers for the products and 3) use a to sort out the signal with the strongest demand. I'll have a speaker for each quality on the rare chance that it's time to increase my or better production blocks, and another set of speakers for raw ingredients. As long as I'm paying attention to my signals, I should only have to recycle the few rare exceptions that would otherwise unbalance the system

[grizzled lagoon]

Ok I finally found it. There was no error, I stupidly forgot to use 4 prod modules instead of 2 prod 2 qual for the last step

[grizzled lagoon]

Hi all
I finished the calculator. I tested it quite extensively and it seems to work without issues. I reproduced the values that were previously shared, amongst others. However, before sharing it outside of this forum, I wanted to post it here first to see if there is any feedback/if you notice any bugs.

Here is the link. You need to make a local copy if you wish to edit the file (for obvious reasons)
https://docs.google.com/spreadsheets/d/1GpszMksTmaaIV4opnKTx_e7JrrBPRaSp/copy

If you wish to see a preview (read only) https://docs.google.com/spreadsheets/d/1GpszMksTmaaIV4opnKTx_e7JrrBPRaSp/edit?usp=sharing&ouid=106564669231106274883&rtpof=true&sd=true

There is 1 feature I know might be useful: choosing whether or not you wish to skip the first assembly from raw inputs to outputs (i.e. are your inputs ingredients or a finished output of normal quality?). But I was afraid it would make the tool less clear without adding much. So currently it is left out.

I'm also wondering, would anyone ever wish to specify modules per quality level that is not legendary quality? I could add that option for just a little bit of extra work. But not sure if there is a good use case for it

[teal wren]

This is very nice file! I've done my calculator some time ago, but it is simulation based and not math formula... but the real reason, I do not know why I need it. It was fun to test what was the most resource efficient way to produce things, but as soon as it is obvious, all alse just a matter of buildign a factory and not thinking much

[grizzled lagoon]

Yeah I also just did it for theorycrafting. Yes the optimal configurations for all best modules are already well known for a long long time. But my reasoning was that 1) before reaching all the best stuff it is not obvious what the best modules are and 2) just like a factory planner it gives you a nice idea of the sheer amount of inputs required to make 1000 legendary things. Especially if you also want to keep some, but not all lower quality stuff.

[teal wren]

In my playtrhoughs I use only efficiency modules up until I get to craft tier 3 at mass scale

Because that + is damn too expensive on electricity and pollution

So I have to make more walls, and more ammo production

[grizzled lagoon]

Well we will go to 4 different planets before having access to everything. So I do expect this to change in SA.

[teal wren]

I have to make them in a first place that is

[teal wren]

Hopefully, because right now for me the only real option before the endgame is efficiency

I just love to let bugs sleep until I start launching rockets every minute and have spidertron army, and they just hit 30% evolution

[grizzled lagoon]

I do the inverse: only prod until rocket, then prod+speed (and efficiency for mods, never for vanilla)

I also mostly wrote the file for other people who wish to embed it in mods/... but are not aware of the mathematics of how to quickly compute infinite repetitions. I've already noticed in some mods that their approach is pretty slow 😄

Damn just noticed the dropdown menu colours are messed up when you make a local copy of the file... that's a pity

[teal wren]

Waiting for "quality calculator" mod which will show the characteristics of selected factories for production

[grizzled lagoon]

Not a modder and no interest to get started in modding. So not for me 😉

[opal haven]

why efficiency modules are such a high priority? flamers will roast any unwanted guests no problem, no matter how high the pollution is
and will pay off in half an hour without any - they are amazingly busted

[teal wren]

I have to build the perimeter in the first place... and I am lazy rushing to the rocket

I just want to take some ammo from mil science and place couple of turrets which will serve me till end of midgame

Just before this is max + 12% per machine, but after +40% and you can have even faster speed

[opal haven]

understandable, although SA will probably mix stuff quite a bit - leaving planet for hours without somewhat reliable defences might end up badly.
so, for me, plan will be something like -> -> -> -> ->

[teal wren]

You mean endgame? Because rocket now is quite early i think

[opal haven]

yep, rockets are blue sci now
plans I mentioned are for first 2-5 h of the game, before first rocket launch
after that - nauvis base must be capable on surviving on its own, without my help

[teal wren]

Huh... this is astute, you are in fact leaving the planet... I need to think about many things, thank you

[half raptor]

As long as you got construction bots, you can remotely "build" your base

[fallen urchin]

Deathworld might get a special strategy: Abandon nauvis, and set up production in a biter-free haven like Fulgora and Vulcanus

Might not needed in normal deathworld, but if someone plays expensive DW or scarce ore DW then this may be worth considering

[obsidian crescent]

Fulgora might be free of enemies, but Vulcanus almost certainly isn't.

[fallen urchin]

Fair point.

There're 2 other reasons against this strat: 1. Space science, which require uranium, is nauvis-specfic AFAIK. 2. Although rocket is a lot earlier, it's still blue science, which means you should already have flamers at this point.

I saw this strat in a SE + rampant playthrough

[grand igloo]

you can get it below 56x

through the power of productivity cheese

[frank warren]

@daring siren do you recall the source of this mystical but wrong 56x number?

[burnt reef]

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

With this straightforward approach, if you want to produce items of legendary quality, and you already have enough legendary quality 3 modules (which is not an easy thing to get in the first place), the legendary items are 56 times more expensive than normal items.

[tall sandal]

its funny because they dont even have the 80x setup which is the lowest anyone could get to

[half raptor]

Devs develop, players optimize

[daring siren]

Except for the white spots

[half raptor]

Lightning is kind of the enemy of Fulgora

If you don't build proper lightning "defenses" your buildings get damaged

[jaunty citrus]

I think there will be enemies in the white spots

[burnt reef]

I think the enemies will have white spots

[daring siren]

I think white spots are enemies

[keen igloo]

i thought the white spots were scrap deposits

[obsidian crescent]

We know that there are "rich" scrap deposits on those islands, but there's something else on top of them. Possibly something that prevents us from making use of them until it is dealt with.

[kindred crater]

like whats been said before, could possibly be enemy factories that manufacture enemy units from the scrap

[fallen urchin]

Precursor warfactories that were not properly shut off

[grand igloo]

these numbers are inaccurate, btw.

community's mathed the fuck out of this, and it's ~80x (I think it's close to like 76? but whatever)

assuming no prod manipulation jank

prod manipulation can bring that down by a lot though

by using prod to preferentially preserve higher quality outputs as best as possible

[daring siren]

79.5x

[kindred crater]

is this just legendary quality modules in AS3s and recyclers?

[obsidian crescent]

Probably. Once you start getting 50% prod buildings involved, lots of things change.

[half raptor]

I hope we get AS4 that blows this math out the water

Though I suppose you can already say that about the EM plant

[obsidian crescent]

I really hope not. EMPs already eat away at a large number of the utility of assemblers. Indeed, rocket fuel is basically the only bulk product assemblers uniquely make that we know of. And if there's a 50% prod chem plant on Aquilo or whatever, then the only thing assemblers really do is make infrastructure and early science packs.

We should still want to have assemblers. They're good.

[half raptor]

Also a lot of the math is how to go from normal to legendary in one craft step. I'm still in favor in one of my save files to try for quality on each step and just have a factory block for every intermediate/quality combination and to recycle excesses until the base is near balanced.

In another of my save files I'm going to the logical thing and prod everything until the final step

Quality seems fun, production seems smart

[frank warren]

quality spoilables seems very hard

[half raptor]

Probably not going for quality spoilables. Or if I do, the non quality products will go toward science and the quality will go toward any quality buildings we can make (recipes still unclear)

[obsidian crescent]

... yes. But even then, it really depends on how long the spoil time is. If you can find a 30-minute spoil time intermediate, you can probably quality cycle on it quickly enough, and then recycle some of those down.

But quality spoilage is probably going to rely a lot more on multi-stage quality boosting than quality cycling. Fortunately, a lot of the things you make with spoilables are not themselves spoilable.

[half raptor]

Until I see the gleba recipes, it's hard to argue

We are going off a few images with no context on what the gleba science even affects

[obsidian crescent]

The other thing is that speed modules clash with quality modules. So Gleba's natural tendency towards faster and faster production is in tension with quality. You may need to use higher-quality biochambers for quality production.

[half raptor]

This planet might be the anti quality planet

[obsidian crescent]

Which is really irritating, because it's the planet where plastic literally grows on trees. And getting quality plastic otherwise is kind of annoying, due to having to get quality coal (either via mining or quality cycling grenades or something).

[half raptor]

I love how quality was the third FFF for this expansion and it's still baffling my base designs

[daring siren]

It's 2 2 in and 4 in ♻️

[kindred crater]

thanks

I don't have the numbers for foundry, but here's another good bit of info for EMP: #friday-facts message

as low as 13.2x with 4 and 1

and yeah, that base prod makes all of the difference, especially for non intermediates

[daring siren]

It's interesting that prod is still king, even when the goal is quality

[kindred crater]

before seeing the numbers I would've guessed going with 100% quality modules. just goes to show the power behind offsetting the recycling consumption cost

[obsidian crescent]

The goal is getting quality for the lowest cost. And prods are all about lowering costs.

[gusty trench]

I'd have expected a higher ratio of quality to prod in the 50% prod machines. Doubling the quality percentage doesn't beat out lowering the prod by 1/6th

[half raptor]

If you get close to the 300% cap, you make 4 items per craft and recycling divides it back down to 1, and recycling gives a chance for quality ingredients. By the time youve done it enough times to get legendary ingredients, that puts you at 4 legendary products for the normal ingredients to make 1

Researching prod bonus is going to be major

It'll probably be interesting when you get a high enough researched prod bonus that you can start replacing prod for quality and still be at the 300 % cap

[kindred crater]

quality processing units and LDS will be good sources of their quality ingredients then

[half raptor]

Yes they will be

[obsidian crescent]

LDS is interesting in two ways.

1. The Foundry's LDS recipe no longer uses plates; it uses molten metals directly. So cycling isn't really viable with its 50% prod bonus. So you're looking at LDS prod 20 before you hit the break-even point.

2. The Foundry's LDS recipe's only solid input is plastic. So if you can make quality plastic, you can benefit from that greatly.

[kindred crater]

https://cdn.factorio.com/assets/blog-sync/fff-387-foundries.mp4 did they change the recipe from #387?

[obsidian crescent]

Yes.

Quality cycling spoilable materials may be weird, but it could be the most resource efficient way to get quality plastic, since you don't have to quality cycle grenades or something.

[kindred crater]

huh.

[frank warren]

#2 is a big deal, becaues you can also recycle the copper and steel out of it!

[kindred crater]

I don't understand what you mean on #1. does it still recycle into its 3 base recipes even for the foundry alternate recipe (steel/copper/plastic)? it is an intermediate so why couldn't you just add whatever combination of quality/productivity to the machine?

what am I missing here

[frank warren]

You can't really 'cycle', it's more of a 'convert plastic into materials'

[obsidian crescent]

Recycling LDS returns plates and plastic. In order to do quality cycling, you need to be able to feed the outputs of the recycler back into the machine. You cannot do that if the machine was a Foundry because it doesn't take plates.

[frank warren]

you can cycle the plastic here, but you'd need to do something with the plates it spits out, which will be of various qualities.

[obsidian crescent]

You can quality cycle into an assembler, with a Foundry feeding this process, but that's all.

[frank warren]

There's no way to melt them back into the foundry.

That said, if you're going to Q-cycle plastic, you an do it here and then also Q-cycle something that uses the steel and copper as inputs.

Assuming you get the ratios to work out. Molten metal is pretty free.

[kindred crater]

ah I see, after the first craft you're basically just using recyclers and AS3 then, losing that base prod

[obsidian crescent]

In fact, one thing you could do is just make plastic on Gleba with qual modules (productivity is less important as the inputs are free) in the Biochamber, then feed higher-tier plastic into Foundries with prod and qual modules, then recycle those things to get the higher quality plates out of it.

[frank warren]

That is what I was saying :)

There's also no reason not to be cycling on Gleba

anything unused should go to a massive bank of quality recyclers.. why not?

[obsidian crescent]

You mean recycling plastic into itself?

[kindred crater]

so all you need is plastic to make LDS essentially?

that makes me feel better about this

thats even simpler than doing it the vulcanus cycling route

[daring siren]

It basically makes very cheap on because grows on trees

[kindred crater]

might as well throw stupid amounts of volume at it to get the quality plastic you need then

even if you don't have the highest quality modules to use in the process

[frank warren]

I'm not sure plastic is even cheaper on Gleba.

Nauvis oil is extremely strong. Gleba would need to be seriously powerful to beat it. Especially with new fluids.

[half raptor]

Well we do still know very little about Gleba. The only thing that got shown off were the bio gatherers and about spoilage

I feel like we know way more about Vulcanus than gleba

[obsidian crescent]

It's really a matter of convenience. Plastic on Nauvis costs coal, which does run out eventually. But on Gleba, plastic production is a matter of a fixed land area. As far as we know, it never "runs out".

[kindred crater]

personally when it comes to quality, I will always choose the methods with the smallest material cost, regardless of the setups or logistics required

[half raptor]

I still don't remember anything about plastic on gleba

Must have missed it

[frank warren]

‘runs out’ has always been a nonsensical definition

[obsidian crescent]

It was the first picture of the Biochamber. It's made from red mash and green cube.

[frank warren]

'red mash and green cube' 😂

I love how we have no idea what this stuff is called

It's better than before when were like "yeah so there's the white spheres and they're related to the green blobs, and you can craft some of them by hand"

It sounded absolutely deranged

[iron root]

Green jello

[frozen fiber]

What is nonsensical about having a word for resources whose patches can run completely dry?

[uncut helm]

🏃🏾‍♂️

[frank warren]

Because in practice, the difference between a resource that can and can't run out is extremely minimal.

It sounds like a real concern, but in practice there's no real convenience difference.

In SA you're going to tap a set number of patches and I expect they will last for 1000s of hours.

Basically: If it lasts longer than you play, it is effectively inexhaustible.

[jaunty citrus]

It definitely matters for a time, just not for very long.

[frank warren]

Yeah, there's a point in the game where it does, but that was not the point these discussions were being made at.

Remember: the subject was why use gleba plastic on nauvis if we just have oil on nauvis

Their reasoning was 'doesn't use coal', which if you follow this idea that coal is not infinite, sounds reasonable

[frozen fiber]

It's just nice that you can set it up and never have to worry about it ever again. That matters to me

[frank warren]

but for all practical purposes, it is.

[frank warren]

I would argue this is mostly in your head, there is a point where they are the same thing. I know it doesn't feel that way, but it still is that way.

[frozen fiber]

Whatever

[frank warren]

For all practical purposes, if it takes you 10k hours to mine up a coal patch, it's as infinite as gleba plants.

It's just not a strong reasoning at all. It has to be something else.

[frozen fiber]

Still, there's no harm in having a word for it, is there?

[frank warren]

There's a bit of harm, it's pretty misleading, because it makes it sound like you'll need to replace patches every 20-50 hours in space age, and like it is actually a meaningful difference.

"A distinction without a (meaningful) difference"

[frozen fiber]

Let's see how big all those patches will turn out to be

[frank warren]

It's not entirely about the patch size, but about our ability to research much deeper, as well as our quality and big miners.

Big miners have lower consumption. We will have loads more mining prod, and quality miners decrease the consumption even more.

[jaunty citrus]

I think wanting to save coal on Nauvis is quite suspect, it's the least utilised resource in my experience, mainly because I never use liquefaction.

[frozen fiber]

In ultra late game yes

[frank warren]

I was about to say, 'coal is the worst example here'

Well, this is a world where you are converting your nauvis plastic production to use gleba plants

you're not exactly early game anymore

[frozen fiber]

I'm still going to keep using that word

[frank warren]

I mean, what do you want me to do about it, arrest you?

I don't think it's wrong to make a distinction. It's what I think is interesting about factorio's different resource types.

[frozen fiber]

Not call it nonsensical I guess?

[jaunty citrus]

🙍‍♂️🔫👮‍♂️

You're under arrest

[frank warren]

It's still nonsensical under this context. I can't help that very much.

[frozen fiber]

You did ask

[frank warren]

I think it's possible for this to be a thing that matters, for what it is worth.

[frozen fiber]

But anyway, I think all points have been made.

[frank warren]

But I think 'gleba plastic is useful because you don't need coal' is questionable at best.

That's why I speculate it has to be something else

So the argument comes in - what do we get from Gleba? I think we can agree it's not 'plastic but without coal'

To be fair, maybe the biochamber makes plastic really really fast.

[frozen fiber]

I'm aware that in late game patches will last for ages btw you're right of course

[frank warren]

I wonder how right I am tbh - mining speed is increasing a ton too.

[jaunty citrus]

I think there is valid but overemphasized concern over the quantity of ore available. (just expand if you have concerns)
Depending on the crafting chain for Gleba plastic, it might be far easier to scale, making it the location of choice for plastic.

[frozen fiber]

I guess we'll find out

[frank warren]

The crafting chain for nauvis plastic is just crude oil + coal + (optionally) water

with fluids 2.0, assuming raiguard doesn't ruin these plans, we can scale oil almost arbitrarily.

[jaunty citrus]

Yeah I do find it hard to believe that it is easier than Nauvis plastic, but there might be room to get rid of excess plastic on Gleba by quality cycling it

[frank warren]

that's roughly my plan on gleba

[jaunty citrus]

Keep the factory running rather than letting intermediaries spoil

[frank warren]

I want to just recycle anything that is going to spoil yes

[frozen fiber]

I sure hope it's not just plastic. That would be pathetic

[high zenith]

The janky quality test play throughs I saw got way better results when they put quality modules in the miners and diverted quality ore to the multi stage upgrade and normal ore to the normal factory

[frank warren]

Yeah, especially if it is - as I was pointing out - plastic that is much slower than nauvis but 'free' because it doesn't need coal.

[jaunty citrus]

There's definitely way more the Gleba than they let on. Unfortunately I'm not clairvoyant so we can only speculate with what we know.

[daring siren]

There's more on every planet

[frank warren]

Yeah, which is 🤯

[jaunty citrus]

Yeah but it feels disproportionate for Gleba.

[high zenith]

It feels even more disproportionate for the 5th planet. That's just because of the schedule

[hoary current]

we do know that plastics take the yellow mash and green gel which are probably one step processing from each fruit type

we know yellow mash is from one step processing of the red fruit

[uncut helm]

yuo know too much

[jaunty citrus]

Arrest them for knowing too much!

[frank warren]

@hoary current How confident are you that green gel is just the other fruit's mash?

[hoary current]

90%

[frank warren]

hmm. That's about my estimate too

[jaunty citrus]

My guess is the round green thing used to make the science packs is "Bioflux"

Wish we had an image of the other tree being harvested

[hoary current]

theres also this thing

[daring siren]

The bell pepper?

[jaunty citrus]

bell bepper, orange slime or weird crystal?

[daring siren]

semisolid fuel

[hoary current]

yellow jelly

[daring siren]

IT'S GREEN

[jaunty citrus]

The icon makes it look semitransparent, the same as the green cube

[daring siren]

It's not chartreuse this time

[jaunty citrus]

Nah it's yellow-red this time not yellow-green

[hoary current]

maybe I’m wrong about the green gel bing the mash of the other fruit. Perhaps the red fruit also has a gel product made from mash

which is the unknown thing on the belt

[uncut helm]

i like to see people analysing my icons upclose

[daring siren]

We're talking about different items

[hoary current]

so the processing chain is
fruit -> mash -> gel

[daring siren]

The bell pepper looks like a semitranparent crystal. The green jelly is green jelly.

[hoary current]

and u combine yellow mash and green gel to plastics

[hoary current]

nah the green jelly is solid fuel made from lube

[daring siren]

solid lube?

[fallen urchin]

I think the main difference is not mining coal, but the lack of long distance logistics

No need to ship in coal from mines, just produce it locally

We already have mining productivity research, will we have mining quality research that increases quality of mined ores?

The main problem with mining quality is, the Foundry will turn Fe/Cu ore into fluid and eliminate all quality involved in it

[frozen fiber]

Only melt the non quality ore

[daring siren]

Is it worth it compared to recycling the "more" items later?

4 quality modules on a BMD is nice though

[frozen fiber]

I don't think the quality ore will be more than 10%, so you don't "lose" that much

[uncut helm]

qualitium

[daring siren]

It gets to 25% with

[frozen fiber]

Sure, but that would be a LOT of very expensive modules

[daring siren]

BMDs are big

It can be worth on

[frozen fiber]

I don't think we're going to need 25% of all ore output to be quality

[daring siren]

On planets where you mine ores, it's better to and smelt in foundries

[frozen fiber]

Or get it to 25% and only keep the 2.5% of rare and above

[hoary current]

tbh the complexity becomes crazy the second u start doing quality intermediates

[half raptor]

I know there is going to be a mathematically perfect answer out there, but I do enjoy the fact that the other strategies people will go with will also be pretty darn good and I mourn the loss of creative solutions the moment a YouTuber shows the "correct" way to do things

Player will find a 5% more efficient method and nuke their hard built original designs in favor of a Nilaus blueprint

[daring siren]

Nilaus BPs are never very efficient, but they are solid and are efficient in what matters to Nilaus

[frozen fiber]

The fact that the best, legendary builds will only be possible in the very late game, should really reduce this

[heavy ember]

ROI of unbeaconed quality setups seems way too long for me. Atleast with legendary modules

[daring siren]

Beacons are going to be used much more, considering the rework

[hoary current]

they’re never efficient in what matters to me

he tend to have BP with a specific base size pre allocated

[daring siren]

You say Nilaus, you mean 100x100 road city blocks

[hoary current]

well yes

[daring siren]

Personally I don't like city blocks, but I kinda like his block style, and will make them so much better

I did that style once here https://storage.googleapis.com/soulburn-mapshot/mapshot/IR2/index.html?x=33.0&y=177.6&z=2.0
It's my most organized base by far, even if each block is a mess 😄

[hoary current]

I just think hes jump start base is too small and the bus base is too big

[daring siren]

There's something nice with having spaghetti contained in blocks

[hoary current]

and its a pain watching him manually building the bus base in 100x100 city block until bots

[daring siren]

Oh yea no lets not do that

[hoary current]

and he almost always do that

[daring siren]

If you notice, my base has just 1 belt or pipe per material

[hoary current]

I always spaghetti upto bots

[daring siren]

Also, IR2 gives you in

Personal, slow, but still bots

[hoary current]

personal bots yeah

quite good for scaling modular BPs

[daring siren]

I like a tight starter base for vanilla, which gets bots quite early... 4-6 hours

[half raptor]

I have a janky base that is one half belt of iron with the other half copper, at the start of the line are 4 assemblers for gears and 2 for (3 ) so that my 2nd belt is circuits and gears. Then it's a simple mall of red science machines and 2 machines making

This janky red base takes a while to fill up on belts and inserters but it's a wonderful red science base

It is better than trying to handcraft the start of my bus base

[heavy ember]

We really should make a bot% a speedrun category. It would actually be useful

[half raptor]

Time is measured in BB and AB "before bots and after bots"

[grand igloo]

the module setup for 'optimal' quality farming is complicated

doubly so if you want to use the prod exploit

[grand igloo]

I'm probably going to go heavy on quality breeders, and exploit the ever-loving hell out of the prod interaction on wildcard recipes.

Prod manipulation is ridiculously strong for quality breeding

I need to check the math but I think it makes recycler-based strategies viable off the bat

[winter saffron]

wildcard recipes?

[grand igloo]

Recipes without a quality requirement set

Essentially, because you aren't changing the recipe, while using wildcard recipes, productivity bonus never resets.

So you can track (prod% modulo (100%)) and favor the highest tier recipe available when that value will reach 100% due to a recipe.

[burnt reef]

iirc bonus products from productivity are always

[grand igloo]

Nope!

They're the quality of the running recipe

[burnt reef]

huh. do you have a source?

b/c that seems disproportionately rewarding for how much micromanagement it requires

[grand igloo]

it's buried far, it'll take a while to find

[grand igloo]

It's a lot more than it seems

on the micromanagement side

for manual crafting it's not bad, but automating it is fairly atrocious

[burnt reef]

no, I mean "very rewarding and very micro-heavy"

[grand igloo]

you can automate it

[burnt reef]

yeah I have a few ideaa on how already

[grand igloo]

#space-age message

the logic of it is fairly straightforward:
ex. you have 25% prod

0 -> 25%: use lowest tier, as normal
25% -> 50%: use lowest tier, as normal
50% -> 75%: use lowest tier, as normal
75% -> 100%: you're going to get a prod output, use it to slam highest rarity possible

though you also need to account for resource levels

so it's more like:

0% -> 25%: use highest tier you have more than 1 recipe of inputs for
25% -> 50%: use highest tier you have more than 1 recipe of inputs for
50% -> 75%: use highest tier you have more than 1 recipe of inputs for
75% -> 100%: you're going to get a prod output, use it to slam highest rarity you have at least 1 recipe of inputs for

[grand igloo]

but it'll be very circuit-intensive per assembler, overall

at some point you'll probably have enough surplus resources and high enough prod that you just stop doing this exploit

bc it stops being worth it

the gains from this method diminish strongly as you get more and more prod

well, independent of modules

because this method actually lets you have a higher tier increase chance per craft

which saves a LOT of resources, counterintuitively

[winter saffron]

One can force-set quality requirements for recipes? e.g. an assembler set as takes only inputs?

[kindred crater]

that was my understanding, you can either select a specific quality or

[daring siren]

For the current set recipe yea

[kindred crater]

but for any, for a given ingredient, they all must be the same quality

[winter saffron]

hmmm. And the machine set as such doesn't require quality modules?

[daring siren]

Of course not

Quality modules are just for a chance of higher quality outputs

[grand igloo]

Yes, you can. Or you can leave it mixed

[winter saffron]

Hmmm hmmm. I feel like there's a way to enable Braxbro's idea via splitters and the least amount of circuits required. It'd be space intensive tho

[grand igloo]

if you use mixed, it uses the minimum quality of all item ingredients (not fluids) to determine process quality

[grand igloo]

The important thing about my idea is that you're stocking in recipes

and need the ability to know how many recipes you have of a given quality ready to go.

Doing that through splitter logic is possible, but horrendous

iirc splitters are turing complete

[winter saffron]

whait what

how

why

[grand igloo]

Because belts can loop back into their own inputs

and you can create AND/XOR through priority input setups

Is it practical? Fuck no

[winter saffron]

ohh god

[grand igloo]

I think the only time I've ever used splitter logic is a very light dose of very simple logic (not simple because implementation lmao) in... I think it was IR3?

It's really not practical.

But yes, splitters and belts are theoretically turing complete

you'll just hate yourself if you try to build a computer with them

[iron root]

turing complete...
without any circuit wires yeah?

[grand igloo]

Yes.

[winter saffron]

essentially transistor level

[iron root]

i mean
a splitter can divide by 2 (unprioritized 1-2), add (including remainders), subtract..
oh god

[winter saffron]

I don't know what to do with this knowledge

[grand igloo]

to count the outputs you'd need to use circuit wires

[grand igloo]

Nothing. Don't do anything with it, for the sake of your sanity.

But anyway

circuit control of this sort of thing would be tedious to set up, but would massively improve quality yield

especially early on

[obsidian crescent]

Circuit control of this sort is not at all difficult to set up, so long as you build it around very specific productivity numbers. For example, if you can get a gear maker building to exactly 50% prod, it's very trivial to do: just alternate inserters that each have a hand limit of 2 plates.

The EMP making copper cables is also quite doable without any prod modules.

But as you get into even the complexity of green circuits, that's when serious inserter controls are needed.

[winter saffron]

I'd just set handsize to 1. This makes it slower but removes a lot of the hassle.

Will the recipe variants have their own prod. research? E.g. and

[obsidian crescent]

I certainly hope not. Because it would also mean that the Vulcanus LDS and steel recipes wouldn't benefit from prod research.

[grand igloo]

You can generalize it pretty easily

it's just tedious to set up

not difficult

you also need circuits, not just inserters, because you don't want to mix quality inputs in the mixed quality machine

(i know, sounds weird)

[winter saffron]

Theoretically you can setup a regular line of assemblers, synch them via circuits, to profit of this. So you have your normal outputs, which can be any qual and/or prod, and every time it reaches 100% you change recipe, and enable legendary input. Reset.

The non-legendary inputs still in the assemblers will be dumped into the assembler's inventory when the change happens. These can be fed back into the line again

The synching & stopping part being the difficulty here.

The legendary input can be achieved via belt weaving or bots

[grand igloo]

Nope, because the logic is more complex than that

there are some things you can parallelize, yes

but I suspect that fully parallizing is not feasible

[winter saffron]

hmmm

Which step would be breaking it?

[grand igloo]

Because the logic is this:

For each assembler, you need to be able to buffer up to 2 recipes of each tier (ideally more, to handle the variance of recycler outputs)

If the assembler isn't going to trigger prod, it'll run the highest quality it has 2 crafts stored for. If it is, it'll run the highest quality it has 1 craft stored for.

It's that 1 and 2 craft buffer that's throwing a wrench in things

Sure, you could centrally compute it

and run the entire thing in parallel, but that would probably make the logic much more complicated to implement

and the logistics less reliable

[winter saffron]

For each assembler, you need to be able to buffer up to 2 recipes of each tier
But this isn't a hard requirement? (Also I'm ignoring recycling for now)

[grand igloo]

Yes it is.

Because you need to have some means of measuring "okay I'm multiplying all the high tier stuff I can, I need to process everything else without multiplication"

if you just alternate highest lowest, you'll have a backlog in the middle

You need to basically go high, then highest w/ surplus

and that surplus needs to be in whole recipe units

so you need 1 of each (in case you can't get a complete set for the one after that to do prod with) plus 1 extra, so you don't need to give up that complete set to run non-multiplied recipes

[winter saffron]

My implementation was strictly for a single product. With any and legendary outputs. I can ensure it doesn't break via a timer which resets the recipes if
e.g. electric circuits and bots and final craft (prod)```

1. I ensure all requesters have at least 3 copper cables and 1 iron of legendary quality.
   If not I reset all recipes to basic or any
2. Only the exact amount of items required will be input into each assembler, so that none may be wasted in the reset craft.
3. reset```

[grand igloo]

with 'any' you waste steps

[winter saffron]

However I can add a decreasing loop:

    if quality = 0 -> reset
    else use :quality:-1
reset

[grand igloo]

you never want to feed mixed anything in

ever

[winter saffron]

I'm not: The requesters have 5 sets of inputs for each quality. I iterate down to find the largest possible quality available and double it in the prod. step

This is apart from a feeding belt for regular crafts with static quality

If you still see a step where I'm wasting quality products you need to tell me precisely where. I don't see it.

[grand igloo]

in your baseline, no?

feeding 'any' in is not sufficient

you want to feed specifically the highest you have at least 1 set worth of excess for

and again, yeah you can centrally calculate and manage this

but wouldn't that be overly complex then

[winter saffron]

you want to feed specifically the highest you have at least 1 set worth of excess for
This is lazily ensured via the decreasing loop. (lazily since it only considers the single product)

The very baseline would be belts of input (or whatever lowest setting one may want). The assemblers should be modded with quality (as to not overcomplicate the prod step calculation). The quality of regular outputs don't matter for me right now since I'm not considering recycling. It also shouldn't starve since the prod step modding only happens if 1) it's possible and 2) if it's possible it uses the highest worth of quality I have excess for.

[winter saffron]

Also any or basic is trivial here. Since the machines are not filled with a higher quality than required at any point. If any machine in the line can't be filled with the same level of quality, ALL machines will reduce their recipe level by one. Repeat. If no possible recipe can be found it will be set to any or basic and will grab from the static feeding belt. NOT the quality set requester chests

[grand igloo]

eh ig

though then you have to worry about synchronization

[winter saffron]

^

lul

Yes, it's expected to run slower than regular assemblers

So I add more assemblers

One can extend this line for as long as the input belt can be a constant feeding source.

[grand igloo]

I mean, it's not horrible to calculate out by circuit, assuming each quality of item can be a different signal

[winter saffron]

I thought so?

[grand igloo]

so you could just calculate out by circuit, set the requesters all at once, then once everything's done, hit go

[winter saffron]

jup

[grand igloo]

you could even modify the algorithm to allow you to handle an entire cycle at once, so you don't have to worry about dead time

[winter saffron]

and if it breaks it just chucks along normally via the reset function

The Selector combinator should make the indexing quite simple

[grand igloo]

Mhm

[winter saffron]

I mean you can change this to make each machines recipe non-dependend on other machines but I suspect this being more circuit intensive than centralizing it and letting it run slightly slower short term

[grand igloo]

You don't need to decentralize to let it run at full speed

you just need to reshuffle the algorithm

so you're calculating the entirety of a cycle in parallel

you can pretty easily calculate how many non-multiplied crafts you have to allocate before your next multiply craft

right

So, in order for it to be safe to use a quality for a craft, you need to have that many recipes buffered, plus one

[winter saffron]

Which is trivial for e.g. 10, 20, 50 prod

[grand igloo]

It's trivial for any prod

It's just a counter with some extra steps

that's advanced when the input buffers for each machines empty

[winter saffron]

I'm lazy

that's advanced when the input buffers for each machines empty
My idea is that it'd just stop everything if the input buffer (primary belt) is not satisfying all machines

[grand igloo]

But if you look at my method, it's absolutely trivial

[winter saffron]

Because if it breaks there some machines will be at different prod level and the whole algo breaks down

[grand igloo]

Well, that's the risk of centralizing.

[winter saffron]

But it heavily reduces circuit calculations

[grand igloo]

Not really.

The operations themselves are fairly simple

granted I've played turing complete so

maybe 'simple' for me is different than it is for you

[winter saffron]

fair

Hmm, I guess I can decentralize it. Make an unique quality-reduction loop and satisfaction check for each assembler.

I havent used circuits enough to know how UPS expensive they are if used in excess

[grand igloo]

But how I might set this up, logically:

X = how many cycles left until the next partial prod hit (if you have over 100% prod)

X + 1 = how many recipes needed to ensure 1 machine can run a recipe of a given tier constantly, from now to the next cycle

Integer divide each buffer's amount (in recipes) by X + 1. That's how many machines can be assigned to a given tier for this craft step.

If the number of assignable machines is greater than or equal to the number of the machines you have: Go down your line of assemblers, filling an input buffer w/ a single recipe of the highest assignable recipe, then decrement that recipe for the next one.

Once all the input buffers are full of the requested items, dump into assemblers and start loading the next batch.

Obviously, you also want a latch to prevent the requests from being changed until they're dumped to the assemblers.

[grand igloo]

once you get to the point where you're concerned about UPS you'll probably be eating the hit to efficiency

and going for a less efficient, less gimmicky setup

[fallen urchin]

Float imprecision will probably make it more complicated

For a machine with 25% prod, it might produce the additional thing at 4th craft, or 5th

[winter saffron]

That's why it's preferable not to use prod modules but only speed or quality

[obsidian crescent]

I highly doubt the productivity bars or productivity percentages are stored as floats.

[fallen urchin]

I think it's float, because I usually handfeed assemblers in warptorio. They have like 8% prod, I feed them 50 ingredients, and they produce 53 outputs with an almost full prod bar

For this situation, I think you can manually insert materials, to make it craft once at 10% prod, so it's 10%-> 35%->60%->85%->110%, this will exclude the effects of float imprecision

[frozen fiber]

I tested it today and if the crafting time isn't an integral number of ticks, crafting and productivity bar do desinc

But if its an integer, they don't, not for tens of thousand of crafts anyway

[obsidian crescent]

Yeah, that makes sense. The problem is, getting anything to a precise integer tick count, especially with speed beacons, is going to be pretty difficult.

No assember has crafting speed 1.00.

[frozen fiber]

There are some options

2 prod 2's and 2 prod 1's in the assembler combined with 3 times the effect of speed module 2's works

The resulting crafting speed is 2

[winter saffron]

Devs, please give us circuit control to readout float prod so we can abuse a glitch huehehhe

[frank warren]

you can make that yourself

[frozen fiber]

In SA, anything other than 1, 4 or 9 beacons will screw it up because you're dividing by a square root and the result won't be integral

Or 2.0 rather, since the beacon changes will come to the base game as well, right?

But if you don't have quality it doesn't matter anyway

[gusty trench]

Or 16

[wary beacon]

Lookin forward to this

[obsidian crescent]

Wait: that wire reach requires +2 reach per quality level, not +1.

Maybe it's +1 in each direction.

[jaunty citrus]

It's +1 radius

https://cdn.factorio.com/assets/blog-sync/fff-397-quality.png

[golden tinsel]

legendary substations are almost like current long rage poles, insane

[grand igloo]

you won’t speed beacon quality breeders. Speed modules have penalties to quality

[obsidian crescent]

Speed modules have penalties for improving quality. We're talking about using prod bonuses to give you effectively 100% productivity for quality inputs to equal quality outputs. That is, you have a lot of Q4 iron plate and Q4 copper cables; how do you feed your EMP making green circuits such that the productivity output is always Q4 without 100% productivity on the machine.

Speed modules don't matter for that.

[grand igloo]

Easily. You just count the recipes.

You don’t need tick timers

Nor do you need 100% prod

sure some floating point fuckery could complicate things but you don’t need 100% prod

[frozen fiber]

If you had 100% prod, there would be no need to think about it

[grand igloo]

Sure, for legendary mats to legendary quality; if you have N * 100% prod then you can just have a dedicated assembler

but you should be filtering out that stuff anyway

since you can't tier it up, qual mods in that machine are a waste

[grand igloo]

this setup is best for the tiers below it

[grand igloo]

Oh also, the "just feed max-quality parts into a separate assembler with max prod" is an optimization for a very specific special case

where the maximum amount of prod you can fit into an assembler is equal to or greater than prod% of the mixed-quality assembler integer divided by 100, multiplied by 100 again, plus 100%. (ex. 25% / 100 -> 0% + 100% - > 100%)

which should be true with modules alone

but might not be once you deal with prod research?

basically, these prod manipulated machines are like two machines in one that take turns

one with (prod% - (prod% mod 100))% productivity, and one with 100% more productivity than that

if you can achieve that 100% more number with modules, you're better off using a dedicated legendary assembler

if you can't

then you're better off just using the manipulation method

I believe

it's weird

[frozen fiber]

You've lost me

[frank warren]

I get it. What parts don’t you understand exactly?

[fallen urchin]

Nevermind, it's about manipulating the prod output

[kindred crater]

i wonder if you can make

presumably higher capacity?

[daring siren]

You can make which is faster. There's no effect of rocket parts or rockets.

[kindred crater]

ah

makes sense

[grand igloo]

what parts need re-explaining

[grand igloo]

would be neat if the rocket gained more cargo weight or something from that

[obsidian crescent]

Unless physical land area is a concern, there'd be no point. Getting a 2x boost in rocket capacity is exactly the same as having 2 silos. But quality goods generally cost more than 2x as much as base quality goods. So it would always be better to just have more lower quality rockets.

[frank warren]

Technically there are things you can launch with a 2x silo that you cannot with a 2 1x silos.

[daring siren]

If it was a thing

It's possible to mod in different silos with different rockets

[grand igloo]

it’d probably be a 6x boost or something

but yeah

[hoary current]

quality science packs are not worth the price, so does quality rocket. remember u can use prod modules instead of quality moduels

[frank warren]

It depends, some sciences have no ability to use prod well

(their ingredients)

giving up speed here is the problem for those

[hoary current]

not able to use speed beacons is a big issue if u want quality output

and if u want the quality item elsewhere u better not use them for science

say electric furnaces

u probably want to save the quality ones for space platform instead of feeding science

also this by , seems like quality is not that vital in SA

most people probably won’t bother micromanaging quality for science

the bonus u get is just too small due to how small the quality upgrade chance is

[frank warren]

and if u want the quality item elsewhere u better not use them for science
not withstanding here; the goal is to discuss whether quality science is viable. That doesn't care about your other needs for these items, as it assumes your base is 'built'

I agree that that tells us quality science is not optimal at any point before final endgame though.

Quality science is provably more resource efficient for

But it's - I would say roughly speaking - almost certainly worse for UPS by a huge margin. and dramatically more complex for little yield.

[grizzled lagoon]

How can it be more resource efficient for some, but not all sciences? You always get 1+bonus science for the price of 1. The bonus only depends on the used modules. Where is the difference?

[charred gyro]

I imagine the number of intermediate parts used for creating the science will determine how many stages of productivity modules can be used, and the total effect changes depending on how many stages prod can be used in.

[grizzled lagoon]

Ok but the amount of highest tier intermediates required before science will still only depend on the modules in the science assembler, no? So you take the module arrangement that minimises the demand. What am I missing?

[obsidian crescent]

The components of most science packs are intermediates, which are prod-able. Since the productivity bonus is bigger (resource cost-wise) than the quality bonus, it is more resource-efficient to use prods on them. However, purple and green science take non-proddable materials. As such, you could put quality modules in them and take some of those outputs to make quality purple science packs.

[grizzled lagoon]

Oh quality in the science ingredients, got it. Thought you meant quality in the science assembler itself.

[hoary current]

quality in the science assembler is always less productive than prod in the science assemblers

[obsidian crescent]

Now, you absolutely should put quality modules in the prod1 and electric furnace assemblers (and probably inserters). But you should do that so that your infrastructure can benefit from higher quality prod 1s and electric furnaces, not so that you can make quality science packs.

[gusty trench]

Gray too

[grizzled lagoon]

Yeah I think I’ll put quality in everything early game and filter away the components with better quality

[daring siren]

There is still a reason to produce science packs, and that is a sink for unbalanced intermediates in some cases

But even then, you'd probably want in your assemblers rather than

[hoary current]

relatively weak reasons. there’s many simpler ways to sink items. u may not get a lot by making quality science using excess items to worth the complexity

[daring siren]

science is a constant sink

[obsidian crescent]

There are always better sinks for such things. Excess quality stone can be turned into brick and later concrete. Excess quality iron can become gears. Excess quality copper can become circuits. Excess quality coal can become plastic and later red circuits.

[iron root]

Making not- science packs with excess quality materials is a neat prospect but it still consumes things in its own defined ratio, which is going to need another overflow into science for that mismatch
It will let you squeeze out a little more science though

[grizzled lagoon]

Can’t imagine that the added complexity is ever worth it. I do think I’ll send low quality expensive intermediates back to science instead of recycling them though. But I don’t think I’ll do it such that the quality isn’t lost. (Eg just fill a trains of blue circuits of quality < 3 to the first thing asking for blue circuits)

[half raptor]

How's everyone doing on this day?

[rotund egret]

40°C outside, walking a friend through the factorio tutorial. poor lad, he's going to be in deep soon

[half raptor]

Here, let's just work through to don't worry you won't get addicted

[grand igloo]

it's a monday

in the summer

it's if that

[hoary current]

[teal wren]

When in your car kicks in:

[grizzled lagoon]

The Cube finally has a worthy opponent

[fallen urchin]

Wonder whether we can get quality ultracubes

[obsidian crescent]

So, quality fuel is actually useful. Maybe that's a good reason not to ship U-235 everywhere; you'll be able to get the benefits of nuclear fuel's acceleration without getting regular shipments.

I think this is of particular use for Gleba, where fuel grows on trees.

[daring siren]

If quality fuel is a thing

[fallen urchin]

According to the Kovarex pic above, it is a thing

[half raptor]

More like you should find places in your base that are sufficient with your lower quality uranium and places that would greatly benefit from quality uranium.

Just realized that recycling might let us sink a ton of u238 into a recycle loop and then kovarex any legendaries that make it through to make legendary u235

If that's possible

[daring siren]

There's no in but I don't see why it won't let you set quality transform u238 to u235

Getting 40 u235 would be tough though

[half raptor]

Did they confirm there's no quality for uranium ore?

[half raptor]

Yes it will be

[daring siren]

No, but they confirmed there's no quality modules in Kovarex Enrichment Process, which is what I wrote with icons.

[half raptor]

So you won't be able to go for quality on that step, but what I'm curious about is doing kovarex with uranium that's already quality

[daring siren]

I don't see a reason why it won't be allowed, so it probably is

i.e. you could 40 u235 + 5 u238 into 41 u235 + 2 u238

[half raptor]

I hate to imagine how much u235 we will have to recycle to get our starter 40 legendaries, problem for future me

[daring siren]

Getting the u238 is easy though, because of the amount and recycling spent power cells

[half raptor]

True enough

I'll have to have a strong reason to go for legendary uranium, seems like it's not very sustainable for trains until you get enough production. Late game problem it sounds like

[obsidian crescent]

Wait, can't you just quality cycle UFCs to get quality U-235/8?

And if you just need quality U-238 (to feed your Kovarex once you get 40 U-235), just quality cycle uranium bullets. Though UFCs will probably be more efficient since you can prod them.

[keen igloo]

quality mining of uranium will be helpful, you get 3 slots there instead of 2, do that on even 2 uranium patches and pretty soon you'll have so much quality radiation you'll be glowing at least at a uncommon level

[obsidian crescent]

Quality nuclear fuel isn't worth that much effort; it's much easier to just quality-cycle UFCs.

Then again, there's the question of what higher quality UFCs do. Kovarex talked about fuel with regard to how they affect trains, but not with regard to how they affect burner device output.

So if quality UFCs have a higher fuel value, that could be a worthwhile reason to invest more heavily in quality uranium sources. Especially if you're exporting them off-world.

[gusty trench]

Fulgora seems like the place to make quality fuel: solid fuel is effectively totally free, so you can quality cycle it even if you don't get enough from scrap recycling, and then add some light oil for quality rocket fuel

[obsidian crescent]

We've seen solid fuel on Gleba, so I was thinking that it might be more efficient to make quality solid fuel from fruits. Not resource efficient, but more time-efficient: you'll be able to make more of it faster for the same land area than Fulgora. Not only would Gleba benefit more from faster fuels (thanks to spoilage), but you might be able to take some of the quality intermediates and make something else out of them.

If you've been to Vulcanus already, then you don't really need to skim quality off of mining uranium; you can just recycle for it.

[opal haven]

there's infinite amount of heavy oil on fulgora - turn it to a solid fuel directly, and add some cracked to light on top to make rocket fuel

[obsidian crescent]

Yes, but building the quality cycling needed for this relatively inefficient process takes up a lot of space. And while you can shove that off to its own islands (this doesn't require anything from scap), its still going to take up a lot of room. Whereas the Gleba option can rely on multi-stage pipelines and quality modules, being much more space efficient and tapping into other Gleba processes for making quality goods.

[hoary current]

if the Gleba route need less machines than chemplants on Fulgora for the same amount of solid fuel output, even if it consumes finite resources I would still go for the less infrastructure option

[gusty trench]

I think the benefit of doing it on Fulgora is simpler infrastructure: a line of chem plants that have one fluid input and one solid output, and recyclers taking that in and spitting quality versions of the same out. Chem plants aren't very expensive, power is easy, and you really just need a train to port out the quality solid fuel to make into rocket fuel elsewhere

Or you could train in water for cracking and then export quality rocket fuel

Genuinely it could take 5 minutes to set up

[frank warren]

this is the craziest news we have gotten so far. What quality fuel do you think is best for UPS?

producing the fuel obviously costs UPS, but presumably faster trains = less trains = better performance. But the cost of quality is exponential, so this may be a case where something like or is actually superior

[kindred crater]

100%. when you do roll a lucky or you could either prioritize that for longer distance trains, or those in your fleet with the slowest acceleration time to minimize slowdowns

[fallen urchin]

What about the first 40 U235?

[kindred crater]

another thought, top speed/acceleration achieves the same result as more joules in fuel in an indirect way

[fallen urchin]

For qual cycling fuel cells, is it better to directly recycle the cells, or burn them through a reactor and use reprocessing?

[obsidian crescent]

Depends on the goal. If you need 40 high-quality U-235, then burning them in a reactor isn't helping (remember: Kovarex forbids quality modules). If you already have 40 high-quality U-235, then what you need is U-238. So "quality cycling" through using them in a reactor is viable. But only to the extent that you can burn lower-quality fuel cells.

Actually quality cycling UFCs will likely be more efficient in terms of resource cost though. You can use more quality modules, so it will take fewer cycles to get good output.

[daring siren]

Hard

I guess you can upcycle 235 you made with

[fallen urchin]

To qual cycle U235, you either cycle or

I think fuel cell is more viable, because you do need the qual U238 after getting all the qual U235

[winter saffron]

When you hit the 300% prod on all recipes you might as well setup quality research to save some UPS.

Don't expect it to amount to much tho

[winter saffron]

I thought you could place in ♻️ ? So no need to craft.

[daring siren]

That is correct. You can upcycle u235 into itself

[obsidian crescent]

Not all recipes have productivity research though; most don't. So prod modules are still better than qual modules even for them. Also, 300% prod research only means that the cost of producing a product of that type with arbitrary quality is the same as the recipe cost. The cost of the base quality is 1/4 the recipe cost.

[burnt reef]

this doesn't seem right to me. why is to the only cost increase?

going input -> output -> input -> output still gives 300% net productivity. why would adding quality modules in the recyclers change that?

[obsidian crescent]

It gives net 0% productivity. 300% productivity means you get 4 outputs for the price of 1. A recycler returns 25% of the inputs for whatever you feed it. Therefore, a full cycle returns, on average... exactly what you gave it.

The only difference is that quality modules in the recycler can boost the quality of what it returns. Thus, you can raise the quality of the output without actually losing any materials.

But for base quality stuff, you don't feed it into a recycler. So the 300% productivity bonus is the actual net bonus.

[iron root]

You still get the 300% productivity with quality though

It just depends if you want the quality on the recipe inputs or the output

Both can level to for no additional material cost though at 300% prod

[obsidian crescent]

In one cycle, possibly. In aggregate across all crafts? No.

[iron root]

So if we’re quality cycling around some 300% prod blue chips
20 green chips and 2 red chips make 4 blue chips, which recycle back into said 20 green and 2 red

So that lets you upgrade quality for ‘free’ via quality mods in the recyclers

But once it’s all , it’s still 20 green chips + 2 red chips to 4 blue chips

At that 300% prod point you can look at it as a recipe with altered ratios that a recycler can undo with 0 loss

[iron root]

To directly link this to what I said
Yes, you feed quality stuff into a recycler, but then you craft it again with the 300% prod

[burnt reef]

yeah it's craft (x4), recycle (/4), craft (x4), which combines to x4

[iron root]

And this can actually be generalized to productivity overall altering ratios and recycling efficiencies (for recipes that properly recycle anyway)

[burnt reef]

so cycling the ingredients to is free, and the finally crafting the item gives x4 output

[iron root]

More or less yeah

Once you can start making recipes hit 300% productivity, the cost of quality is based on how many steps away a material is from said recipe’s inputs or outputs

Or really steps down from the inputs I guess
Steps up from the outputs is… funky

[half raptor]

slowly over the last year, despite stubbornly clinging to the for guidance, I've been slowly coerced into accepting into my heart. I plan to do 2 runs, one for a quality heavy run and the other for a prod heavy run, but once you hit end-game it probably will run similarly

[iron root]

is just so weird because it’s benefits have a different relation with time compared to

Cause the benefits from quality whatever persists after you remove the modules.
Productivity doesn’t do that

[winter saffron]

You could setup a low priority set of train stations which upcycle input to legendary (or whatever) to fill a building mall. That way excess produce gets upcycled automatically.

doesn't have to be anything massive just something to keep running in the background

[rotund egret]

re: uranium fuel

I'm probably going to recycle ore itself since it's the earliest step, therefore benefitting most from prod/quality

[obsidian crescent]

... how? If you recycle the ore, you eat up over 75% of the ore. There's no prodding in the cycling whatsoever. Whereas if you quality cycle UFCs, you can prod the UFC production, thus reducing the cost of the overall setup.

[heavy ember]

Oh no.
I can see that power from reactors to be byproduct if there's a need for quality uranium isotopes

[jaunty citrus]

0.25 * 1.4 * 1.4 is the same as 1.4 * 1.4 * 0.25. If you're doing the same steps it doesn't matter where you do the loop.
However looping UFC's gives you an extra step to take advantage of prod and quality modules with.

[frank warren]

so many people miss this 'communitive argument' - I'm glad to see it here

[rotund egret]

it does require less infrastructure the earlier in the chain you do it

since you still need to mine the uranium ore

[jaunty citrus]

It would require significantly more as you can't benefit from productivity in an ore recycling loop

Recycling UFCs would let you use productivity modules and quality modules in the crafting step and the recycling step

[rotund egret]

can't you still prod the uranium after you obtain the high quality versions?

[jaunty citrus]

With ore cycling you only get the recycling step every loop, no crafting step

[rotund egret]

what about quality-ing the miners?

[jaunty citrus]

You could do that anyway, but it only applies once, not every time the ore loops

If you cycle with an item that you craft, getting a legendary item (assuming legendary modules are used) can take only ~80x the resource cost of a normal quality item.

If you only use quality modules in the crafting and recycling step, it goes up to 116x times

If you skip the crafting step modules entirely (like you would with ore cycling) the cost goes way up to 2727x the cost

So if you cycle ore straight out of a mine, every 2727 normal ore you mine will turn into 1 legendary ore

[grand igloo]

can't you theoretically get to 325% prod?

oh wait right they capped it

so the high module slot, innate prod buildings just get you to the magic 300% faster

[daring siren]

ye

once you get high enough, you can start using qual instead of prod

[half raptor]

I can't wait until we can get the game and see what the maximum we can research it. I'm hoping we can research all the way to 300% so even assemblers will be capped

[daring siren]

They did say you can

but it's very expensive

[teal wren]

Yes, there were no denial that lvl 30 is possible. So even assemblers can be capped. But going past level 25 is extremely hard

[daring siren]

You'll want a 1MSPM base for those researches 😉

[half raptor]

Let's be frank. Expensive just means all the reason to make a long term robust base

[teal wren]

You'll need such base

Hmm... it will only take 6.4 hours on 1M SPM to get to 1 level 30 research

[half raptor]

How do you know how expensive a lvl 30 research is?

[teal wren]

1500 starter price with 1.5 exponential step

[half raptor]

Fair

[teal wren]

If all starter prices are the same

But RCU is (was) end-game item, so I don't think it will be higher than 1500

[half raptor]

I plan to just research in order of prices

[teal wren]

Most viable path is to research lowest cost productivity research

Except mining I suppose, as it grows linearly

[half raptor]

Since prod bonus is multiplicative across the production chain anyway. Cheap upgrades helps the whole base

[obsidian crescent]

Note that the productivity researches don't appear to use yellow science. They still use space science, but that's much cheaper now since launch costs are down. They're still expensive, and exponential scaling is exponential.

So hitting productivity 13 for blue circuits (the earliest that you can hit the break-even point) will probably require serious megabasing.

[half raptor]

We'll know when we know

[rotund egret]

I have a weird hot take

I think yellow science has been cut from SA, since almost all its researches have been moved off-planet and automating rockets is now a space science thing

[obsidian crescent]

And yet, they keep showing labs with yellow science in them.

My take is that there are some new automation/remote techs that yellow science provides. Like, if there's an earlier game remote-controlled vehicle for the RTS, it makes sense to shove it in yellow science.

[burnt reef]

if productivity research cost scaled linearly, then the net cost per level would decrease

so level 16 would effectively be cheaper than level 15

[obsidian crescent]

It would depend on what exactly you research productivity for and how you make it. Steel productivity wouldn't matter if you use the Foundry LDS recipe, for example.

[burnt reef]

yeah I'm abstracting several details out but the point remains

if lab prod had linear cost and effect scaling, the effective cost per level would have a finite cap

further ingredient productivity just makes it cheaper yet

[burnt reef]

if it was +10% research prod, +100% cost per level, the effective cost would never go above 10x level 1's cost

there are other ways to prevent this (sublinear effect scaling, polynomial cost scaling) but I think linear effect & exponential cost is most likely

[frank warren]

We know the cost function

@rotund egret I would like to mention that because yellow and purple science are optional before rocket, we will have technologies on the first planets that do require each of them, so if you rush for example Vulcanus first - there will be vulcanus techs that require neither, yellow, purple, and yellow+purple

[rotund egret]

yes

I was just thinking since the devs mentioned they've been trying to crunch as much of the game together, as it's already quite long. science-wise, fusing yellow and purple science together would make sense, even if it's not currently implemented

in the end I did say it was a hot take 😄

[kindred crater]

not sure if you're joking or not but thats my plan lol

I got a few more months to save up money to get a dedicated server that will run the game unpaused 24/7

[charred gyro]

Isn't that something that'll take... months?

[kindred crater]

1k spm is like 138 IRL days to reach lvl 30 for a single prod tech

and from what we've seen so far in SA, we'll blow right past that

[frank warren]

lol funnyduck is running with this new found knowledge

just because your deck can do 10kspm doesn't mean you can do 10kspm

[charred gyro]

I am expecting "infinite" techs to take a ridiculously long time to complete in increasing requirements. Otherwise they wouldn't really match their namesake.

[obsidian crescent]

You don't need level 30 for any productivity tech. Level 20 + 4x Q5 prod 3s is enough for anything, and anything that comes from a 50% prod building can stop at 15. Blue circuits get to stop at 13 because the EMP has 5 slots.

[frank warren]

yes but level 30 lets you replace the prod in the machine with speed

[kindred crater]

thats right

[frank warren]

which is not nearly as impactful, but there still are benefits.

[kindred crater]

higher performance builds, more UPS savings

[frank warren]

after level 20 cost increases super exponentially, because you aren't getting a prod benefit anymore, just a speed benefit.

so the cost per extra rate is doubly higher

[obsidian crescent]

That's pretty meaningless if you're speed-beaconing that machine already. What's +125%*4 when you're already at +2000%?

[frank warren]

(same increase in cost structure, but dramatic decrease in benefit structure)

[kindred crater]

pretty much in line with all prod research and quality

in 1.1 I think people have reached 50k spm. with quality and prod research it could be 10x that

if its an aquilo building, it would have to be something powerful and special

[frank warren]

50kspm 60ups has been achieved.

can't download it on steam deck tho

meet gabe newell speedrun

[obsidian crescent]

With quality and the new beacon change, everything's already fast enough. A single building can fully stack several green belts on some recipes. There is zero reason to have an even better beacon.

There is one more building typically combined with foundry and electromagnetic plants, which will be covered later.

The only thing that you "typically" use with these is a chemical plant for plastic and sulfuric acid.

[charred gyro]

Yup. And it'd lead to even more of the stuff that people complained about. Optimal builds would involve multiple rows of these beacons since they could overlap each other and still reach the same crafting machines.

[frank warren]

yes because we thought there were 3 unique machines .. that would be on the first 3 planets

[obsidian crescent]

We also thought there would be new oil processing setups on Gleba, and that was correct.

[frank warren]

from the suggestion that you use the aquilo machine with the other two, it sounds like only the second half was wrong.

there are 3 unique machines that work together

[charred gyro]

Eh... we didn't get a list of recipes that the biolab machine can do, did we? Not like the details we got for the other planets' unique crafting machines

[frank warren]

but it also feels weird as hell that the aquilo machine would just do a few oil things... not very thematic

[obsidian crescent]

We didn't get a list of Foundry recipes until the Factoriopedia FFF. And while they told us what the EMP makes, the EMP is more of an assembler than a refinery. The Biochamber is much more of a refinery-type building, a thing that makes intermediates, not the final product.

[fallen urchin]

Considering SPM is measured in bottles consumed, a 1M SPM base will take less than 1000 minutes to research a 1B cost tech

The SPM only measures the base size, and lab productivity needs to be calculated after that

[frozen fiber]

True. But it was probably still several 100k SPM

[daring siren]

I call it eSPM

[teal wren]

What is it? SPM counting all lab productivity? Or SPM per real-world minute?

[daring siren]

Effective SPM

i.e. the rate at which science is getting completed

[teal wren]

That's what the new science icon on the chart?

[daring siren]

This value https://cdn.factorio.com/assets/blog-sync/fff-408-science-graph.png

[teal wren]

Very useful. But damn this placeholder needs to be updated

[daring siren]

hmm why?

[teal wren]

Also as I was saying also need to be updated to something more slick

The thickness of lines not even the same on the corners

[teal wren]

Looks very cheap and flat for my taste

[daring siren]

Oh? It's the "selected" visual from the game

[teal wren]

Also wait what, how science consumption graphs can be different? aren't they all should be perfectly aligned

nvm

Because each potion has it's own indepentent resource level

[daring siren]

Well I would expect them to be the same in the last 10 minutes consider they are working on the same science

so your question does make sense

[frank warren]

Correct, all though tbf I think 1M SPM is very much achievable in space age from what we know.

[daring siren]

But very hard

It'll be like doing 25kSPM in vanilla

[frank warren]

Agreed.

If you really think that you expect 2M in SA though :P

I have lower confidence in the estimates overall, I think 1M will still really be pushing it.

More like 50kSPM in vanilla + hardware improvements

Finally we can put the discussion of 'what is a megabase' behind us - it's a base that produces 1Mspm :P

[terse lagoon]

What i wonder, 1m spm they talked in fff, was it 1m science bottles per minute, or 1m spm after prod multipliers?

Cause if second it kinda not legit for me

[frank warren]

after prods

no confirm on it, but kovarex isn't a UPS master, and all of our knowledge says that 1M science is in the 'believable but difficult' area

[terse lagoon]

Ew

[frank warren]

It sorta means nothing, but I expect lab prod to be so expensive I doubt he has more than 200-300%

[winter saffron]

Not a fan of prod. science either.

Maybe it has a 300% cap. Which I'd prefer if they want it no matter what

[frank warren]

It shouldn't have a cap

You can't recycle science points

[fallen urchin]

MP shouldn't have that cap either, you can't recycle ores into the ground either

[frank warren]

And it doesn't.

[daring siren]

There's no limit for lab prod, but it's likely an exponential science

[teal wren]

Yeah... if this research wasn't - I think it would end up being no-cost (in terms of real science packs) increase research. As productivity goes linearly as does price

[frozen fiber]

It would still get more expensive

Just like mining productivity does

You pay way more for 100th 10% increase than for the first

[teal wren]

Yeah, but with mining productivity you eventually will mine out all resources

[frozen fiber]

So?

[teal wren]

But I am talking about lab's productivity bonus compared to research cost of next level of lab productivity

so assuming each level is 1 research unit, so the next is 2 and then 3

an lab productivity goes like 1.1, 1.2 and 1.3

And this ratio is asymptotic

[frozen fiber]

Do you mean productivity(n+1)/productivity(n) ?

[teal wren]

No, i mean research cost / productivity. I.e. x / (1 + x / 10)

[frank warren]

Akara is correct

Welll, somewhat correct.

The price would never change if it went up linearly

[frozen fiber]

But the cost of every 10% step does increase

Are you guys saying mining prod 100 is as cheap as mining prod 1?

Because it obviously is not

[opal haven]

ratio might be 1 -> 1.2 -> 1.4 and so on

then cost will increase over time

[teal wren]

That sounds like exponential.. but with some extra steps?

[opal haven]

cost formula will be 1+0.2*level - looks non-exponential to me

[teal wren]

But this is exactly linear

Lab research cost cannot be linear or otherwise ratio will end up being some fixed asymptote

[frozen fiber]

She said prices won't go up, ironically by arguing that prices do go up (but only linearly). So no, she is just wrong

[burnt reef]

we're saying that, in terms of resources in the ground, mining prod 100 is nearly as expensive as mining prod 200

[teal wren]

1st. I am not "he".
2nd. Not the price will go up, but the ratio of tech price to lab productivity

[frozen fiber]

Ok, sorry about misgendering you

[teal wren]

So it will end up that each next research will take the exact same amount of time

That does not happen with mining productivity, because it does not affect what labs are doing, so maybe in terms of resources you are unrestricted, but you still need to expand

[frozen fiber]

This is a completely different statement though

[teal wren]

That is the same statement when I said that ratio will go to the some fixed asymptote

[teal wren]

And what previous statement are you reffering to?

[frozen fiber]

Well if you say "price" I'm going to talk about the price

[burnt reef]

mining prod 100 costs ~9.2x as much as mining prod 1
mining prod 200 costs ~9.6x as much as mining prod 1

[frozen fiber]

That the price won't go up

[burnt reef]

it's just the processing that gets more expensive

[teal wren]

Where did I said that? 🤔

[frozen fiber]

Here

[burnt reef]

and research prod comes after processing, so that's not a factor

[frozen fiber]

And some other places as well

[teal wren]

Well, it increases in technology's price cost, but in terms of real science packs you are putting in - the cost is the same.

Sorry for confusion

Will get the same as you go higher and higher in level

[frozen fiber]

Ok, you're right that effective cost per step is asymptotic

But assuming 10% prod per step, e.g step 10 is still 5 times the cost of step 1

Our rather it would be, because the price increase will most likely be exponential anyway

[teal wren]

Yes, for such powerful research if would make sence to make it exponential

Also because being non-exponential comes to real price stall

[gusty trench]

I thought all the infinite prod recipe bonuses were exponential in cost? Something like 50% increase for each one

[frozen fiber]

It was a discussion about a hypothetical

What if the cost of research prod increased linearly

[gusty trench]

Ah, gotcha. With the way prod can compound, I wouldn't be surprised if some combination of prod researches could get cheaper over time

[burnt reef]

if there was uncapped science pack productivity research that also scaled linearly in cost, it would get cheaper over time

[daring siren]

it won't be linear

[burnt reef]

(recipe productivity is capped b/c recycling and therefore is asymptotically constant, but science packs can't be recycled and therefore don't need the cap)

yeah but if it was

or, perhaps, this is why it won't be linear

[gusty trench]

Among other reasons, yeah

[opal haven]

hm, true, I was incorrect here. research must be exponential

[frozen fiber]

It could be quadratic

Anything above linear really

Would solve that issue

[burnt reef]

plot twist: cost scaling turns out to be O(n log n)

or even O(n log* n)

[fallen urchin]

In terms of ore in ground, somewhat yes. In terms of factory manufacturing, no.

MP cost of ore in ground will eventually converge to a fixed value, but MP1 is nowhere near that value

I think MP100 will cost like 6x~7x of MP1

[fallen urchin]

There's another game "Final Factory"

It's basically oriented around productivity progression

[frozen fiber]

It costs 12.4 times as much, in fact

While MP200 is only 6% more expensive than MP100, in terms of ore in the ground

By MP1 I mean the first of the infinite researches, so technically MP4 in the game

[grand igloo]

The cap probably applies for everything, just in case people make packs recyclable

[obsidian crescent]

Mods can change the cap, so if someone wants to install a pack recyclable mod, they've already decided that they don't care about the cap.

Also, I really think getting level 30 on any of these productivity packs in a reasonable timeframe is just wishful thinking. Hitting level 15 will already require a lot of SPM or just days of leaving the machine idle (and hoping that your defense hold).

[frank warren]

Math with what we’ve seen says it won’t take so long.

[daring siren]

1MSPM would help

[obsidian crescent]

If you can get 1M SPM without having max-quality everything... why do you care about getting max-quality everything?

[final nimbus]

For more SPM of course.

[daring siren]

[heavy ember]

What's the best guess on how the prod techs scale?

[frank warren]

we have the exact formula assuming you can find the message.

you know whose messages to search

[daring siren]

was it a K or BK message?

This is part: #friday-facts message

200M sounds reasonable at 1MSPM 😛

[jaunty citrus]

It's 1000 * 1.5^L
#friday-facts message
So it get's ridiculously expensive in the later levels.
Some cumulative costs at break points:
LVL4: ~12.2k (40%)
LVL10: ~170k (100%)
LVL13: ~580k (300% with EMP)
LVL15: ~1.31m (300% with Foundry)
LVL20: ~9.97m (300% with )
LVL30: ~575m (300%)

So you can get 98% of the benefit at level 20, which is less than 2% of the cumulative cost of level 30

[obsidian crescent]

So what does it take to make 580k space science? Isn't that quite a lot of uranium? If one launch of U-235 can make 1000 space science packs, then that means you need 580 rockets full of U-235. So... how much is that?

Let's say 500 U-235 per launch, so a 1:2 ratio of U-235 to space science. So a total of 290,000 U-235 total. Assuming a 3:1.5 U-238:U-235 Kovarex ratio (ie: 50% productivity), that's nearly 580,000 U-238. So that's 5.8M uranium ore that you'd need (less with prods).

So in Space Age, you might actually exhaust a uranium patch 😉

[jaunty citrus]

You had me for a bit there hahaha

"Oh no I might need a second uranium patch 😱 "

[obsidian crescent]

Well, you probably want more than one patch just to get through it faster.

But one interesting point is that blue circuits and LDS productivity both require space science, while we know that steel doesn't. Because of that, you can actually research steel on multiple planets simultaneously. That is, all planets can have their own labs and science makers, with all of them contributing to steel productivity research.

You can't do that with anything gated by space science, since that can only be made on Nauvis.

Granted, you can't quality cycle steel the way you can LDS or blue circuits, so it's a moot point for quality. The same goes for plastic.

[jaunty citrus]

With many key items able to be made in machines like the EMP, I think the viability of quality cycling materials for "zero cost" legendaries gets lower.
If you just cycle the end product you don't need ridiculous levels of Prod research to make it viable, and you don't need to do any sorting/balancing infrastructure for reverse crafting items from intermediaries like and

The biggest ones being modules, beacons, and the EMP itself, as well as the foundry in the foundry.

I wouldn't be surprised if the last machine is a similar 50% prod building that can make itself as well.

The EMP brings down quality cycling from 116x cost in assemblers to just 23x assuming all legendary quality modules in a non-prodable recipe.

The foundry with one less module slot is 32x

[daring siren]

You don't need even close to "zero cost" for it to be super viable

Also, lets remember lab prod is a thing we'll also be investing in

Getting lv20 or even 10 of lab prod would make the rest cheaper

That said, there are many sciences, so it doubles up again

[jaunty citrus]

My problem is that you're limited to what materials even have productivity research. I'm not sure that any of them involve concrete or iron ore in their recipes, which would count out advanced machines like the foundry and recycler from this reverse crafting from low cost legendary intermediaries significantly less viable.

Same thing with tungsten carbide actually, there might be a recipe using/producing it that gets prod research, but if it doesn't that's another severe bottleneck, alleviated by just cycling the foundries/BMDs directly.

Also when doing the reverse crafting from the cheap legendaries, you necessarily increase the cost based on how many steps back you need to recycle.

[obsidian crescent]

That kind of assumes that tungsten carbide only gets used for those two things.

[obsidian crescent]

Quality stone brick on Vulcanus just kinda happens since every lava processing machine will be spitting out mostly useless stone. You can run stone brick though a furnace with qual modules, and even quality cycle away the leftover brick instead of just casting them back into the lava. And the Foundry recipe for concrete gets its iron in molten form, so the only thing that determines its quality is stone brick (or maybe stone if it doesn't use brick). So that's pretty cheap.

Oh, and quality cycling tungsten carbide gets you quality carbon and tungsten ore, which can be used to make quality tungsteel.

[heavy ember]

There's two recipes for space science though

[jaunty citrus]

There's only one recipe that is viable beyond initial space platform research though

[jaunty citrus]

It does sound quite reasonable, I guess it's just tough without knowing what all the recipes are yet.

[obsidian crescent]

I have to assume that each planet has at least one intermediate that gets a productivity recipe. It is very weird that tungsten carbide is not an ingredient in tungsteel, despite tungsteel using the same ingredients as carbide (plus molten iron). So it's not clear which one would make the most sense as having a productivity research. If carbide is useful for anything other than the Foundry or BMD, then giving it prod research would be more impactful since it's made in an assembler (no 50% prod building).

[grand igloo]

isn't concrete cyclable?

[jaunty citrus]

Concrete isn't prodable and probably doesn't have a prod research

[grand igloo]

yeah but it's pretty cheap

iron ore, stone brick

[wary beacon]

once you get to that point, you're going to have some serious prod mod action helping you at every step of the and space science process, plus launching rockets is going to be practically free

it's very nice to have a use for uranium besides just blowing stuff up and nuke plants as well, i've modded a simple space science recipe into my game and it's pretty swell

[obsidian crescent]

I just realized. Because the space science in space recipe is made... in space, you could craft science packs on the way to other planets and just drop them off, then return to Nauvis to pick up more U-235 to do it again. The other ingredients are all space stuff, after all.

Which means you still can use the "build all the science packs on each planet" trick to move things along.

I'm not sure that Fulgora is a good place to try that on though, since some of the earlier packs are weirdly expensive there.

[jaunty citrus]

is 1000 of and maybe worth the value of one rocket to send it back to nauvis?

You could make it all locally, or you could just make a rocket and send it to one place. Not sure about all the planets but given the weird costs on Fulgora I agree it would be especially bad to make the other science locally

[obsidian crescent]

The goal was, instead of localizing all packs to one planet, to use every planet's resources to make all of the packs needed for the productivity sciences. So your Nauvis base might be able to hit 1k SPM, your Vulcanus base hits 2k SPM, your Gleba base gets 800 SPM, and your Fulgora base can hit maybe 500-600 SPM. This kind of distributed production allows you to take advantage of all of the resources available to you, rather than trying to build one planet up to the point where it produces all the packs through purple at 3k SPM.