Friday Facts #375 - Quality

[KuuLightwing]

You will only need a recycling loop at any one level of construction. If you choose to do it at a low level, so that you have top-quality basic ingredients, then everything made from those ingredients will be top-quality automatically. Conversely, if you choose to do it at the final stage (the bots themselves), then you will only need ordinary ingredients at all the previous levels.

I feel that it's honestly kind of an issue on itself. It feels very all-or-nothing system, as if you want to use quality anywhere in your production chain, you have to now go all out on it, an make all the ingredients you are using of all the possible qualities you could get. Which is probably doable with bots, but sounds incredibly tedious.

So you either grind every base component to max quality at some early stage before it spiraled into tens of intermediates, or you just loop the final product itself. Probably latter, cause seems like you'll really want to have productivity for everything else.

[lolleroni]

Love these new optional mechanics and I enjoyed the beginning quote in the patch notes. Rewatched the matrix and excited to play the expo!

[TheHeratic]

my main question is if the naming scheme is final, or if the graphical indicators are, right now it seems like like adding quality and more adding rarity.
the use of colors to differentiate the types of material is fine but I think that adding more of a change to the item itself would be better, granted adding slight changes to the 100's of items in the game would be excessive; but the naming scheme is the big problem for me. I should state that all of this is purely my own subjective opinion. the labeling of what are essentially rarities being , normal, uncommon, rare, epic, and legendary make it seem like a loot based game, like diablo or borderlands, but factorio is a factory game, less focused on collecting loot and more on mass producing it which you can do in this game, however the naming scheme does not follow that. I think the naming of the qualities is what's causing people to get upset, it doesn't feel like it matches the rest of the game.
idk, its hard to say exactly how I feel since I haven't had an opportunity to mess around with it. It's a big change for sure so it's bound to have a lot of conflicting feelings from the community, It'll be interesting to see what else will change in the expansion

[silver27]

One thot what didn't let me sleep: for the game that so consistent what you could solve it from the beginning such randomness seems out of place.
All production chains are one sided because it is only improvements and you never want to recycle anything. Only exceptions is pistol, burner miners, stone furnaces and maybe wooden power poles.

And now there is 5 tiers of Quality, and now you get things what you don't want and need to recycle them.
Ah recycler, only 25% why so low? To avoid a net positive recycling loop. For new prod modules there is cap at +300% which is 4 times and recycler gives back 1/4 of original cost, it gives back exactly 100% of best productivity you cannot reach without Quality modules.
So if only purpouse of the recycler is give a rerol, it would be much more convenient to make "rerole machine" instead. Or quality improvement for intermediates only.

All Quality features are ether inconsistent or expensive which made them bad for early-mid game. If you suppose to stall it until you reach very high production and then spent infinite resources on quality improvement then there is 3 intermediate tiers of good for nothing stuff what you barely use.

My conclusion: Quality may be good but it need EXTREME balancing and Definitely won't be an optional.

[Tricorius]

Except uranium takes your entire argument behind the woodshed: the existence of kovarex proves how important determinism is to the game. The entire U-235 chain would be significantly less useful if all we got was a trashcan to clear up the system. Quality is both significantly more complex and wider ranging, but completely lacks any mechanism to put a lid on the slot machine.

Actually, Uranium and Koverex is an excellent parallel. It’s mathematically proven that over time, the standard refinement even on a small patch of uranium, is sufficient WITHOUT koverex enrichment to power all but the most crazy huge bases. And those bases usually move AWAY from nuclear back to solar (though that is due to game engine issues).

You don’t NEED quality improvements. Just as you don’t need to setup Koverex.

They have already said there are NEW MECHANISMS yet to be revealed that will help with quality.

[pwootage]

I'm sure most of what I'm going to say has been said before, but this is bothering me enough that I decided I ought to add my 2 cents of feedback. Ok, after writing it, more like 45 cents. This is a lot longer than I originally expected it to be. I guess I feel strongly about this

TL;DR: Quality adds too much complexity, in a way that persists throughout the majority of gameplay (from modules unlock to the end of the game).


I, of course, have not playtested this, and I'm sure a lot of design thought has gone into these systems; but I have a number of major concerns about a system like this. It seems to fundamentally change the gameplay (once unlocked) to the point that it makes Factorio an entirely different game to play, and one that violates some fundamental rules of (current) Factorio.

In the base game, every factory can be perfectly designed to have perfect ratios using simple machinery, with simple tier management and straightforward scaling rules. Everything is predictable. Production rates can be calculated to exacting precision. The tools available to you are concise and have obvious tradeoffs and benefits. Building a mall to automatically construct all your tooling feels natural and compact, even without resorting to logistic bots. Crafting processes are straightforward, all being able to be broken down to a list of inputs and outputs. There are no mandatory circular recipes, and even multi-output processes (oil and uranium) have mechanics for which you can manage the balance of the outputs. It's satisfying, completable, and easy to reason about.

Adding (at the very least, rng, cycle-based, filter-based, as described in this article) quality completely invalidates much of this process.

It is no longer possible to build an optimal factory, as it would require legendary of every product, which is simply not feasible. Scaling rules are complex, as now you have 5(!) versions of each machine of each tier and module you need to take into account. Is it better to use a Rare Assembler 2? Or a Basic Assembler 3? It's difficult to intuit and reason about, due to sheer scale.

This dramatically increases complexity for the vast majorty of gameplay, for any item for which quality is helpful. Every single time you want to have a quality version of an item, you *must* build a disassembler loop. The sheer fact that a disassembler exists reveals the fundamental problem with this scheme: almost every single item in the game is now a circular recipe. Particularly nastily, this affects specifically the items in a resource mall far more aggressively than most, since these are the machines you would want better versions of. Dozens of items, all of which you now have to consider building a new loop for. And this is true from the moment you unlock quality modules, for the rest of the game. Which just got longer, thanks to the expansion. Even for items that you don't need quality for: how do you know that you don't need quality? It's extremely difficult to reason about, there are simply too many additional factors that you now have to consider.

This fundamentally changes the core gameplay production chain design logic from the moment its unlocked. No longer am I thinking about production trees and scaling, I'm thinking about how and when I'm going to fit recycling loops into 50% of the things I'm producing, so that I can hack in better performance with those things in specific places. Which also requires continuous maintenance as you unlock more quality tiers and produce better quality modules and machines.

In addition, a small list of less major issues that this system causes:

• This increases the number of items in your inventory significantly (likely 2-3 versions of some of the machines you would carry around normally). This is both distracting and adds additional inventory space requirements.
• The visual clutter from indicating the level of everything is just brutal. Just look at the example posted in the blog post. If you zoom out at all, it's nearly illegible.
• Blueprints are tricky, especially in a blueprint using multiple qualities of things.
• Related, cut/copy/paste, upgrades are going to have some of the same issues.
• This quality system boils down to "increase cost by (x)%, increase electricity usage by (x)%, take (x)% longer, add a bunch of filtering logic." It's a new recipe with more steps, mandatory filtering/circuit logic instead of production chain, and very low/slow yield.
• The buffs from quality are *absurdly* strong. Based on the numbers in the article, quality is more powerful than modules, and by a good amount too. Enough that they likely affect game balance for people who don't want to use them.
• Factorio has always felt to me like a game about horizontal scaling. The copy/paste/blueprint feature, to me, defines Factorio. I feel like there's plenty of vertical scaling through modules already (almost too much, frankly).

Thankfully, modding exists, so I can simply remove quality from my game (or rebalance, or otherwise modify, if that ends up making more sense). I would rather have an assembler 4, than 5 versions of assembler 3. In the end I would like the unmodified xpac to at least make sense and maintain that feeling of "an expansion to Factorio" rather than "A game similar to Factorio".

I'm sure I'll play it regardless. Factorio is a very good game, which is why I felt so compelled to provide feedback. I love it now, and I'm sure I'll love it for years to come.

Thanks for reading <3

[maxp779]

Ooft this was a divisive feature! For anyone around when they added bots and beacons and modules, were they equally divisive? I got this game after all those things were in place.

Hahahahahahahhahahahaha :: deep breath :: hahahahahahahahah.

You *still* occasionally hear someone say how cheaty bots are because they avoid the “belt puzzle” that is the “core of Factorio”.

Sounds like some people just don't respect that others play Factorio differently to them. Im glad logistic bots etc stayed in the game, very cool feature.

[greatstarguy]

Will jump back into the flame wars on a couple subjects that are key points:

1. Kovarex and uranium are the only non-deterministic recipe in Factorio so far, and have been brought up again and again when debating quality.

What uranium does before Kovarex:
The extremely low drop rate (~0.7%) for U-235 tells you, as the player, that you shouldn't plan around having too much of it at this stage in the game. Looking at the map tells you that uranium isn't common enough to rely on this process for U-235, so it's enough to have a filter inserter leave it in a chest. If you're hurting for energy, you can rush nuclear without Kovarex and do fine, but otherwise, it's basically just uranium ammo.

Kovarex is a signposted (through the research tree) way to get around this bottleneck by turning mostly-useless U-238 into very useful U-235. The saved U-235 from earlier comes in handy now, and this is when you can start thinking about large-scale applications of nuclear, like reactors 2x4 or bigger, and atomic bomb. Kovarex feels good because it turns uranium processing, which is pretty mediocre on its own, into something that lets you think at mega-scales: make power for a mega-base, deliver boom by the mega-ton, kill biters by the millions.

Why quality system is materially different from uranium:
With quality, everything that results in a quality+ is kept and is useful immediately. But actually mixing high-quality and low-quality buildings and items together results in serious inventory clog, so there's competing incentives at work here. Furthermore, the low chance of getting quality - only 10% upgrade with 4 T3 modules is steep - means that in order to get large amounts of high-quality items, you need to build really wide - each tier costs something like 6x the previous tier. The big issue with this is that there's no way to meaningfully reduce the grind except by building more inputs, more mines, and just feeding raw resources in. Quality needs some kind of Kovarex-like system to significantly shortcut grind once certain technology levels are reached. Otherwise, the signposting is not as clear, it becomes a noob trap, and the feeling of progression is lost as you build more of the same instead of something new.


Maybe a better way to look at quality is that each tier of quality represents some level of "compression": this Q3 red circuit took 100 copper, 50 green, and 50 plastic instead of 5,2,2. (Fans of modded Minecraft see octuple compressed cobble and agree.) But then why is there this +0.1% chance for a medium product to become a super-highroll? Practically, all this does is add clutter in that you have to filter out 3 or 4 byproducts, and add a belt line or something that will be used once a playthrough.

I'm a big fan of fewer, but more distinct tiers, like has been suggested previously. There's no additional challenge between automating Q2 -> Q3 and Q3 -> Q4, as the same design can be repeated. And I would prefer the chance for anything greater than a 1-level increase to be removed, or at least relegated to a different machine that isn't an assembler. This cuts down on a lot of the extra bloat that would otherwise be needed, and makes the process seem "fairer".

[greatstarguy]

2. Trade-offs between different levels and quality tiers.

Different items have different use-cases.
This is probably clearest from how inserters are separated.

• Burner inserter demands fuel, moves slowly, but is super cheap. It tells you, the player, that it's meant to be used super early-game when coal is being used everywhere (smelting, steam power) and then transitioned away from.
• Inserter is iconic and constantly useful. Cheap, easy to make, but functional and will move items from A to B. In early-game, when throughput is not as much of a concern, it is always ok, and in late-game, inserters not being fast enough signals the player to upgrade their machines and designs.
• Long inserter is another basic building-block. Once recipes become more complicated and demand more ingredients, it becomes useful to reach into another belt. Similarly to the basic inserter, long-inserter not being fast enough tells you to fix your designs and prepare for megabase-style building.
• Stack inserter (and stack filter inserter) is the high-throughput way to go. Significantly more expensive than fast inserter or regular inserter, but much greater item transfer speed, which is essential for late-game production. Using this tells the player that they're working with end-game, top-tier technology and should plan their builds accordingly.

From just the properties of the inserters, Factorio communicates to the player what they're doing at each stage of the game, and how they should be thinking about their designs. Every inserter has their distinct role to play: inserters in cases where speed is not a concern, stack inserters for loading and unloading mass quantities, and fast inserters for in-between cases. Making many different variants of each tier of inserters blurs this distinction: now maybe the epic inserter is better for loading than this normal stack inserter. Maybe not. But asking the player to make this choice between 6-15 different flavors of inserters messes up this distinction and doesn't actually improve the experience for the player except at the top level, where we effectively choose between stack inserter and stack inserter+. The intermediate levels really don't matter nearly as much, and that's true for most of the machines which have both tiers and quality differences.

[TBC_x]

Quality is tiny feature in the scope of the expansion.

tbc_x_a_storm_inside_a_glass_of_water_7d7e7429-0987-4b15-8a97-fad1d7582f9e.png (1.33 MiB) Viewed 3391 times

[coppercoil]

I'm a big fan of fewer, but more distinct tiers, like has been suggested previously. There's no additional challenge between automating Q2 -> Q3 and Q3 -> Q4, as the same design can be repeated.

I think this is very valuable observation. I like Kovarex process, but Kovarex x 5 doesn't give x5 satisfaction. I think, every tier of quality should have different (orthogonal) mechanics. Randomness is good for one tier. Maybe second tier just requires an additional rare ingredient? Third tier requires enormous amount of resources, energy and time? Fourth tier requires hardening in a cosmic radiation (launching in a rocket)? Let us to brainstorm those mechanics, we like that

[KuuLightwing]

Actually, Uranium and Koverex is an excellent parallel. It’s mathematically proven that over time, the standard refinement even on a small patch of uranium, is sufficient WITHOUT koverex enrichment to power all but the most crazy huge bases. And those bases usually move AWAY from nuclear back to solar (though that is due to game engine issues).

You don’t NEED quality improvements. Just as you don’t need to setup Koverex.

They have already said there are NEW MECHANISMS yet to be revealed that will help with quality.

That is incorrect analysis for uranium processing and kovarex.

First of all, Uranium has additional uses other than powering bases. U235 is used for nuclear fuel and atomic bombs, which are likely to consume more U235 than reactors. While this is also not strictly necessary, there's strong incentive to use those, thus strong incentive to use kovarex to reliably produce atomic bombs and nuclear fuel.

That aside, uranium processing generates huge surplus of U238, and the only thing that consume it without also consuming U235 aside from kovarex processing is uranium ammo which alone probably isn't going to be enough to consume all the U238 you going to produce. So in that sense saying that kovarex processing is not needed is kinda like saying that you don't need cracking for your oil processing - while technically true, that also means that you need to store byproduct in an ever increasing storage (or shoot the box from time to time)

Even ignoring all that and speaking strictly about using uranium for power, yes I can power giant bases without kovarex and without uranium. The comparison to quality is still incorrect though, cause quality is the only way to get 100% productivity bonus, at least as far as we understand, while getting enough power for a base could be achieved via different routes

[EvanT]

I think, every tier of quality should have different (orthogonal) mechanics. Randomness is good for one tier. Maybe second tier just requires an additional rare ingredient? Third tier requires enormous amount of resources, energy and time? Fourth tier requires hardening in a cosmic radiation (launching in a rocket)? Let us to brainstorm those mechanics, we like that

That actually sounds good. Clear additional / incremental processing steps to increase quality.

[gGeorg]

I was thinking about the quality fears, for few days.
It looks to me, the main fear is forced grind. Requirement to build up SPECIAL GRIND sections of factory to produce and recycle loops to reach high quality. That is a play style I dont like too.

However, if we see the quality as random chance to get better part from STANDARD production. Then it adds complexity >> rising up complexity is the main engine of Factorio gameplay fun.

So the question is, how to make sure, the player can not build those special grind sections ?
Well, if recycler could only recycle final products like - Armor, Weapons, D. Towers, Assemblers then player have to subtract higher quality random parts (like green chips) from standard production line, then move it to Hig-Quallity-Mall (tm) for further processing. It also means delivering low grade components from High-Q-Mall back to Standard_line.

Also, if Reecycler unit is very very expensive (at least nuclear reactor price) with high volume recycle capacity (volume of three full stack inserters input) and high base power consumption ( high idle consumption) >> it would create lots of interesting headache >> Put several types of low grade items to one recycler, then sort out all different parts on output and distribute them back into standard lines. LOL. Pure mega puzzle.

Limitation of recycler usage (to final products), would bring more complexity to current Factories, without pure (boring) grinding loops sections.

[KuuLightwing]

I was thinking about the quality fears, for few days.
It looks to me, the main fear is forced grind. Requirement to build up SPECIAL GRIND sections of factory to produce and recycle loops to reach high quality. That is a play style I dont like too.

However, if we see the quality as random chance to get better part from STANDARD production. Then it adds complexity >> rising up complexity is the main engine of Factorio gameplay fun.

So the question is, how to make sure, the player can not build those special grind sections ?
Well, if recycler could only recycle final products like - Armor, Weapons, Towers,Assemblers then player have to subtract higer quality random parts (like green chips) from standard production line, then move it to Hig-Quallity-Mall for further processing.

You really don't want that. Problem with quality design as presented is that once you introduce quality to the system, you have to introduce it across the entire production chain, because mixing quality in a recipe is going to essentially void the higher quality product. Therefore if you use it as a part of standard production, you have to use it everywhere, or it's literally not going to do anything. So, after each production step you have to sort items by quality and route them to dedicated production lines for every tier of quality. While this would certainly introduce complexity, I don't think it does so in a good way, because it's pretty much a "all or nothing" situation, not something you can just gradually build upon.

Recycle grinder at least could be contained to one part of the factory (probably final product anyway).

[Mango]

Interesting.
My humble opinions:

1) As everyone, i think the quality tiers should have better naming. This sounds like WoW. Even the colors are WoW like

2) The quality mechanics is interesting, but the RNG part of getting higher quality feels wrong. In current version, you can always backtrack where is the bottleneck of your production with 100% probability.

Maybe if the probability was changed to additional output of higher quality similar to production modules - something like a bar that continually fills when working and after it fills you are guaranteed to produce higher quality item next time. In this case quality modules make filling of the bar faster and higher tier modules allow higher quality.

3a) Quality is second tiering which in some cases does the same thing as we already have.

For example higher tier ammo is the same thing as researching damage.

3b) Tiering things which are already tiered can create duplicates.

Speed/production module 1 with 'legendary' quality has the same primary bonus as normal module 3.

4) The recycer looks nice, but it is weird that the input box is on the back while inserters can add items to the machine from anywhere. Maybe it could be one big shredder where you insert items on top?

[Necronium]

When people are crying that there is % to some recipe most people will go and automate it and then assemblers will go brrrr cause they dont care about it.

Except, again, there's only one recipe in the game that has this and the way people deal with it is by bypassing it with the later research. And there absolutely were complaints about it when it first came out, but the devs decided to leave it in anyway. Eventually people either leave or just deal with it, correct, but that doesn't make the complaints any less valid.

People are doing what they want but both recipes in uranium process are needed and they focusing on different things. We wanted to get more of whats Factorio already had and we got it now. Devs copied and expanded uranium processing to other items. And people will always complain about something but it doesnt meant that devs should listen to taht. We get uranium chain and it is working fine and it is optional. Now we get excatly same thing for other items but this time rather power production we can get new toys.

Except uranium takes your entire argument behind the woodshed: the existence of kovarex proves how important determinism is to the game. The entire U-235 chain would be significantly less useful if all we got was a trashcan to clear up the system. Quality is both significantly more complex and wider ranging, but completely lacks any mechanism to put a lid on the slot machine.

Seriously, uranium and this side of quality mechanic are really the same. if someone fails to understand how Factorio works it doesnt mean that they are different from each other. Difference is that there are 5 levels when Uranium had 1 and kovarex process is your trash can and upgrader when in quality you have separate trash can and upgrader, thats all. Quality isnt that much more complex, it is just finally we have more options how to play and people overcomplicate simple subject to extremes cause now they have to decide about loops or which quality to use. We even got gif from devs how you can set up this thing. If state of Factorio players is that they cant solve problems and use given tools I fear that Factorio wont go anywhere other than put x items get 1 out and add more research cause something is better.

This accomplishes nothing that researches and recipes wouldn't accomplish better.

Yeah lets delete trains bots or belts depending on what problem players are having with them right now cause there is better solution. Same for power generation and modules when we are at it. There is nothing they provide that cant be done with more recipes and research.

We also can delete fluids cause they are uncessary and are kiling determinism within game and biters too cause they are rng

[Nosferatu]

... Problem with quality design as presented is that once you introduce quality to the system, you have to introduce it across the entire production chain, because mixing quality in a recipe is going to essentially void the higher quality product. Therefore if you use it as a part of standard production, you have to use it everywhere, or it's literally not going to do anything. So, after each production step you have to sort items by quality and route them to dedicated production lines for every tier of quality. While this would certainly introduce complexity, I don't think it does so in a good way, because it's pretty much a "all or nothing" situation, not something you can just gradually build upon.

Recycle grinder at least could be contained to one part of the factory (probably final product anyway).

As I understood it and plan for my factory:
1.) Quality materials and products are only a thing inside your "mall factory"
2.) Get high quality ingredients --> High quality output is guarenteed --> No need for recycler.
3.) Mall will get are real sorting mess because I have to choose where I want high quality output and which items will be used to use up lower quality ingridients.
4.) Unwanted "to low Quality" ingreditens could be used up in the base factory (Science production)

4. Is my main concern because if I try to get rid of ingreditens with quality there it could accidently result in products with quality and then result in stacking problems...

[KuuLightwing]

As I understood it and plan for my factory:
1.) Quality materials and products are only a thing inside your "mall factory"
2.) Get high quality ingredients --> High quality output is guarenteed --> No need for recycler.
3.) Mall will get are real sorting mess because I have to choose where I want high quality output and which items will be used to use up lower quality ingridients.
4.) Unwanted "to low Quality" ingreditens could be used up in the base factory (Science production)

4. Is my main concern because if I try to get rid of ingreditens with quality there it could accidently result in products with quality and then result in stacking problems...

1. Yes, I think trying to introduce quality to the main factory is just a bad idea due to overcomplication of the production lines (and also inability to use productivity modules)

2. You need recycler anyway because no matter how you slice it, you will have products of lower than desired quality, unless you specifically set up a loop that ensures maximum available quality. Sometimes it's going to be the end product, which you have no use for and it's not even used by main factory.

3. If you go this route, you essentially have to use bots for this because they will basically solve the sorting problem automatically.

4. If you were to reintroduce some amount of quality items back to science production lines, it's likely to get diluted to the base quality most of the time because mixing qualities results in lowest one. I still kinda doubt it's a good idea overall though.

[neltera]

i will probably ignore qualities for most of the game due to farming for qualities being wasteful, non-intuitive and practically gambling.
i'd feel shit that my factory does not have production standards anymore.
it is understandable that the quality feature is practically loot as an addition to the game, but i dont play factorio for loot.
in my playthroughs i am deliberately slow and conscious about the pollution and generally have a pretty chill approach to the biters due to favouring blue waters.