Friday Facts #375 - Quality

[Cellidor]

Mirroring others, the only big change needed would be the naming scheme. The best one I'd heard of so far was to avoid different names altogether and just go with simple numbers. Something like "Tier I-V", or "Quality I-V" (Q1-Q5). Keeps it clean, completely unambiguous as to which is better or not, and on top of that, it's more modder friendly, as someone wishing to add higher levels of quality could just add Q6, Q7, with all relevant changes and such without having to come up with some -new- names for each quality level.

Something simple, something elegant, and something fitting with the game's overall feel and themes, that does not imply that tier 1 is bad, only that it can get better from there. "Gamer" terms like rare/epic just don't feel...right. Just like it wouldn't feel right to call something 'magical' or 'mystical'.

[kirkbauer]

I have a suggestion for you. Especially for managing player inventory, although I think it could be applied elsewhere.

A simple setting on the entity (e.g. player, cargo wagon, etc) to enable/disable nesting of quality. If you enable nesting, the items are stored in mixed stacks and there is no indication of quality (or perhaps an indicator that the quality is mixed). This way, your cargo wagon, or your tool belt, can still carry just as much as before.

If you uncheck that, then everything splits into separate slots, space permitting. So if the entity is full when you uncheck it, nothing will happen until space frees up, then entities will split into various rarities.

With this system, a train can carry mixed quality items and you can sort it out at the destination station, for example. Or, if I'm carrying items in my toolbelt and I don't care about rarity, it doesn't mess up my inventory slots.

As a bonus feature, if you click to select an item from a mixed-rarity stack, it could pop a small menu showing the rarities and counts and you could select 'all' or one specific rarity.

[morsk]

It is not easy to come up with names that are fun. I ran through a thesaurus on "quality" and "excellent", and the only one I like is Superior. There are a lot of bad words out there.

My best are (Standard, Quality, Superior, High-Grade Premium, Optimized). I cheated to name the first one after quality, but why not? WoW has 2 "rarity" names + 2 "thematic" ones, about being a hero. So I tried for 2 "quality" + 2 "good engineering", but I can't think of any engineering that are fun and not awkward, except Optimal / Optimized, and used a 3rd quality one. Maybe this will inspire someone to come up with something better. I don't mind the Warcraft names. I want a Legendary power armor, not an Optimized one. Although at least Optimized Fish is funny.

If hate doesn't die down, they can always buy love by letting us /quality 4 Eldritch to nickname them at runtime. Eldritch turrets! Yes we can mod them, and that's cleaner code, but telling people to find a mod does not buy love from haters. I am suggesting bribery here. Bribery usually works.

[Tertius]

The last comment gives me a cheeky idea.
How about naming the qualities according to famous conductors. We do already have conductors in Factorio, don't we?

(empty), Toscanini, Stokowski, Karajan, Bernstein.

Notice the order by the starting letter.

[aka13]

The last comment gives me a cheeky idea.
How about naming the qualities according to famous conductors. We do already have conductors in Factorio, don't we?

(empty), Toscanini, Stokowski, Karajan, Bernstein.

Notice the order by the starting letter.

I have no idea who all those people are, I'd go with the fantasy names then

[mmmPI]

The last comment gives me a cheeky idea.
How about naming the qualities according to famous conductors. We do already have conductors in Factorio, don't we?

(empty), Toscanini, Stokowski, Karajan, Bernstein.

Notice the order by the starting letter.

TSKB ? was that the name of their car brand ? I must not have been born yet, i'm more from the era of Schumacher

[Snabrun]

Oh yes, this is what we need. I'm looking forward to it with my friends!!!

[blazespinnaker]

correct me if I am wrong, but one massive implication of Quality will be the importance of the satellite base.

For throughput you want to pack as much as you can into a train before shipping it off. One pattern that I think works well, is singular resource intermediates are always built at patches. steel, for example, packs nicely. It sounds like quality should be able to pack it in even better, assuming stacking rates don't change and higher quality science.

of course, recipes may change dramatically and mess with this, but interesting

[draslin]

I like the idea in principle, but the randomness of the quality seems contradictory to the nature of engineering. IRL, Nobody would tolerate building a car, which randomly attains the desired quality, and subsequently recycling multiple runs of it until it comes out perfect by chance.

Hilariously, that is/was the method of American car production.

I recall reading an article about American engineers going to a Japanese vehicle production plant for a tour and asking where their bone yard was. The Japanese were confused and everyone thought there was a translation issue.

It turned out that the American engineers would just toss numerous bad parts/cars off to the side. The Japanese didn't have such a situation. They said "when we produce something bad, we examine it to find out why it was bad and make corrections so that it doesn't happen again".

Fair enough, but I think you're ignoring one important component there, how many did they throw away? The numbers associated with quality in factorio suggest that at best, for every 10 you make, you'll get one you can use. Which is just idiocy in terms of engineering.

[wizcreations]

Better names for quality:
Poor
Basic
Normal
Fair
Good
Well Made
Superior
Excellent
Exceptional
Masterful

I’m sure you can use these to come up with better names for the quality levels.

[Saiph]

Hmm. I feel somewhat uneasy about this "quality" concept. Firstly, the names you've given to the different levels sound like Factorio is turning into an RPG, and I think that will spoil the feel of the game for a lot of players.

More seriously though, I think it's always a mistake to introduce more RNG into a game to add complexity. You say you've playtested quality to make sure it works, and that things remain balanced. I really hope that's true, because in my experience it only takes about five minutes from release for real-world players to break RNG-based game systems, and end up with embarrasingly crazy (and sometimes hilarious) results.

And how exactly have you QA tested quality? RNG systems are a nightmare for software QA test engineers, because combinations of RNG lead to emergent behaviour, where unexpected results are commonplace. QA engineers can't write proper test scripts and test plans for complex RNG systems, it just doesn't work. So it's not possible to properly test an RNG system.

[Mike5000]

We have made it so that quality is 'invisible' in the game until quality modules are unlocked, so you won't see anything related to quality if you haven't researched quality module or are playing the base game. This includes all the GUIs and interactions as mentioned earlier.

As long as this holds and it doesn't slow things noticeably I'm happy.

[blazespinnaker]

And how exactly have you QA tested quality? RNG systems are a nightmare for software QA test engineers, because combinations of RNG lead to emergent behaviour, where unexpected results are commonplace. QA engineers can't write proper test scripts and test plans for complex RNG systems, it just doesn't work. So it's not possible to properly test an RNG system.

Everything is already HRNG, human random number generator, in factorio already. So, randomness probably isn't an issue for testing.

[mochito]

I tried a search but didn’t manage to find if this has been asked. How will recycling support (or break) alternate recipes? For example a big feature of Space Exploration is that research can unlock more efficient recipes for certain items. Would a recycled green chip return wood, stone tablets, or neither? I could see any supported behavior here being (ab)used to convert wood to stone or vice versa.

This may be an issue that only mods have to address. Off the top of my head this only appears in the base game for uranium and oil, neither of which should necessarily be subject to recycling. I just hope that this doesn’t break support for alternative recipes, which I see as a core and interesting gameplay feature for mods such as SE

[morsk]

I tried a search but didn’t manage to find if this has been asked. How will recycling support (or break) alternate recipes? For example a big feature of Space Exploration is that research can unlock more efficient recipes for certain items. Would a recycled green chip return wood, stone tablets, or neither? I could see any supported behavior here being (ab)used to convert wood to stone or vice versa.

I saw this answered on Discord earlier:
That's up to the mod author to decide if they want a recipe to have a recycled variant. It generates the recycling variant from the recipe itself. So if you define multiple ways to craft an item it won't make multiple recycling recipes for the same item. The first will get used and or the mod author can decide which one gets used.

[PhasmaNL]

I don't like it. It's so "gamey". Doesn't fit Facorio at all imo.

I've always found Beacons gamey too. (And personally I'd prefer if they were removed from the game.) But adding this quality level is a step in the wrong direction. And while I get it must have been a big job to get all of this working/implemented, I still think it's a cheap way to add "content". Instead of actually adding more stuff, we now get programatically generated multiple versions of all existing items.

It would have been way cooler if a completely different type/category of items were introduced, with different mechanics. Like for example the Nuclear Power enrichtment process did. In a way you could say that was a quality system as well, but this was based on reality and specific to these items. A generic system is just not good. For example a new system based on Quantum Mechanics would have been way more interesting, even if it was partially (science) fiction.

I'm afraid that's it's already way too late to really get this sytem reconsidered, but still, I think it should...

[ochrid-fact]

Most important thing first, my take on quality names: miracle (that it hold together at all), crude, ok-ish, pogo, punk.

With that out of the door, I am thrilled and want to create quality refining loops until I've got my punk armor mk2.

Actually I do not even care about an improved mk2 armor. I can kill everything with mk2 armor already just fine. What I want is: I want to build those quality refining loops.

When can I buy this expansion? I'd pay 35 Euro bucks for a non-multiplayer pre-version now, and another 35 bucks once it's ready.

[quineotio]

One of the benefits of the proposed colour/naming scheme is that it's familiar. Something to consider.

[whitecold]

I'm trying to still figure out why the quality system is bothering me so much. First off, the names are bad, really bad, and make no sense at all, but that is really a superficial issue.

One thing is that the graphics seem to become confusing. Everything can have quality overlays, and that is an issue: The machine, the modules, the recipe... that is just way too much. It makes alt-view from a clear description into a confusing mess.
The quality of inserters/powerpoles or assemblers and such is something I would not want to see almost all of the time, unlike the recipes of the machines. At the same time, items on belts are "real", they are not representations of things, they are the thing, so it doesn't make sense for them to have a weird icon in the corner, as they are the item itself. All of this makes quality a visually unsatisfying system.

My biggest and largest gripe is with the recycling mechanic. I absolutely love the looping recipes of Space Exploration, and they feel right, while this does not. Those looping recipes are representing refinement processes, where raw ingredients are being step by step modified and processed. Here it feels very natural that you get side products with which you can try again and again until you got out all you can.
This is not the same with manufacturing. Once you assemble something, taking it apart again and trying over is really not going to make it any better. If this is representing picking the best parts, then you only get one try, and if that doesn't work, use new components. This would really encourage skimming good parts from your main line, instead of weird quality "distilling". Recycling should not deliver any quality items, instead giving you base at best, if not raw ingredients. That may require additional rebalance, but quality modules could have higher chances at the cost of production, to still make quality items consume plenty of resources.

[johnny7]

Before getting to the meat of my post I would like to thank the devs for creating such an interesting and engaging game. I have played over 2,000 hours of Factorio and have enjoyed every minute of it!

I've been reading the various posts regarding quality modules and would like to add my two-cents: First, the idea of incorporating the property of quality into the game should provide a more varied and interesting approach to crafting items and in managing resources and inventory. In parallel with other's comments, however, I do find the random nature of its implementation to be a bit disconcerting.

What do I perceive to be at the root of the quality-protesters objections? It's that quality modules do not produce quality products! Rather, the as-proposed quality modules effect a lottery system that sometimes spits out better stuff. In rebuttal, it has been said that the law of large numbers makes the randomness predictable. While mathematically true, such an approach to making quality products would be ruinous to any business due to the sheer scope of waste: make 10,000 widgets and 100 will be quality.

In over six decades of life experience I have observed that quality results from a confluence of design, materials, tools, procedures and an experienced workforce. While the workforce is unavailable to us on Nauvis, the other aspects of producing quality items are available or potentially available. In other words quality (or the lack thereof) is predictable, luck is random.

That being said, the idea of a Quality module is an excellent one as it is an in-game representation of better design, tooling and procedures, such as quality control, that are necessary for the reliable, repeatable production of quality items.

Allow me to outline a different take on implementing quality and quality modules:

1. Limit the number of quality levels to three. Others have touched upon the taxonomy of quality levels but, for this discussion, I will simply use "Good" (standard, no modules), "Better" and "Best".

2. The presence of a "Better" quality module allows an assembler, refinery, electric smelter, etc. to consistently make "Better" quality products. Likewise having a "Best" quality module would allow consistent production of "Best" quality items.

3. Just as in construction, where all of the workers have to use the same set of plans, you may not mix Better modules with Best modules. Further, if speed modules are present then all quality modules are disabled because, as stated if FF #375, haste makes waste. Along the same line of reasoning, quality modules will lower (significantly?) the rate of assembler output.

4. The number of quality modules in an assembler, smelter, chem plant (etc) determines the ratio of parts being produced that pass or fail quality control (QC). For example, with one quality module installed we'll have 4-of-5 items pass QC. With two quality modules installed that ratio might improve to 6-of-7 items passing QC. Items that fail QC can either be recycled to recover some of their quality components or can be used as Good (standard) grade items.

5. Because making Better-quality items requires the use of Better-quality ingredients the output quality should be limited to that of the lowest quality input ingredient. This requirement entails the existence of quality raw materials - read on!

6. Real world ore and oil patches are not created equally. Their location and the presence of impurities renders some sources more or less desirable than others. I propose that each patch of ore or oil in Factorio would provide materials of varied quality. For example, if you put down 100 miners on an ore patch, maybe 96 of them would produce "good"-grade ore, three might produce "better"-grade ore and one might produce "best"-grade ore. Likewise, in a group of several oil wells there may be one or two wells producing higher quality crude.

7. Because resource quality is determined by nature, quality modules have no affect on miners and pumpjacks.

8. Miners and pumpjacks are enhanced to provide the circuit network with an indication of the quality of the resource they are sitting upon. This information could then be used to selectively enable or disable miners and pumpjacks via combinator logic. Any miners found to be sitting atop Better or Best resources would be ideal candidates for 3-dot productivity modules.

Proposing resource patches having varying levels of quality once again raises the randomness issue. But, in real life, you don't fully know what's under the ground until you dig it up so, in the case of ores and crude, let's say we're simulating the unknown rather than randomness and call it good enough.

It would be an interesting game dynamic if smaller ore and oil patches tended to produce an appreciably higher ratio of quality materials. Such a game mechanic might encourage players to pay attention to small resource patches rather than bypass them in favor of large patches.

9. For any baseball fans out there here's a wild pitch: if items of differing quality are commingled in a chest or train wagon they then lose their enhanced quality and become "good" (standard) quality. Why, you ask? If, in real life, several high quality items were commingled with a multitude of ordinary items the resulting effort to find and separate the extraordinary from the ordinary would be prohibitive.

The risk of over-producing Better and Best quality items using my proposals would be constrained by the low density of higher quality raw materials. Such materials are, theoretically, infinite because the world map is infinite but the practicality of making everything "Best" quality would be a Herculean task.

Okay, my "two-cents" may have gotten a bit long, but thanks for reading to the end. For the non-U.S. readers the "two-cents" idiom means "here are my thoughts on the subject".