Friday Facts #375 - Quality

[epr]

HAHAHAHA, I really thought this was a joke at first.

But seriously though, this is going to be kind of fun for what? The first playthrough?
After that it's just going to be tedious and boring to deal with.

RNG is fun, right?
Gambling is exciting, right? right??

lol, no. Shan't be buying this one, chief.

[Ignotum]

The Recycler.
"Not only it solves one of the biggest problems in Factorio of how to get rid of all the pistols Trupen made." (FFF #375)
I think the recycler should decompose the starting pistol into 42 fish!

[Hamster_Furious]

I'm going to have to agree with the point that the stylization of the quality feel very mobile rpg-esque, I dont think that the color and classic "rarity" tiers mesh very well with the game's art direction, (pun intended) but the concept itself seems mostly innocent.

[Unknow0059]

Seems cool.

But I abhor the quality icons. It's overly complicated, adding to already extant visual noise.

You can communicate the same thing with one single colored notch, rather than needing to have dots.

Players can easily remember the color hierarchy.

Whether it fits the theme and tone of the game, I agree, the saturated colors are very RPG-like. If there's a Factory-themed alternative, that would be better.

[adam_bise]

Some have expressed their dislike of introducing RNG features into Factorio.

But how random does it have to be really?

I am no expert on statistics, but I know that casinos are able to precisely predict how much money they will make in relation to winnings payouts because of the way an average becomes more precise given the number of iterations.

In statistics, I believe, this is referred to as confidence intervals in standard deviations.

So then how many iteration would be performed in an "improvement setup"? Would this be a high enough number to result in a precise result prediction?

[FuryoftheStars]

Some have expressed their dislike of introducing RNG features into Factorio.

Hmm, for me I think it's a couple main issues. Everyone's MMV.

1. Between the RNG and names, it really gives it that RPG/fantasy feel, which was something that I liked not being in this game.
   Even with better names, I think the level to which they went with the RNG would still heavily give it that feel. Don't get me wrong, though, I'm not 100% against the concept of RNG in this game. I'd just rather it be in more minor capacities.
2. In some ways it feels like a cheap feature. I'm sure that some actual work had to go into the backend to support it and all, but it's just a matter of slapping some Quality modules into the machines for a chance of better output, then filter out the results we don't want and recycle.

[Koub]

I am no expert on statistics, but I know that casinos are able to precisely predict how much money they will make in relation to winnings payouts because of the way an average becomes more precise given the number of iterations.

I'm OK with playing in a casino if I'm the one who wins every time, and the ones who lose are the others. You want to be my opponent ?

Same thing with RNG : I don't mind RGNing over something I'll be producing by millions no matter what. However, I remember having to farm a dungeon back when I was into MMORPGs, because I wanted a loot with 25% drop chance, and I got my loot at the 23rd run. I'm sure that over the entirety of the runs, the average would have been 25% chance, but from a personal perspective, it was closer to a 4% chance.

I prefer a reliable consistent 1/X chance, because I get 1 every X crafts guaranteed (the same way productivity bonus works) as opposed to a RNG "it averages out over big numbers", because not everything is big numbers. I'm not expecting to craft the most endgame armour a few thousand times, just to get X% of the maximum quality. I just need one, and the only way to get the best armour will be to fight against RNG.

I have come to trust the devs about their design decisions, because over time, they proved me they designed things in a way I enjoy. So I'll wait and see if my first bad feeling about quality is just a bias of a past trauma, or if it effectively makes the game experience less enjoyable.

[adam_bise]

OK then let's look at some data

the_legend_of_gear_1236784618.png (32.64 KiB) Viewed 3875 times

Looking at the chart, it would seem that one Q3 module adds 2.5%

You can estimate the missing bottom line by noticing the chances increment by a factor of 10. So does that mean having 100% uncommon input items increases the same level output item by 90% + 10% from the modules?

If so, then the chances for one uncommon gear given one uncommon lead plate appears to be 100% (with 4 Q3 modules)

So then looking at Power Armor MK2 input items are:
Efficiency Module 2
Electric Engine Unit
Low Density Structure
Blue Chip
Speed Module 2

These items are made many times, and I am willing to bet that if you input 100% legendary items into a machine with 4x Q3 modules, the chances of a legendary Power Armor MK2 would be 100% (The missing line in the picture)

Do you think this is the case?

This is further hinted by the gif in FFF 376 The bottom left assembler seems to be actually "set" to uncommon assembling machine 3 which would require 2 uncommon assembly machine 2's and 4 uncommon speed module 1's. You can see the output is always at least uncommon in the gif - a 100% chance (missing in the picture) and 10% chance for rare.

[torne]

These items are made many times, and I am willing to bet that if you input 100% legendary items into a machine with 4x Q3 modules, the chances of a legendary Power Armor MK2 would be 100% (The missing line in the picture)

Do you think this is the case?

The original post stated "an important part of the mechanic is, that the quality of the ingredients is the base for the quality of the product" and IIRC it was explicitly confirmed in one of the threads that the output of a recipe is never lower than the lowest-quality ingredient regardless of the quality modules used, so yes, if you use entirely legendary inputs for a machine you will always get a legendary output, even if you have no quality modules in the machine at all.

[Loewchen]

These items are made many times, and I am willing to bet that if you input 100% legendary items into a machine with 4x Q3 modules, the chances of a legendary Power Armor MK2 would be 100% (The missing line in the picture)

Again, you don't need Quality modules to produce items of the same quality as your input, you only need them for a chance to IMPROVE the quality.

[adam_bise]

These items are made many times, and I am willing to bet that if you input 100% legendary items into a machine with 4x Q3 modules, the chances of a legendary Power Armor MK2 would be 100% (The missing line in the picture)

Do you think this is the case?

The original post stated "an important part of the mechanic is, that the quality of the ingredients is the base for the quality of the product" and IIRC it was explicitly confirmed in one of the threads that the output of a recipe is never lower than the lowest-quality ingredient regardless of the quality modules used, so yes, if you use entirely legendary inputs for a machine you will always get a legendary output, even if you have no quality modules in the machine at all.

Great! So the only thing random will be low level materials, which are made many, many, many times. Then everything else is legendary once the intermediates were improved to the max.

I will assume that for items made via fluids, the fluid doesn't count as a quality, and the rule that "the quality of the ingredients determines the output quality" still applies.

[mmmPI]

I prefer a reliable consistent 1/X chance, because I get 1 every X crafts guaranteed (the same way productivity bonus works) as opposed to a RNG "it averages out over big numbers", because not everything is big numbers. I'm not expecting to craft the most endgame armour a few thousand times, just to get X% of the maximum quality. I just need one, and the only way to get the best armour will be to fight against RNG.

Hey i'm not so sure you have to "fight RNG" this "only" way, you can make your armor from high quality component, and be 100% sure to get one of that same quality. The "RNG" then can be completly nullified. I mean instead of having a 25 % chance to get a quality upgrade when crafting it and try it several time which can conduct to expensive quality 4 material being wasted there would be less weight given to the "random" if instead this quality upgrading was done at the plate or circuit level and then those being assembled in an armor with 100% succes rate.

So how many times do you want to try and craft your armor ?

with 25% success chance I would say you need 8 attempts otherwise to get 90% chance to get one : at least that's what Wolfram Alpha is saying when i give it this formula : 1-((0.75)^x)=0.90 ==> ( 100 % - (( failure rate ) ^ number of attempts )) = probability of success, and 16 attempts gets you to 99% chance. So that there is 1 % chance that you feel curse and need 17 attempts or more to get your armor which is not so far away from your legendary bad luck run.

[Chrius]

I don't like it.

The reason I enjoy the game is because it's precise. I put 1 iron ore in, I get 1 iron plate in x amount of time, etc.
If we have different machines that have different speeds, that can possibly screw the ratios making it impossibly harder to balance production...

Then there's the problem of higher quality equipment...
As it is today, there's a progression to the equipment available... if I unlock something new, it most likely is better than whatever I had before. With quality added in the mix, every time I make something new I have to check if it's worth upgrading, because it might just be a shit item.

There's also the problem of being overpowered. If the base items will stay the same and quality materials will make better ones, that means one of 2 things.
Either the game will be balanced for the base items, and if you make a higher quality one you're OP, or the game is balanced for higher quality items and you are forced to make them. Either way, this means people will want better quality items, so they will have to sift through their production to find the random higher quality materials to have a better chance of making a higher quality equipment. They will then fail, because the chance is random and will have to scrap all the items they made with higher quality materials...

I'm not against the idea of having higher quality items per se. Just against the randomness.

Suggestion...

Imo quality modules should be extremely expensive to make, but *guarantee* a higher quality item. This way we can make specific production lines that will always make/use materials of a specific quality.
It should also always require quality modules to keep that production like at a higher quality. Ex: I put 1 quality module in a miner, I get uncommon ores. I put 2 quality modules, I get rare ores, etc...
Then same thing for the furnace, but it would require a base material of a certain rarity to make that type of plate. So if I had uncommon ores, the best I could get is uncommon plates, regardless of how many quality modules there are on the furnace. And if the furnace has no quality modules, or a lower amount than the ore needs, it reduces the quality of the plates (either down to whatever amount of quality modules there are, or back to normal if there are none).

Also player crafted items could always be the highest quality. The only problem is you need at least 2 quality modules (1 for the miner and 1 for a furnace), but then you can use those materials to craft better stuff yourself (before getting more upgrade modules for a complete production line).

Another thing. Quality Modules shouldn't stack. Instead of having 4 of the same quality modules in a machine, there should be 1 for uncommon items, 1 for rare, 1 for epic and 1 for legendary (or whatever name they end up with). This way uncommon modules can be made at the beginning of the game (still expensive, but with basic materials) and legendary ones can be made near endgame with much more complex items.

[Fist]

I generally just play the game and don't engage outside of that, but this is honestly dreadful to me to extent I thought this was an old april 1st joke that steam had accidentally reposted.

Quality never works as a bonus, it's always a baseline. If you don't have quality things by the very nature of it you have bad things.

Spending even any time using trash things like burner inserters already feels bad, what on earth compelled anyone to think 'what if everything felt like using burner inserters until your gacha gave you better ones' was a good idea? This isn't an ARPG, true RNG rubbish has no place here.

Better efficiency should inherently come from better designs not a lottery. Never a lottery.

You have reinvented a problem that was already solved by having actual better machines and equipment and made the solution a terrible luck based system that makes the game worse for anyone who doesn't want to engage with a shitty skinnerbox loop. This is so absolutely unfathomable and alien to me it's like the dev team were replaced by dusty mexican mummies.

This has literally unsold the expansion to me. Considering this was probably meant to be a hype update that is very impressive.

[NoriSilverrage]

Sheesh, people sure like to complain. This sounds like a fun addition. For those that don't like it, it also appears 100% optional...

I do agree that the names do not fit Factorio but that's a very minor thing.

[Maric]

for bonus content more optional than modules this sounds prefectly fine and the only thing i dont like are the names, i prefer words like precise/optimized/perfect over uncommon/epic/legendary

for the people that are sad about the inconsistency lets look at some math here:
https://stattrek.com/online-calculator/binomial
if you put in your number of crafts as the number of trials, lets say 1000 (the mk2 power armor has ~50k crafts in its chain, the depth is about 4-6 "chances" each)
the probability is 0.1 for just the 10% upgrade chance
the number of successes would be the upgrades that happen

if you put in 100, you get a reasonable 50/50 (well 51.5% chance) that you actually reach 100 or more quality upgraded results
if you put in 90, you can be 0.867% certain that you get that, so better than 5/6
for 80, its 98.6% that you can get there or better and for 70 its 99.9 and you would need some dream level luck to not get 60.
so even if in the WORST case (more unlikely than getting hit by lightning every day for the next year) you can just rely on getting 50 quality ups per 1000 crafts.
Its not that you can rely on a flat 5%, but the larger your scale the more unlikely it becomes to not hit the quality ups

for 10k crafts you can rely on getting at least 900 quality ups. so even in the worst case you have 20x "wasted" material to eliminate luck to a degree that makes it more likely that through a hilarious amount of random bitflips you just find a chest full of power armors anyway, more likely you are going to be faster than 10x resources (again, with no recycling) because on average everything upgrades by one tier 10% of the time and more than one tier sometimes. the challenge in the build is to capture the bonus speed that proper filtering for higher tier items can grant you.

[adam_bise]

Yes, designing those will be fun for sure.

At this point my biggest question is this: Will I catch a legendary fish?

[DiegoPro77]

On another similar game, called Dyson Sphere Program, I like the feature to increase the "quality" of the machine outputs (speed or production) with spryers.

Let me explain. In this game you have conveyor belts similar to the factorio ones. On these belts you can put spyers that, guess what, spry what are called "proliferators" on the items of the belt.
Proliferators have some features that affect fuel durability or some other parameters, if we talk about fuels (there are no weapons in this game, it's a complete sandbox at the moment) but, if all the recipe items of a determinate item are proliferated at the same level (there are 3 tiers of proliferators) you unlock special bonuses on that recipe, as faster speed in the production of the item itself or as extra production, similarly to factorio modules.

I liked this mechanic in DSP.

[thriem]

I do not like this "quality" aspect either. For some aspects it might be ok, but it produces a lot of overhead just to get rid of things.
If it has to be that way, I'd prefer to have a chance, like the Uran enrichment - you set the desired quality (and if it is allowed to be better ??) and the assembler shows you a chances of the output - upon fail, only % of resources are used - or yielded back (since it is consumed upon crafting, not result)?

Much better I think would be upgrades, similar we have already with Modules-Upgrader and other facilities. But, instead of a replacement, entities use certain materials to upgrade.
So you dont toss away 50 spidertrons just to get the lucky roll… but, depending on the recipe, have something yet work towards to.
Even better, technically enemies can be crucial for battle-upgrades, like auto-targeting needing sample entities to "train AI / computer recognition" or whatever. So, a new evolvement or tier of enemies pose much more of a threat, if only a fraction of turrets is shooting at them because there targeting system malfunction - whereas the enemies do not need much of a buff themselves and can be still be handled reasonably manually - after enough samples have been gathered and systems are "recalibrated".
For other entities, it kinda feels a bit pointless to me. Mean, nice and good - but a research could fix it too - and if it needs to scale with industry, how about researching an "module slot" or whatever, so you still have to craft for each entity.
When it comes to gears, circuits and other components… I'd rather have a % output chance and nothing upon fail. Waste management just is not a thing everyone will be into (like Circuits Networks or proper Train management) - or have a "QA" entity, where I can feed into, and has 2 outputs for "Pass" and "Fail" - so someone can rather simply filter stuff - sure, you can do this via belts and splitter-filter, but I see a lot of conveyor-belt-knots incoming for any base that wants to be tight just to get the waste out for proper management - a whole entity "reserves" the space while maintaining a bit of readability of the belt-flows.

[stoicfaux]

First thing I did was check whether the Friday Facts was posted on April 1st.

As someone who crafted the upper tier modules in Space Exploration and fell into the trap of increasing base productivity just to create better productivity modules, I'm less than enthused by 'quality' items. (The bureaucracy expanded to meet the needs of the expanding bureaucracy to become more efficient.)

Most importantly, why aren't there any quality belts/undergrounds/splitters?