Friday Facts #409 - Diminishing beacons

[morsk]

But there is a way to prevent that! All that needs to be done is increase the base beacon energy consumption. Double it! (maybe make it scale with the number of machines serviced). By doubling or tripling the base beacon energy consumption, the optimal count of beacon source per machine will become closer to 5:x / 7:x, rather than 8:8.

People already make solar fields 10x the size of their bases, in order to power 12-beacon. (Literally 10x, not a joke.) There is no limit to what they'll do to reduce active entity count. As much as I hate their bases, we should have mercy on them and not make the bases even worse with these solar fields.

[MeduSalem]

To be honest I don't know how to feel about the change. ^^

But mostly that is because it changes nothing about the meta. People will still do 12b/1m setups... or alternating rows of beacons & machines. Simply because that is the best you can do with the range Beacons have.

There are only 2 things that really change...

1) Quality Beacons make the setup even better now because less transmission loss.
2) The diminishing return calculation is more annoying and less straight forward than the linear one.

The latter doesn't matter in practice however because as a rule of thumb you will still aim to have as many beacons as possible unless space is a concern (like on platforms).

[Disie]

This is a nice change in the right direction and I really appreciate the devs taking their time on this before 2.0 release. However, as somebody with over 10k hours spent megabasing I do not see how this will change anything in the endgame.

We will still end up putting beacons in every available space possible, direct insert builds will get a slight buff but nothing substantial and now the megabasing is also locked behind tens of thousands of legendary quality beacons along with double the number of LV3 legendary speed modules and who knows how many prod modules. There are no disadvantages to going with more beacons late game, power is free, space is virtually infinite, there will never be a scenario where 6 beacon build is better than a 7 beacon build even if you consider power or cost. It doesn't matter if I need to make 10k or 100k legendary quality beacons, it will still be more beneficial to spam as many beacons as possible, except now I will need to spend hundreds of hours grinding to get all these legendary components to make a ups efficient megabase.

I think that beacon overload is actually an amazing idea IF done differently to what Space Ex does and what's been outlined in this FFF.

Instead of having the machines overload if hit by more than 1 beacon why not have them overload if hit by n% of transmission effect. So for example in 1.1 a beacon has 50% transmission effect, a maximum number of beacons that 1 machine can have is 12 so in this case it would be 600% transmission.
Why not take that and have anything above 600% overload the machines? If you went with buffing the beacons to have 100% transmission effect in 2.0 that would mean one machine can accept up to 6 regular beacons and if it's hit by 7 it will stop working.

This would make quality beacons optional, incentivise players to craft them eventually to cut down on the amount needed to hit desired speeds but also not make it overly grindy as we wouldn't need that many of them. Investing in quality beacons would mean the factory using less power, less speed modules and having more freedom in build layouts as less beacons would be needed compared to non-quality beacons. This would also mean that layouts would need to be carefully planned to have every machine under 600% transmission effect instead of spamming beacons in every available space like we do now and like we will continue doing in 2.0 if this goes live as is.
For example if the regular beacon gives 100% transmission effect we could have the legendary go up to 200% meaning now we can only have 3 beacons on each machine and it would be an interesting challenge to design builds around this principle. It wouldn't be as boring as putting down 1 beacon and surrounding it with machines or vice versa and at the same time players would be incentivised to make quality beacons but it would not be required to make a megabase.

I would be interested to find if this idea or similar was already considered and if there are any major drawbacks to this idea that I'm missing that makes the current beacons diminishing returns idea superior.

[Shingen]

So, uh, 1 minor nitpick @boskid:

this is quite clearly a list of sqrt(n)/n, so the new efficiency multiplier of each beacon affecting given building, but the values seem truncated instead of being rounded.
for example,
sqrt(3)/3 = 0.57735..., should round up to 0.5774 (both with classic rounding up from 5, and half to even/statistician's/whatever rounding), but it's 0.5773 there
sqrt(7)/7 = 0.37796..., should round up to 0.3780 but it's 0.3779 there
i mean, maybe it's like that on purpose, but i thought i'd mention that

[Alsadius]

I like to build my end-game design to optimize the energy cost per craft (assuming maximum productivity modules). In 1.1, that means rows of beacons and machines (1:8 and 8:1), which is boring.

The diminishing ratios sounds great. But I'm afraid that the most efficient designs will still be rows of beacons in endgame.

But there is a way to prevent that! All that needs to be done is increase the base beacon energy consumption. Double it! (maybe make it scale with the number of machines serviced). By doubling or tripling the base beacon energy consumption, the optimal count of beacon source per machine will become closer to 5:x / 7:x, rather than 8:8.

I doubt they'll add that, but this would be a really trivial mod to make. And probably one worth installing if beacon-maxing bores you, tbh. (I might even go to 3x or 4x, to be honest. Keep them for where they really matter, drop the spam.)

What if we put 12 beacons and load 6 of them with speed and 6 of them with efficiency, do we get 2.45 of each effect or 3.46 of total?

Seconding this question.

[gGeorg]

the need for more clear presentation of throughput numbers becomes even more paramount, and I did not feel that concern was addressed.

Agree 100% with this. You boys realize that you have now effectively required the use of online calculators for mere mortals to get accurate throughput numbers? Since calculators are now a requirement, are you including one in-game?

Smooth curves in math model are a pleasure for a math freak, however, for players it could be pain. Even linear dependency is brain teaser for me. Build-in calculator or even easy accesible edditor-designer-blueprintProject surface would be nice QoL.

[jurjen909]

Loved the Friday Fact of this week, but It was a bit difficult reading the graph when they only showed numbers and no identifying titles and units. You guys probably know what each thing is, but just looking at the graph without any knowledge of what exactly was being tested and shown was a bit rough. In the end I got that the x-axis was the amount of beacons used, and the y-axis the Beacon Effect (multiplier).
Perhaps I am too used to making and reading graphs that are used in scientific papers that have a requirement of being able to be readable without prior knowledge or reading the paper itself.

[Mooncat]

For those who missed some words in the FFF and wonder why their implementation doesn't fix the problem of maxing beacons, like me,

In their conclusion,

In 2.0, maximizing beacon count will still generally be a good practice to have things produce quickly and save on computer performance. However, there will now be a lot more room for creativity, and with quality return on investment comes into question more.

Apparently they've already known it (ofc, right?), but they don't have intention of completely getting rid of those layouts. People who love to max beacons are still gonna max them.

But now I wonder, will the equation be moddable so that we will have mods that don't allow beacon stacking at all.

[maniak1349]

No more manual calculations for me I guess.

[kitters]

Nothing beats the fact I would have to use tables or calculator instead of simple multiplication.
And it almost doesn't change 8 and 12 beacons design (10% efficiency - nothing).
You want to buff earlygame and late quality game? Okey, let it be instead of E=0.5*n to E=1.2+0.3*n.
Linear. *3 efficiency on 1 beacon, -10% efficiency on 8 beacons, -20% on 12 beacons. Buff for higher qualities as you see fit.
Easy to calculate. Nothing diminishes, just a straight bonus for "first connection" to buff early game and motivate to use it early or for good use of green modules.

[Alsadius]

For those who missed some words in the FFF and wonder why their implementation doesn't fix the problem of maxing beacons, like me,

In their conclusion,

In 2.0, maximizing beacon count will still generally be a good practice to have things produce quickly and save on computer performance. However, there will now be a lot more room for creativity, and with quality return on investment comes into question more.

Apparently they've already known it (ofc, right?), but they don't have intention of completely getting rid of those layouts. People who love to max beacons are still gonna max them.

But now I wonder, will the equation be moddable so that we will have mods that don't allow beacon stacking at all.

Yes, that's what the Lua list at the bottom of the blog post was about. If you want to get rid of stacking, you just change those numbers to {1,0,0,0,0,...}, and only the first beacon will have any effect.

[bman212121]

For the people who want the "best" solution to change, nothing can really fix that. Because the game is in a grid layout and it's a calculated value, there will ALWAYS be one best answer. So whatever that answer is, that's what the optimizers will do when they solve the equation. If you don't like that, then don't put that restriction on yourself where you have to do it because it's the best solution.

This change is one of those it's about the journey not the destination. If you think about the actual journey from start -> end there is a lot more variation in setups that can be done, because the scaling has changed making it more enticing to do things that are "good enough" to get from A to B. Quality definitely adds even more to the variation and maybe a quality boost gets a factory enough production to overrun your input, so you don't need to build another beacon. I'm definitely looking forward to seeing how many different ways you can put down 1 or 2 beacons and use them as temporary boosts with good results.

[Mooncat]

For those who missed some words in the FFF and wonder why their implementation doesn't fix the problem of maxing beacons, like me,

In their conclusion,

In 2.0, maximizing beacon count will still generally be a good practice to have things produce quickly and save on computer performance. However, there will now be a lot more room for creativity, and with quality return on investment comes into question more.

Apparently they've already known it (ofc, right?), but they don't have intention of completely getting rid of those layouts. People who love to max beacons are still gonna max them.

But now I wonder, will the equation be moddable so that we will have mods that don't allow beacon stacking at all.

Yes, that's what the Lua list at the bottom of the blog post was about. If you want to get rid of stacking, you just change those numbers to {1,0,0,0,0,...}, and only the first beacon will have any effect.

Oh that's great. I have missed another paragraph again.
I blame me not having enough sleep.

[Agamemnon]

Hottake: If a player is so far into Factory optimisation as to spam beacons and obsessing over exact production statistics, they can be expected to use a calculator.

Getting powercreep under control is an important design goal, so yeah - beacon changes are good.

Only one thing stated in the blog post I would strongly disagree with:
The SE Beacon Overload mechanic was extremely satisfying to design my factory around. Finding a way to route all (up to 8, solid or liquid) inputs and outputs to and from a beaconflower with 12 machine-petals was a fun puzzle. Or rather a bouquet of puzzles. (but mainly because I installed the adjustable inserters mod as well)
Same for coming up with good designs for 8x5² buildings around a single beacon.
I was getting a lot of mileage out of just 1 small beacon, too - way more than it's building cost would suggest - so I just pretended they were super expensive to justify my efforts of designing everything so maximally compact.

TL:DR I had more fun emergent gameplay with Beacon-overload mechanics (+ adjustable inserters) than with copying long-ago solved designs for spamming regular beacons in a line.

[bman212121]

Here it is! (beep boop)

https://factorio.com/blog/post/fff-409

Really like the diminishing returns approach, but I wonder if there could be some further interest added by having a proximity bonus to beacons so that a beacon directly next to a building has a greater effect than one further away. This could make beacon designs really interesting as that high proximity space is already highly contested for inserters etc. but having this as an added factor also further complicates any calculations.
Would be interested to consider though.
Overall, really loving these updates though and appreciate the effort you guys are putting in!

I like your proximity bonus idea. It would likely cause more variations for non perfect builds. For people doing beacons around a single factory it's unlikely to change anything, but that's fine.

My only other thought of how to add variation into the game would be to make the beacon output a static number, and then divide it by the number of entities that it affects. IE if you have a value of 100, and it touches 4 factories, each factory gets 25. If you have 7 factories it touches, then it gets 100 / 7 = ~14.29 output. I don't think that would make any difference for builds which rely on single factory outputs, but it would change things up in many other cases. The biggest complaint I can see is that it annoys the people who need to calculate every single factory's output by hand because each factory would probably have a different number. But for those who enjoy the puzzle you'd be drawing out configurations and having to calculate different balances based upon belt throughput where you might have unbalanced designs due to belt limitations.

It would be neat to see a mod add your idea and mine, to make it to the point where you just play the game and not worry about calculating it because it's too complex to care!

[Sliverious]

My reaction to these updates have become a trend at this point. For the time I spent playing factorio, I am more than willing to buy the expansion when it comes out, and probably will. However, I doubt I will ever play with any of the new features enabled. Maybe it is a case of getting used to these things, but I really feel like modding out almost everything we've seen in FFFs so far. To be honest I don't know how I feel about that sentiment. Does anybody else feel the same?

EDIT: to clarify, I don't really see any point in most of these changes. The early game will just be "suffer through with yellow belts and whatever builds produce enough to progress science", and after that you produce some scaled up builds that will help you create a base so big that you don't care about quality or power consumption. The main challenge of the expansion seems to be how fast of a machine gun rocket launcher you can build to get resources from one planet to another.

[argbla]

Perhaps I missed it, but one of the concerns about meddling with the beacons was whether the user would be able to calculate the benefits quickly in their head. I am not saying head calculus should be a showstopper, but the need for more clear presentation of throughput numbers becomes even more paramount, and I did not feel that concern was addressed.

Agree 100% with this. You boys realize that you have now effectively required the use of online calculators for mere mortals to get accurate throughput numbers? Since calculators are now a requirement, are you including one in-game?

Calculators? What are those? Machine go fast, engineer happy.

[Alsadius]

My reaction to these updates have become a trend at this point. For the time I spent playing factorio, I am more than willing to buy the expansion when it comes out, and probably will. However, I doubt I will ever play with any of the new features enabled. Maybe it is a case of getting used to these things, but I really feel like modding out almost everything we've seen in FFFs so far. To be honest I don't know how I feel about that sentiment. Does anybody else feel the same?

Personally, I'm pretty excited for the quality of life changes (smarter bots, better combinator UI, improve map generation, etc.), and the new planets look really cool so far. I expect quality and improved throughput to be a major focus when actually playing, and my monkey brain does love making numbers bigger, but they'll likely be a bit less interesting in terms of giving us new challenges. This beacon change I quite like overall, though - it feels like a deep vein of new reasons to tweak builds, much more than it might look like.

But I think the part I'm most looking forward to is having a "campaign" feel to it, where there's a large set of related challenges before you get to the end, without it being the Godzilla-sized monster that is something like Space Exploration. I keep thinking I'd kind of like to try SE, but I'd probably finish a Space Age run first, even when I can't start it for like six months yet. Plus, an official product is almost always going to be smoother and better-balanced than a mod, even a really good mod like that one.

So yeah, I'm excited for SA. My only big issue is making sure I'm not too burned out on Factorio to be interested by the time it comes out (since I tend to play games in fits and starts). But even then, I'll probably enjoy it a lot in 2027 or something.

[CXZman]

I second the opinions here that this "Diminishing Beacon" update kinda changes nothing.

(TL;DR at the bottom)

The problems to be solved, as stated by the devs, are:

1. All beacon layouts end up looking the same
   Nothing has changed here. Diminishing return never actually makes the gain plateau or get worse if you add more beacons. More beacons is still better than less beacons. It already was a question players had to ask themselves: whether they want to maximize everything or just content themselves with an easy design that just more or less do the trick. Those who just want a quick and dirty layout could already do it and will still be able to do it. On the other hand, maximizers will always maximize (hence their name) and sooo... they will maximize. Nothing has changed for them and once a good layout will emerge, it's going to get shared and implemented everywhere. Back to square 1 on this issue.
2. It's just going to look like a big grid of stamped factories
   Yes. Because that's what the tools of the game implicitly tells us to do anyway. We build on square grids, rotation of building is by 90° increments, buildings are mostly 3x3 or 2x2 (mostly 3x3), modularity works, most buildings don't give a damn about geolocalization (it doesn't matter WHERE the building is, it always work the same) or range (notable exceptions would be fluids and heat)... all of which means we're going to copy paste blueprints all over the place. Making Beacon effect have a diminishing return changes nothing of those parameters. We will keep on copy-pasting our layouts just the same.
3. It's hard to figure out the final output of factories, so we need online calculators to balance production
   This one just got worse (if it is actually seen as a problem. I do, but that's me). Mental algebra that players needed to make used to be linear. Now it includes some square roots stuff. We are pretty much pushed back into "winging it" or going online to use an actual calculator. There is no middle ground anymore of actually trying the calculation by hand. I might be wrong on this one, please raise your hand in the audience if you enjoy calculation square roots of decimal values for fun on paper.

Of all stated problems, none seem to be solved with the update (the calculation issue seems to have worsen).

For starters, I don't even understand what's the issue with beacons working as they do right now. I'm trying to accept it for the sake of the discussion but if the issue is that "it's always looking the same", then it's not a Beacon problem, it's a Factorio "problem". Considering optimal layouts for factories, solar arrays, splitter arrays, malls, there will ALWAYS be a maximum to be reached, an ideal layout given some parameters. It's way too discrete and deterministic for a global maximum not to exist. What I mean here is that a global maximum would translate as the "it all looks the same" problem, meaning it all looks the same because it is "the correct solution" to a given production challenge in the game.

The only way I can think of (please show me how I'm wrong on this) to help this issue is to make it so there are local maximas that are similar in "height" and also different enough. Then, choosing which maxima to use will depend on other parameters, probably just taste or mood, maybe other contextual parameters... but even then, you will still get similar factories, stamped out layouts and grid-looking megabases. Slightly more variability (as much as there are local maximas) but once a given layout has been chosen by the player, they are SURELY going to stamp it everywhere.

If landscaping wasn't a thing in Factorio, then terrain would be your solution. Because of its randomness, it would always force players to improvise specific and local solutions. Maybe those solutions would be based on pre-designed "best layouts" (= maximums) but you would have to adapt them anyway. Since we can heavily change the terrain to suit our needs, it eventually doesn't really matter and one can always just re-use whatever they've designed from a previous game, or from anyone playing the game.

Another solution to mix it up would have been to make all buildings different, and make it so that Assemblers are not always the same shape depending on what they're going to build... and now it's a whole other game. I guess that's partly why the new factory buildings are going to be different in those aspects (larger buildings), but you can still stamp them out cookie-cutter style once you have a good modular layout.
So maybe make it so Quality actually affects the shape of the building of the location of the input/output (for fluids)? So you can level up the Quality of your layout, but you would need to do so re-designing at each step, and now it becomes complicated to always have "the best" layout. Finding the "one global maximum" is more complex.
... yeah but once you've found it, you can copy paste it and send it online for others to grab and stamp it everywhere. It makes blueprint books grow significantly larger, but at a given quality stage, base will still all look the same.

Because it's Factorio. It's what the game does anyway.

New beacons do not solve stated problems in my humble opinion. Stated problems are a result of all of Factorio's gamedesign. Fighting these problems feels like fighting the fundamentals of the game themselves.

[Sliverious]

But I think the part I'm most looking forward to is having a "campaign" feel to it, where there's a large set of related challenges before you get to the end, without it being the Godzilla-sized monster that is something like Space Exploration. I keep thinking I'd kind of like to try SE, but I'd probably finish a Space Age run first, even when I can't start it for like six months yet. Plus, an official product is almost always going to be smoother and better-balanced than a mod, even a really good mod like that one.

This might be one of the main reasons I am not excited about these changes. I don't play a lot of games with campaigns or story modes, and whenever I do, I tend to press whatever button will make me skip it. To me, factorio is a puzzle about logistics, not so much "space adventures".