Friday Facts #224 - Bots versus belts

[Zavian]

I once tried to setup a smelter array with bots. It turned out to be less effective as if i just had build it with belts, no matter how many bots and ports i threw it it.
As ive seen in others peoples bases, bots, for a high end efficient base, actally take alot of thought too.

I think this is part of the problem that Twinsen is trying to talk about. There are people who occasionally show up in the support forums who get to bots, and seem to think "Bots are the new hotness. All my new production is going to be bot based". So then they build an assembler with a requester and provider. Make that a blueprint and stamp it down repeatedly. No thought or planning to it, just add more copies of this same blueprint anytime they want more production of something. To me that's a really boring way to play the game. They occasionally show up in the support forums complaining about bots not recharging or wanting smarter bot charging/pathing etc. They have missed the point about bots being just another tool, made bots their only tool, and haven't understood the way they charge and path. Some of them show-up asking for help and advise, some complaining that the existing bot mechanics are broken, none of them understanding that the root cause of their problems is in the design (or lack of design) of their base. It's probably impossible to know how many casual players end up getting to robots and then playing this way, but it's likely that only some of them show up in the forums.


The best idea I can think of is to give the players another belt type, with new mechanics, that is also unlocked around the same time, and says "Here's a new toy with new capabilities you should be using as well".

[Avezo]

Kinda funny how this is belts vs bots this week and conclusion seems to be - bots are fine, belts need buffs and week earlier it was pipes vs bots and conclusion was similar, that pipes need buffs (although bots were not neccesarily fine).

[bill1970]

A lot of the arguments against bots seem to revolve around them being overpowered compared to belts. Couldn't that argument also be applied to trains? In theory you could build a base without trains by building progressively longer belts to bring in resources. If you think of it that way, trains are also "cheating".

To me, belts are overpowered, because they literally require no power. They "magically" run without needing energy.

Leave my bots alone!

Bill

[leftofzen]

Construction bots need to be available earlier in the game, and logistics bots need to be available later in the game. While I have 10k logistics bots flying around my current duo base, belts are certainly still being assembled and used; they will never stop being useful. For me the "bot-only" base is kind of a cool idea but doesn't appeal to me, and I will never build a "bot-only" base. The key thought here though is that you need to let the player decide how they build their base. You can't say "I think this is OP so you can't have it anymore" because that isn't how everyone views logistics bots. Bots however are a big part of the game so I think they need to be 'fleshed' out some more. I'll detail below.

It's also interesting that a lot of people think logistics bots are OP...I think that in my end-game megabase they are UP and I've got a few mods on that give research to increase their speed and carrying capacity.

So my suggestions, in the spirit of the FFF, would be to:
* Make construction bots available earlier in the game and make logistics bots available later in the game. Do this by allowing research into flight earlier and then having a 'logistics bot' research that is costly/requires a few more prerequisites.
* Make roboport logistics zones smaller initially, and have research that increases this. This will ease the player into logistics bots and allow progression at the cost of time and resources. Same for construction zone size.
* Allow research to increase the battery capacity of robots. It's fine to keep this low when you first get bots, but having them have to recharge every few seconds in an end-game megabase is ridiculous.
* Add boxing/palletting, and don't let bots carry boxes/pallets. I found at least 3 mods that do this already. It will allow you to increase belt throughput substantially (making belts more powerful) at the cost of slightly more complexity (machines to pack/unpack items).
* Add more variations of inserters, such as 90-degree inserters. Again this implicitly makes belts more powerful.

Secondary/additional ideas include:

* Add bot 'zones', akin to the staff zones in RCT/RCT2. These are user-defined zones in which a certain group of user-defined bots is restricted to. While this makes bots more powerful in the hands of an experienced player, the additional complexity could provide a ramp up from the factories where bots fly around everywhere. In other words you get to 'optimse' bot usage.
* Making belts require a small amount of energy. I never liked the fact that they are free.

[RMJ]

Kinda funny how this is belts vs bots this week and conclusion seems to be - bots are fine, belts need buffs and week earlier it was pipes vs bots and conclusion was similar, that pipes need buffs (although bots were not neccesarily fine).

The problem is bots isn't fine. If they buff belts, the trains will be even more inferior than they already are, which is a crying shame. Because Trains are so beautiful animated and designed in this game, its so awesome to just see them doing their thing.

Bots are too good at everything.. Either they should be master of x, or jack of all trades master of none.

[auroq]

tl;dr skip to the last two paragraphs

I have always felt that bots have an inherent disadvantage over belts in their difficulties with moving large groups of items across long distances. People have stated that this can be "solved" by just making more bots, but that doesn't truly solve the problem. It just causes you to waste a lot of resources and use a lot more power.

This game to me is all about efficiency. Belts and bots and trains for that matter all have distinct advantages and disadvantages. Bots should not replace belts, and belts should not replace bots. Bots should be used for short distances or for a one time resource transfer. In my experience anything else just doesn't work when you consider the ultimate goal of efficient automation.

However, it's far too easy to fall into the trap of building "The Megabase." As a developer I see this issue a lot. (Those who are unfamiliar with software development, please bear with me.) As a small startup dev team it is easy and often logical to put all of your software/services into a single "Big Build" process. But in the long run as the codebase, services, platform, etc... grow it makes much more sense logically to split everything into microservices, microbuilds, small individually deploy able pieces. I try to play factorio the same way. Rather than building one big Megabase with 10000000 untrackable robots flying around in circles, it is more efficient and logical to build several smaller individually purposed bases that connect (like well organized software dependency trees).


tl;dr- skip to this part
My proposal is rather than nerfing, buffing, doing away with, etc... I would prevent logistics networks from becoming too large and unwieldy. More specifically, don't allow the player to connect infinite number of roboports. That is what leads to massive uncontrollable networks. Set a size limit for logistics networks. This can either be done by limiting the number of roboports that can be connected by a network. Or it could just set explicit size limits, so for example, whether I have 2 roboports or 50 in this network, the network still won't span beyond 1000x1000 (or whatever size makes sense). If I need to transport items between networks, I must rely on belts, trains, inserters, etc...

I feel that this not only solves the problem of bots being overpowered, but it also creates an inherent interdependency between belts, bots, and trains.

[golfmiketango]

Sure, we can do the "always assume the worst"-game and panic at every thought experiment, but then the developers won't talk freely with us anymore but assume the worst as well. That would be the end of open dialog.

+1. However, I think to some extent this is to underestimate how hardened the core team at Wube are to situations like this.

I'm pretty sure when they posted this FFF they knew full well that it would be prone to misinterpretation and hyperbolic, emotional responses, but opted to go ahead and publish anyhow, hoping that some plurality of readers would have useful insights or just enjoy hearing the developers' thoughts. I think it's unlikely they will lose a ton of sleep over anyone in this thread that either did not read carefully, or read between the lines and got the wrong idea (although I confess this is simply my own attempt to divine what is in the minds of the authors).

[bobucles]

Bots aren't the end all ultimate logistic option on their own. They have to work together with TRAINS to reveal their true power. Keep in mind that players spent months if not years refining this system, mastering the use of trains and trying just about every other game option before agreeing on the modular bot base as the ultimate base design.

An ordinary player doesn't reach to those extremes. The casual logistic system is basically a single blob that covers a huge base. That kind of blob is the worst case scenario for bots as they lose efficiency with bigger networks. A single bot in a tiny system can easily satisfy 10 requests in the same time it takes that same bot to do one thing in a huge system. That "smaller is sweeter" factor is what makes the modular bot base so effective, while the same number of bots in a super blob base won't be able to keep up with extreme endgame demands.

Unfortunately you can't force players to build gigantic doom networks where 90% of the flight time is wasted effort. Those doom networks are the only time where you can't really solve logistic problems by laying in more bots. It simply gets too crowded and the low efficiency can't keep up.

[krail9]

Why is so much opposition to a limitation or soft cap on the maximum throughput of a bot network? It makes sense to me, that since bots are "easier" and more flexible than belts, shouldn't they have lower maximum throughput? All your bot-based designs would still work, it could just be 5% slower than an ideal setup using belts. And if you want to do crazy botswarms you could loosen or remove the cap easily with a mod?

A lot of the arguments against bots seem to revolve around them being overpowered compared to belts. Couldn't that argument also be applied to trains? In theory you could build a base without trains by building progressively longer belts to bring in resources. If you think of it that way, trains are also "cheating".

To me, belts are overpowered, because they literally require no power. They "magically" run without needing energy.

Leave my bots alone!

Bill

Trains are actually a great example of balance because despite having massively increased throughput compared to belts, they have their own pretty hefty limitations that keep belts relevant. The proof is there - high end bases are very often 100% bots with few or no belts, but 100% train bases with no belts are a novelty at best.

ATM:

Code: Select all

TYPE        USAGE                                                THROUGHPUT
Trains      Bulky, clunky mechanics                              High
Belts       Flexible but requires a fair bit of space/planning   Medium
Bots        Little/no real estate, ultimate flexibility          Medium-High??*

*Shouldn't that be low? Don't remove them or nerf to the ground, I just think there's a fair point that bots are too dominant.

[zebediah49]

Why is so much opposition to a limitation or soft cap on the maximum throughput of a bot network? It makes sense to me, that since bots are "easier" and more flexible than belts, shouldn't they have lower maximum throughput? All your bot-based designs would still work, it could just be 5% slower than an ideal setup using belts. And if you want to do crazy botswarms you could loosen or remove the cap easily with a mod?

IMO the biggest problem with that is it's a practical impossibility, as long as bots scale linearly. Either you make them even harder to get initially, you put some kind of weirdly artificial scaling cap in, or endgame you can just build as many bots as you have to. Forcing players to build more robots to accomplish the same thing is just a waste of time and UPS.

Trains are actually a great example of balance because despite having massively increased throughput compared to belts, they have their own pretty hefty limitations that keep belts relevant. The proof is there - high end bases are very often 100% bots with few or no belts, but 100% train bases with no belts are a novelty at best.

ATM:

Code: Select all

TYPE        USAGE                                                THROUGHPUT
Trains      Bulky, clunky mechanics                              High
Belts       Flexible but requires a fair bit of space/planning   Medium
Bots        Little/no real estate, ultimate flexibility          Medium-High??*

*Shouldn't that be low? Don't remove them or nerf to the ground, I just think there's a fair point that bots are too dominant.

Since bots can be packed in arbitrarily well, lowing throughput is sort-of a fool's errand. Unless you did something like impose a bots/second chest access rate limitation, which I suppose could put a solid cap on bot throughput.

I say the second column is the better place to correct. Make the logistics system take a bit more space (say, via 3x3 requesters), and it will return to the same kind of footing as the rest. You would gain flexibility, and, with investment, throughput, compared to belts, but it would take a little bit more real estate than belts. Ideally it would be balanced such that for ultra optimized designs, belts would be the most compact option.

[Enduinn]

Assuming construction and player QoL help remains the same:
If there is a way to make setting up logistic robots a logistical problem like for belts, they could remain interesting. The problem is that the game figures out everything for you. Every bot can help with every move everywhere (unless you limit it); there is no logistical problem to solve to use them, unlike belts.

Trains are different from belts because they transfer items long distances but are terrible/useless for maneuvering items. Bots transport similar distances but maneuver items far better. They need some clear distinction in purpose.

However, watching swarms of bots running around automating things is very satisfying, so there is definitely some gameplay value. My "vote" would be:
First, require logistical challenge to using them.
If not that, give them some distinct purpose that does not make belts obsolete (like trains, and bot-player QoL help).
If not that, make them VERY late game. Once you've almost beaten the game already, you obtain this extreme power that lets you take things to another level.

[Zavian]

Since bots can be packed in arbitrarily well, lowing throughput is sort-of a fool's errand. Unless you did something like impose a bots/second chest access rate limitation, which I suppose could put a solid cap on bot throughput.

If you don't touch the size of logistics chests, you can actually fit 6 chests into an 8x8 beaconed build. That could be 5 input chest and an output chests. If 6 chests aren't enough player could switch to a 12 beacon 12 chest build with a craft speed of 8, so you would need to nerf chest throughput pretty hard to actually achieve anything this way.

I say the second column is the better place to correct. Make the logistics system take a bit more space (say, via 3x3 requesters), and it will return to the same kind of footing as the rest. You would gain flexibility, and, with investment, throughput, compared to belts, but it would take a little bit more real estate than belts. Ideally it would be balanced such that for ultra optimized designs, belts would be the most compact option.

If they want to stop the 8x8 bot based builds this is potentially interesting, as even a 2x2 chest takes more room than you typically have in those builds. (It also breaks the 12 beacon per assembler builds). But personally I think Twinsen is more worried about players who switch all their new production to using an assembler with requester and provider chests, and just stamps those down mindlessly, every time the need more of something. And I agree that scaling production that way is a boring way to play. Neither of these solutions really does anything to stop that. (There is also the player who just mindlessly stamps down assembler + 2 chests surrounded by 8 or 12 beacons. You could force them to change their blueprints, but even with chests 4x4 they could still fit 10 beacons per assembler).

[devaking]

im with universe two.

if bots where only a thing of speculation, my suggestion would be:

1- to have a limit of bots in each roboports, lets say one per empty space in current UI.
2- active bots air space cannot overlap one another/ they will not respond outside there own air space
3- station could be put in 3 mode
first one: each bot acts like a separate filter inserter.
second one: the roboport is one big filter inserter
third one: repairing/building blueprint if sufficient materials are available in airspace.

of course those modes could be change using combinators to switch between them depending on the demand of your base.

4-having a robo-wagon-port that can only repair ( or build blueprint if sufficient materials are available in the train wagons)

[GenBOOM]

100% train bases with no belts are a novelty at best.

you literally only need belts to get the resources from the electric mining drills onto the train. thats it.
technically you can go 100% no belts by mining dircetly into chests and then inserting that into the train but it would limit your mining output due to spacing but it requires more power, though it would cost less to setup than blue belts

you take ores and smelt them directly from the train cargo and place the plates back into that cargo wagon.
then you have that same train stop at a station for shuffling resources so that you get the correct amount
then you have that train stop at a station for assembly, load the end item and then have it stop at a storage location where things are organized
this can all be done with circuits and combinators and its actually not that hard but without bots the setup time is even higher than belts because transporting resources long distances just to build something like this means this will take hours longer before everything is online because you will constantly need some small item that you now have to hand deliver

[AndrewIRL]

Power Creep

There's a lot of discussion (I didn't read all of it) saying thing like:
Nerfs suck, only buff things
Devs are trying to lock in a "correct" way of playing

When trying to decide whether a feature needs to be buffed or nerfed the most important consideration is how it fits within the overall game. No nerf or buff is inherently good or bad. Any change which moves something closer to the power curve is the correct one, as it improves game balance. It is easy to mod your way into a completely unbalanced and boring game. The devs have to make more careful evaluations and think of overall game balance.

Think about the power curve and power creep.

I want to talk about the more interesting type of power creep, the accidental kind which inevitably happens as designers work on increasingly complex sets of interlocking pieces.

EXTRA CREDITS S11 • E8
Power Creep in Hearthstone - What It Teaches Us About Games - Extra Credits

Power creep.png (434.74 KiB) Viewed 7988 times

Youtube: Power Creep in Hearthstone - What It Teaches Us About Games - Extra Credits[/url]

[agmike]

When trying to decide whether a feature needs to be buffed or nerfed the most important consideration is how it fits within the overall game. No nerf or buff is inherently good or bad. Any change which moves something closer to the power curve is the correct one, as it improves game balance. It is easy to mod your way into a completely unbalanced and boring game. The devs have to make more careful evaluations and think of overall game balance.

The balance currently is that the bots are end-game tech with infinite research buffing them indefinitely, while blue belts are practically mid-game items with no further upgrades at all. You can't really expect for them to be comparable, right? The question is, would you remove some power from bots moving them mid-game, or give belts end-game power? I think the answer is pretty obvious.

[greaman]

But personally I think Twinsen is more worried about players who switch all their new production to using an assembler with requester and provider chests, and just stamps those down mindlessly, every time the need more of something. And I agree that scaling production that way is a boring way to play. Neither of these solutions really does anything to stop that. (There is also the player who just mindlessly stamps down assembler + 2 chests surrounded by 8 or 12 beacons. You could force them to change their blueprints, but even with chests 4x4 they could still fit 10 beacons per assembler).

I wholeheartly agree.

Plus:
- the problems discussed to be corrected (bots being to powerful) have been recently created by the devs in the first place (nuclear power, infinite research)

I still have a megabase on vanilla running and expanding since months, but if the mods would nerf bots with that kind of reasoning that would be the last vanilla game if have started.

I heavily depend on bots for storage as I tend to accumulate ressources for weeks whilst being away and then start science/expansion and as the devs have neither implemented a science queue nor decent mass storage in vanilla, that would really take the fun out of the game...
(And I still remember my rediculous chest-inserter-chest-inserter-chest-inserter {...} setups some game versions ago...)

[mushroomxchen]

Sorry for my English, I know it's not very good.

Factorio is a sandbox game. It should focus on encouraging and helping players to create what they can imagine, not limiting them.
Balance between bots and belts is some issue, yes. But not more important than what we need to create a awesome base.

To me, I can't agree belts are more fun at late game. Think about a 32 to 32 belt balancer, how many people can accually make one by themself?
I can't even make a 8 to 8 properly without google some example. I don't think such things should be fundamental to people who are trying to play this game.

When start building a bot based megabase, I felt it's more like a new phase of base designing.
Finnally I can stop thinking about those small belt corners/mergers/splitters, so I can think in a bigger picture.
With bots, numbers on the paper became achievable, input and output can be much higher, so I have the module to make a really big base.
And it's not boring, although one single module is simplified by bots, it's much more complicated on the sytem trying providing enough input and output, the rail sysytem.

I think what I really need is not a gigantic belt balancer at my train station, but a better system to make my train go to the stop I want.

Back to the balance thing.
To make belt viable, inserter should be able get/put items from/onto belts at the same speed as they do with chests. This will make numbers on the paper achievable.
There should be 2 kinds of logistic bots.
Long range bots can move few items to a far location. I would use them to transfter rocket fuel to nearby train stations or some other light job. I don't want to build any belts to do that, It's a waste.
Short range bots should be able to move only in a small area, not from one port to another. In this area, they can transfter logistic items from one chest to another at very hight speed. At least its' a much better balancer.
With some limit on short range bots number per area and area size, I think belts could find some usefulness in my base.

Just my opinion.

[schigum]

Hiho, here my two cents:
in the given context the biggest issue I see with bots is that there is no limit on the number of bots working a route.
So you can crank up the throughput almost indefinitely by just adding more bots (and charging stations), since bots can stack as much as they want.
Getting rid of this stacking is probably too demanding performance wise, but one could work around the issue by introducing a frequency limit for the interaction of bots with items.
Say, only one bot per second can interact with a chest.
This would limit bots to low throughput scenarios (high level items), while the mayority of the base would have to be build with belts. Yet you still have the flexiblity and comfort of the logistics system. Also, this would allow for a switch to easily swap universe 1 and 2...

Cheers,
Schigum

[golfmiketango]

Power Creep

There's a lot of discussion (I didn't read all of it) saying thing like:
Nerfs suck, only buff things
Devs are trying to lock in a "correct" way of playing

When trying to decide whether a feature needs to be buffed or nerfed the most important consideration is how it fits within the overall game. No nerf or buff is inherently good or bad. Any change which moves something closer to the power curve is the correct one, as it improves game balance. It is easy to mod your way into a completely unbalanced and boring game. The devs have to make more careful evaluations and think of overall game balance.

Right. Everyone keeps saying they want the devs not to force something down their throats but that's the wrong answer to the wrong question. First, devs really aren't doing that. That's why they've provided an extensive LUA API which allows (within reasonable limits) sufficiently motivated customers to realize whatever vision of factorio they would like to have. But, the base game is a different thing. The base game is in fact not there to always give everybody all the options they could possibly want. That's the official factorio experience that will be the jumping off point from which every player's experience of the game begins, and, for a plurality of players, probably where it ends as well.

Imagine, for example, if the base game was changed to be the Creative Mode mod. Nobody could tell you how to play factorio, true. But new players? There would not be so many. Newbies would find themselves in a world of limitless freedom to assemble wierd inscrutable factories building multi colored flasks.. but why bother? Pretty soon they'd just decide they had better things to do. Newbies need artificial challenges presented to them so they can tackle the learning curve and understand why and how the game engine is a powerful potential catalyst for fun. Until they've had that education, they mostly will have no use for the limitless freedom to play different styles of factorio.