Friday Facts #225 - Bots versus belts (part 2)

[golfmiketango]

To an extent I suspect the reputation of belts as throughput limiting is a function of the main-bus design. Main-bus is a fantastic design for zipping through the early game but it encourages space-constrained, linear layouts that do not gracefully accommodate expansion later on.

I'll use myself as an example: beginning in mid-game, and increasingly so, in "post-game" play, my bus tends to become a bad habit that I can't let go of. This mindset forces me to prop up my legacy production designs with robotic augmentation. Never, because I wanted to. But, instead, because I didn't feel like tearing my base up, and I was too lazy to sit down and make careful plans, so... well looks like I need some more iron plates here, and here and here, so let's see if robots can squeeze their way in and inject the material I removed further back. When I start ramping up science and wanting to run all that machinery at the same time, I start to realize I didn't leave enough room for expansion.

The end result: horrible spaghetti and my base winds up looking like the opening screen of factorio after the progress bar. Furthermore I find it difficult to go past about 600SPM this way, no matter how many bots or roboports or whatever I want to throw at it. Also: ~35 UPS running science at full tilt.

To me it's pretty clear that this is not anything to do with belts being "totally unplayable," at least, not in 0.16, now that they are so much more UPS-efficient. It's to do with a determined attempt to "force" 20 lanes of iron plates down a 4-lane bus. Factorio gives me two dimensions but because "bus" I'm only exploiting one of them for bulk delivery. As soon as I leave my main factory and carve out some room for myself, suddenly a whole bunch of new approaches are possible. In a wide-open area to build things, belts suddenly feel a lot less constraining and I find they can, indeed, be superior to bots and trains for certain tasks.

Speaking of which, ever tried to build a bus with bots? It can be done but it isn't pretty. Nor is it min-max optimal, especially in high-throughput scenarios (although maybe buffer chests have ameliorated this limitation; I'll let you know when they get fixed, or I get fixed to understand them properly, so I can actually use them).

[Bigbigcheese]

How about instead of stacking the items, you stack the belts...? ie you can run one belt on top of another belt to double the throughput? Though you'd have to solve all the height change differences, ramp objects or something

[vtx]

The stacking on belt only seem the best way to keep belt pertinent in middle to end game vs bots. Assembly machines should be able to get directly stacked items too, to keep belt on same level of the bots.

bots througthput 326 items.
blue belt 40 items.

Even a small stack of 2 items will keep the usual 4 lanes main hub will bring the belts on same level of the bots.

[Dangdle]

The problem with bots is the sheer number of them, I recon roboport capacity should have a limit to 5 bots (remove bots stacking to 50), have normal research that progressively takes the capacity to 10, then space science that adds 2/4/8/16... and so on (maybe multiples is too strong?), this would make them stay just as powerful for lategame megabases, and underpowered early game, so that they are useful mostly for resupply or construction.

[Caine]

bots througthput 326 items.
blue belt 40 items.

Theoretically if you braid belts you can raise that to 80, but it is still no competition.

[RobertTerwilliger]

Bot nerfs just won't work. Longer charge = multiple roboport chunks. Not fun. Maybe give them item pickup\placing time, or acceleration from quite low speed, to make them inefficient for short range transportation?
Or add some ACHIEVEMENTS to encourage players at least to make few playthroughs, say, without any requester chests (bots for personal supply only), or no bots at all.

[blueblue]

I'd like to take another look at the question of why bots really need to be nerfed.

People have made different experiments on how bots compare to belts with regard to throughput, and in some experiments bots hold up equally well or better than belts, see the subreddit, specifically see this, this. Note that in the FFF experiment, the bots have all upgrades, which takes quite a bit of research and that should be worth something.

It is dubious wether bot mega bases are even the correct course of action in 0.16 if UPS is your limit, one of those posts suggests otherwise. In any case, no one is currently building belt (or any) mega bases, because the belt compression question has not been answered. This is a bad time to look at the mega base meta because in 0.16 it doesnt exist yet.

EDIT for clarification - Currently we don't know wether sideloading will compress belts in the near future or not. So if you are building a mega base with belts now, you may have to redesign it if sideloading compression is added again. Some experiments suggest that belts might be better or as good as bots for mega bases if you're trying to get the maximum science per second on your device. So if you are building a mega base with bots right now you might find that you should be using belts later. Hence not many of the people who would like to are building a mega base currently. This is referring especially to streamers who the FFF post referred to.

For casual play, similarly bots seem to have a better throughput than belts, but again this is after unlocking all the technologies. When you unlock bots, they are in a very sad state indeed. Bots have other drawbacks, such as ridiculous power costs, being hard to set up and you need to put everything into provider chests. So the issue might also be with the drawbacks of bots being less visible than those of belts. When I got logistic bots the first time I thought that they were quite overpowered and unfun and used them sparingly. The game doesnt tell you: "Sure, it's as simple as putting down an assembler and a requester and a passive provider chest, but you also need to multiply your power if you want to do that often. Circuits and smelting too? Go multiply again then. And if you actually want them to do something in time, you'd better have 5k of all science packs ready".

I still think bots are a bit too bad when you unlock them and a bit too good when you get the speed upgrades, and they should not be able to handle high throughput as well as they do currently. But they are not in need of drastic nerfs. Instead, their cost should be made more evident to players. Increasing their power cost a bit and making it generally a bit harder to get power would work. Belt qol improvements, obviously.

This forum needs kudos so much.

Yep.

[FasterJump]

• I totally agree with what you guys said in that FFF.

• I'm very enthusiast about the splitter buff (so much new options!)

• Imo, increasing bot charging time will affect personal transportation...

• Some random ideas:
-Increase normal inserters capacity from 3 to 4
-Reduce robot capacity from 3 to 2
-Increase (=nerf) charging time by factor 4, buff battery size by factor 2. (so player supply is buffed while chests supply is nerfed)
-set a fixed throughput to requester chest, like a cooldown of 1 second each time a robot supply it
-set a mandatory robot maintenance time of 30 seconds at robotports each time a robot supply a chest/player

[ThirteenEra]

Would it perhaps make sense to leave the bots speed etc as-is, but nerf the type of items they can carry?

Make it so that the items contain a [Lightweight] flag, and the bots can only carry items with a flag of 1 - for example, bots can transport green/red/blue circuits, plastic, research bottles, maybe stone, efficiency/speed/etc boosts, batteries and sulfur, coal, etc. But the bots would not be able to move copper/iron/steel/oil barrells/engines/electric poles/crafting machines/solid fuel/solar panels/etc. They would still be placeable by construction bots, just not carrieable by logistics ones.

The FFF talked quite a bit about how Death spell simplified "older" content while still keeping new stuff challenging. Isn't this what the "weight" solution would offer? It would allow you to simplify the "old" noodles for the circuits, batteries etc, while still making sure you have to plan the logistics for the "current" noodles aka rocket launching. And the fact that these values would be moddable would make it so that almost anyone could be pleased - want the "old" bots back? Mod all items to have flag 1. Want more challenge? Make bots able to move less items. Want less challenge? Add copper/iron plates. Etc.

Im by no means a "serious" factorio guy, but i feel like this could be a potential solution?

[vampiricdust]

The stacking on belt only seem the best way to keep belt pertinent in middle to end game vs bots. Assembly machines should be able to get directly stacked items too, to keep belt on same level of the bots.

bots througthput 326 items.
blue belt 40 items.

Even a small stack of 2 items will keep the usual 4 lanes main hub will bring the belts on same level of the bots.

What are calculating here? Blue belt moves 40 items in one second. A bot at 100% speed moves 3 tiles. Every item on a blue belt moves 5.625 tiles per second. Each belt holds 7.11 items on average. So a blue belt is 2 bots at 4 capacity with 187.5% speed boost. Each bot is worth 3.4 blue belts, which is longer than distance these 2 bots can go. It takes 267% for 2 bots to move 8 items faster then a blue belt can. Meanwhile, those 6.8 blue belts can hold 48.3 items, which is 22 bots with capcity 4.... it takes at least 2 bots with 1.5 million resources of research to just move 1 belt worth of resources. To match cpacity, you would 10 logistic bots per second for every second of travel time from the bots starting point, to the starting point of production, to the end point of production, back to a roboport, and for every second waiting to charge.

[AileTheAlien]

I haven't read the full 22 pages, so hopefully I'm not repeating what somebody else said...

The death of filer inserters - While we're at it, let's get rid of filter inserters and stack filter inserters. Let those be unlocked from the same tech upgrade that filter inserters are given to us. There is no real reason to have a separate inserter entity that cannot be reverted back to fast inserter (like you can convert blues to purples).

I'd go even further, and have even fewer inserters. The game's UI would be a lot cleaner if we only had a small number of well-defined insterers. Blue overlaps too much with both yellow and stack, and the fact that non-stack inserters get stack bonuses always feels wrong to me. Half of all of the inserters in the game are basically useless at different stages of the game, cluttering up the UI.If it were up to me, I'd make it:
- burner: as is now, but gets speed bonuses from fuels (as proposed elsewhere)
- yellow: mostly unchanged, but new research to upgrade speed (and more power used); final speed upgrade is prerequisite for stack inserters, and their swing-speeds match at this research level?
- red: mostly unchanged, but gets partial speed bonus of yellow's research, to compensate for no stacks
- green: basically unchanged?

This would actually be a boost to reds, but only the rate at which they get usefulness boosts. i.e. sooner in the game than the current game's stack bonuses. The blue/yellows would effectively be merged, and the research/speed bonuses should be balanceable. Blues cost more than yellows now, but the overall cost of the new blue/yellow replacement should be balanceable with power usage and research costs. (more solar, steam gens, etc, would put an indirect cost increase on them, for resources)

[zoony-247]

I've had enough of this rant for 2 weeks now, I had to register an account to say how I feel.



I first want to thank you, the devs for making such an amazing game. as someone who suffers from depression this game has helped me out tremendously

Please leave the bots how they are. As you know how passionate your community is about your game, you have figured out that there is no easy fix to this "problem". There is no problem. Everyone plays the game in their own way and trying to balance the way each player builds their factory in a single player game(mostly) is crazy.

" it's our jobs as game developers to make sure that the most optimal way of play is also the most fun way of play" <----- this is crazy to me, I'm open to talking about this but i just don't understand this

[Culex]

I don't see the point in nerfing bots. No matter what changes are made, the answer will always be "make more bots" unless the throughput of bots is artificially throttled (which I think is what most of the suggestions in the thread come down to). My own couple of suggestions, along different (beacon hating) lines:

• Remove beacons from the game. I really think beacons are a significant contributor to the weakness of belts. They are clunky, ugly buildings that are redundant from a gameplay perspective since they have the same function as modules. I hate having to design my bases around them and they make a lot of belt-based designs impossible because they take up so much space. You could buff modules (add additional slots, etc.) and have the exact same gameplay without them.
• If beacons are kept in the game, remove the stacking and increase the effectiveness/area of effect of a single beacon. Just any change that doesn't require multiple 3x3 beacon buildings to be clustered around your buildings would be nice. This would create more space for belts as well.
• As has been suggested multiple times, specific late game technology (such as loaders) for rapidly loading and unloading trains onto belts in a seamless fashion. I dislike the use of stack inserters for this, since it feels like our loading and unloading is being artificially limited here in a way that doesn't add fun or challenge to the game. I'm iffy on whether the same tech should exist for chests, but I think it would be nice on balance.
• Generally adding late game technology that makes belts easier to compress, easier to load and unload, etc. in a visually pleasing fashion. I think the improved splitters are a good first step here. In particular, I don't understand why late game belt compression should be difficult. It's a weird problem to have to solve in the late game, especially considering how many quality of life improvements you've received for other aspects of the game.

In the end I don't really care whether belts or bots are better for megabases, but I'm all in favor of belts having more fun features and being easier to use later on.

[Mobius1]

Sticking to the point:
We all agree that in the long term, bots are superior to belts from the moment they're available to build, even nerfing them they'd still be superior.
So, buffing belts, or increasing their speed would be just another blueprint copy-paste thing, nothing new and challenging would be put there (see what I did there Kovarex? )
(BTW, you used way more furnaces with the bot transporting than from the belt transporting, the truth is if you plop down same lanes you would see 1.2x difference on the final production. The main purpose of bots is that they can be used with multiple inputs recipes that are impossible to achieve with belts, also they're faster than belts when you need a big base, belts are useless.)

Instead, what about using belts as real-life logistics happens, inside the assembly machine! Inserters takes from the belt, put on the machine inside the factory building, that whole production line will output a product every x seconds based on the speed of the machine, inserters and belts. So the idea here would be something similar to Factorissimo where it would be available in the vanilla game and the factory building would replace the assembly machines, there would be no more assembling machines in the game, instead they would be an factory building where you have to design the production block inside your factory building using belts, containers, machines that will do stuff with the raw materials, inserters to move intermediates around and that factory building would output in a box outside, in the real world, where logistic bots would then grab the finished products and separate them for delivery either by train, car, truck, whatever means the player would choose from.
But there is a catch, you would have to prepare that factory building to craft 1 product at a time, just like an assembly machine does, but the productivity from the factory building would depend on the player's designing inside it. That way you make the logistic part of the game more interesting, there would forcefully make belts worthy somehow since they would be necessary inside the factory buildings (OFC I would still download the mod that would enable logistic bots inside bcoz belts sux, but that's just me) and everyone would be happy, excited, glad to play exactly the way they want... (it would be a shitshow, obviously, but you know, I had to )

I hope something useful will come from those words and kudos for the FFF #224 Twinsen you managed to light a fire without a fire, I loved it!

[Xuhybrid]

After reading several posts, and giving it some thought. I believe that no matter what you do about bots your gonna draw the ire of the bot lovers. They are powerful.

As far as belts are concerned. I think that the biggest point of contention is getting material down the bus from its point of origin, whether it be from smelter or train. I think that the stack inserters should be left alone. Higher tiered belts would certainly solve the issue of super wide buses. I would think that belt balancer / compressors would most definitely be designed with the given tools. There are two ways that material can be picked up by inserters at the assembler, 1. Have the super-fast belt compress to a point that the inserter can grab what it needs, or 2. Put down slower belts to slow the material enough for inserters to grab. Either process will work. Most people enjoy the satisfaction of seeing compressed belts and will work to maintain that status quo. It will most definitely drive creative design.

I'm leaning more towards the idea of having research late game that increases belt speed and inserter speed. This would lower the gap between them and bots while not adding any more infrastructure to worry about or upgrade to.

[JCav]

I find belts to be tedious. I massively enjoyed building my "bot cell" system for my "megabase", solving the problems of moving items quickly enough to get to ~2k sci/min was quite the challenge. It was so much fun that every subsequent restart has seen me get bored of the fiddling around with belts in the early game, so I've never gotten to even blue science, let alone bots again. I think that you need to counterbalance this discussion quite a bit with the fact that a large minority do really enjoy bot based play. It's not the zeitgeist on this forum, that's for certain, which is why I rarely contribute here, but things already feel drastically nerfed since the 15 change to logistics system research. You have a difficult challenge ahead - it feels like there's two games in development here, and you're focusing on one, and ignoring/actively nerfing the other. Ultimately it's your choice, but I'm probably just going to go back to playing not-factorio for now.

If you've never made it back to bots again, why should they cater to your desire for no nerf to bots? You don't play, and when you do, you haven't even used the feature being discussed again.

As it stands, bots are a FOOS. First Order Optimal Strategy. They are in fact the best way to move items from one location to another. That does not make for good gameplay. The fact that this game is in early access still means it is a perfectly fine opportunity to make changes such as these.

[ahydra447]

My 2c:

My brother and I enjoy this game a lot. We are very casual players - we play for 5hrs or so per week, and our current project is building a 0.2 SPM (that's one rocket per 5 minutes) megabase, which judging by some of the other numbers I've seen is barely anything We're so casual that we don't use beacons or blue belts, and build huge oil-fuelled steam engine stations for power rather than mass solar farms.

Our megabase is set up as follows:

• main base: belt-based using splitters to feed various small production lines, including science
• rocket smelters: over 1000 furnaces, using yellow belt as input and red belt as output. Trains bring ore and robots unload it into chests; ore is then belted into the furnaces. Resulting plates are belted into provider chests where robots feed them into the modules (see below) in a separate logistic network.
• rocket factory: 70 small modules (2xRCU, 2xLDS, 2xRF AM2s) arranged into 5 blocks of 14. Each block is its own logistic network where the robots take the inputs (iron, copper, solid fuel, plastic bar) and deliver it to the modules. Within the modules we use very short belts. This design was chosen to be easily blueprint-able and also to self-load-balance, and generally works very well, but obviously could be even more compact if we used robots alone.
• rocket silo (just one because we're filthy casuals): output of modules travels on very long belts (no problem as the output rate is very slow) to reach the rocket silo. Space science is transferred by robot simply because the rocket silo is moderately far from the labs (not worth a train, but rail lines are in the way of belts).

So overall, we use about 50/50 mixture of belts and bots, and that's fine for us. Bots feel at the right kind of power level: the bot tech comes very late on; roboports are not only power-hungry but require a lot of stuff to build, and in return you only get a moderate-sized coverage area and 4 charging points; and robot speed research is very expensive. There are a few reasons why we generally prefer to use bots though, which are no doubt amplified as your base gets bigger and bigger. But I think that primarily if the devs want belts to be used more, I think it should be more about making belts easier to use rather than nerfing bots.

• In early game, you spend all your time building belts to get the main bus going. You have to prepare your main bus for lategame, so must leave space for 4x iron, 4x copper, 2x green circuits, 1x coal, stone, etc... belts which increases distance from the bus to the factories. Additionally it seems iron plate throughput off the bus is often very poor, leading to spaghetti as you try to boost a factory's income.
• Underground belt reach feels very small. It only just reaches under a train track and often isn't good enough to solve a spaghetti issue. If it was the same size as pipe-to-ground it would be much more useful.
• Belts are expensive. Even red belt requires a TON of iron to make en masse; blue belts are so costly we don't bother with them. One would rather be spending the iron trying to produce more green circuits since you can never have enough of the damn things . If we want to move more towards a belt-driven game, we need cheaper belts (or perhaps just make it easier to obtain iron/copper in general, e.g. increased mining drill output - I was a bit disappointed to see resource layouts nerfed). A nice change would be for iron gears to cost only 1 iron rather than 2, as this would help a lot with getting enough iron off the main bus for science packs. Some recipes (given the topic at hand, most notably roboports) can then be changed to require more gears to compensate.
• Unloading a train quickly using belts is very hard. I'm sure you can imagine the belt spaghetti nightmare that you would get trying to organise 24 belts (4 cars x 12 inserters unloading in pairs onto 6 belts) from a train. Using robots and fully-upgraded stack inserters, it can be done in about 17 seconds, and the bots take everything away immediately.
• Belt compression issues (inserters refusing to fill holes in belts), which I believe was worsened in 0.16 but is being fixed.
• (the most frustrating) Player inventory, even with the best armour available, is very small at only 90 stacks and generally half or more of that will be occupied with things like belt, inserters, pipes, blueprints, personal robots, etc. (With biters, add ammo, turrets, defenders to that list) So without the ability to plop down a requester chest and ask for whatever it is you need to build this particular power plant / train station / etc, you have to make a lot of trips back and forth in order to pick up ingredients, which is extremely tedious, even with 5 exoskeletons and concrete paths everywhere.

After all this rambling, I'm not even sure what point I'm trying to make here . Perhaps: bots feel like they're in a good place, if they're going to be nerfed it should perhaps be their charge capacity / charge rate, to avoid them being used to carry things over long distance rather than longer belts and trains; belts are expensive and sometimes fiddly to use; and please buff player inventory size a lot in the late game to avoid having to make so many repeated trips.

On a semi-related note, I really love the look of the new splitter options. Perhaps that will help simplify some main bus layouts.

[Bi0nicM4n]

Players would still be able to build robot only factory, belt only factory or combination of those, but the strongest strategy would be to combine all types of transport, each for the part where they are the strongest. My suggestion would be to either completely remove the Worker robot cargo size research, or more likely, to multiply the charging time several times. For factory transport, both would have the same effect, but the latter wouldn't hurt personal transportation that much, as robots would transport all the missing items to player before they would need to be recharged, and then they would recharge when the player is already supplied and doing something else.

If anything, don't delete cargo size research. If you decide to increase robot charging time, it could still be modded to one's liking. Restoring deleted functionality - not so much.

[SagaciousZed]

Brining back loaders given the changes to compression. Also let loaders load from trains onto a belt with no buffer? I think will be the end game less headache buff to belts. Doesn't really increase throughput for belts, but it would increase throughput from train to belt so you can easily make it even as powerful as bots by running more express belt out of each train wagon.

[AlmightyCrumpet]

I'm not sure if this has already been suggested, there are many pages on this thread. It is probably more of an early game thing as it starts but what if you had assembly machines that had belts feed directly into them (rather than having inserters feed off the belts). There is a mod that adds this sort of thing called Assembly Line, it does recipes that can take up to three items. It would certainly make things interesting in building designs, not to mention you can always expand it by increasing the building sizes to allow for more inputs.

As for my opinion on the Bot vs belts. I prefer to build with belts, it looks more visually pleasing to see the vast array of colourful item filled belts and personally it gives me a better idea on what goes where and if anything is lacking. I still use bots as they are convenient for requesting items and building.