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

[vampiricdust]

That's what I was trying to say, just put in other words. The game about belts and the game about bots are, in a way, two different games. So Factorio as a whole, being, by the very-late-game, about bots (as they provide higher throughput, more convenience and lower CPU load than belts) is mainly a matter of the game-design decision. Some would like it, some wouldn't. Personally I don't like the uselessness of belt-like mechanics for higher throughputs, and consequently don't like being forced to use bots as a replacement for belts. By the extension, I don't really like how useful bots are for higher throughputs (it just doesn't feel right to have both high flexibility and high throughputs in one transportation methods). But as these are my personal preferences, they don't really matter for the development of Factorio As for what would be more "fun" for the "playerbase as a whole", I have no idea.

There is no game "about belts" or "about bots". Neither are required to "win" the game, but both make easier in their own ways.

What I was getting at is that belts are for when you build the factory from the microcosm of making all the initial builds, building up production lines & sorting out what goes where. Once you hit the rocket launch stage, the game transitions to a macrocosm of slapping down all these pieces you've figured out and got working to mass produce items. You have blueprints of your micro factories which used to be puzzles of belts, inserters, assemblers, & etc, but now are pieces of blueprints being connected together. Logistic bots simplify the connections by freeing the player from having to constantly connect each piece of the puzzle together. To me, if you have to run pipes or belts, you should blueprint the bigger scale and use that.

Logistic bots allow macrocosm thinking with the microcosm blueprints. Everyone says the logistics bots are overpowered or make the game too easy, but I say they're thinking too small. Eventually your view of the game grows and logistics bot go from cheaty / overpowered to allies with their own flaws and quirks. Trains are the bridge that ties players' transition from a microcosm play into the macrocosm play.

[raidho36]

Seconding the idea of limiting bots to only have one of them access an individual chest at a time. That way you can't just make more bots to increase throughput, you'll have to actually design the logistic network properly. Particularly high throughput areas will not work with bots altogether and will require blue belts instead. That's not to mention, having to build more bots is just adding pointless grind, not something you want to have in the game. Switching beacon effects from area based to entity count based (instead of affecting everything in specific radius, it affects specific amount of nearest entities) (so that belts occupying space is not a factor) will help the situation as well. Alternatively, make beacons work over electric network, possibly with an extra type of wire to transfer module boosts. Every entity connected to these wires then could set how much effects it wants to consume, and the network will share beacons' combined boost according to individual entities' settings.

[Lubricus]

Seconding the idea of limiting bots to only have one of them access an individual chest at a time. That way you can't just make more bots to increase throughput, you'll have to actually design the logistic network properly. Particularly high throughput areas will not work with bots altogether and will require blue belts instead. That's not to mention, having to build more bots is just adding pointless grind, not something you want to have in the game. Switching beacon effects from area based to entity count based (instead of affecting everything in specific radius, it affects specific amount of nearest entities) (so that belts occupying space is not a factor) will help the situation as well. Alternatively, make beacons work over electric network, possibly with an extra type of wire to transfer module boosts. Every entity connected to these wires then could set how much effects it wants to consume, and the network will share beacons' combined boost according to individual entities' settings.

When using bots for automate the logistics you don't need more throughput from one chest than what an assembler needs. It's when you rebuild or move parts of your factory you need high throughput for the chests. So that change will mostly nerf the task I want the bots be good at not automate science production.

[BenSeidel]

What would be the mechanism that would cause the bot to go to get maintained?
If you have some form of external "repair" entity then It's probable that one bot will always remain in the air never going into a roboport so can never be "removed to be repaired". This will mean that you always need to have a bot supply chain to replace the broken ones.

Another question would be: What are you adding to the game? if all you are doing in effect is increasing the effective size of the roboport footprint (by having inserters and/or repair stations etc) then what is it's value?

I only bring this up because it reminds me of the mod wear and tear. It came out of a thread similar to this except focused on assembly machines. I can't find the thread but it's outcome was essentially that it was just placing a requirement on the player to cover everything in roboports, not a creative solution.

[E-37]

Hey! Ive been playing Factorio for a while now but just got around to joining the forum. Here are my thoughts:

- let robots crash when they run out of fuel. Crashed robots could either be lost or be recovered by other bots.

This makes sense that the robots would do that, however it would need to be implemented in a special way. What would happen if the bot crashed? If it turned into an item, it could land on a belt and clog up production. There would possibly need to be a research to do that or have bots not crash when they are waiting to charge.

- have blue belts go at x4 speed (up from x3 currently) so upgrading to each new tier is a doubling of capacity and is easier to plan for

Having another tier of belts would be nice. I wouldn't change the speed of blue belts because that could mess up designs that rely on its speed. I know that at some point fast inserters cannot pick up off a belt. I am using a mod that adds a x6 belt and fast inserters can't use it (similar to how normal inserters can't use blue belts) I would add another tier of belts rather than modifying blue belts.

- make it easier to compress belts by ensuring that all gaps on a belt are big enough for an item to be inserted into it

How would you propose to add this change? It seems tough to implement and wouldn't change anything since bots become very strong in megabase phase. At that phase being able to insert a few more items on a belt wouldn't matter at all. If you are having trouble with partly compressed belts, try doing the math with the belt speed vs output to ensure a good ratio.

As for having bots be unable to occupy the same airspace, it would be nice visually, but would probably slow the game down a lot. The pathfinding for 10000+ bots flying and colliding at once in a large base would be awful. It would force you to use belts over bots simply because your computer couldn't handle the bots.

[Krazykrl]

Essentially you're talking about a progressive tax on running bots.

Individual tracking of bots will be prohibitively expensive; especially since bots are currently "cattle" instead of "pets".

If you really want to add a tax to bots... just give them a 1% chance of being destroyed when they are empty and attempting to land. This still doesn't negate the fact that bots are scale-able, but boring... and belts are not as scale-able, but more difficult to setup and manage. Once again this leads to the fact that bots aren't overpowered (especially in a game of practical infinities); but the fact that the alternatives cannot scale to the extent of bots.

[raidho36]

If you had limited bot lift rate to 1 per second per chest, then you'd be choked for many recipes, most of which are ubiqutous such as green circuits. Making bots unfeasible for extremely large volume transportation for applications like this is the whole reason this topic is brought up.

[Koub]

[Koub] There is no added value leaving this topic on its own while this has been suggested many times (and discussed) in the bots vs belts FFF threads. Merging it to keep everything together.

[Daid]

- let robots crash when they run out of fuel. Crashed robots could either be lost or be recovered by other bots.

This used to be the case, but annoying as hell. So they changed it to moving very slow.

Giving bots collision would most likely be a huge performance impact. Making bots a UPS problem. Thousands of free moving objects colliding isn't light on a CPU.

[Koub]

[Koub] Merging this in the FFF #225 Topic, which hosts the second (and hopefully last) part of the bots vs belts discussion.

[Tricorius]

I keep hearing the “same airspace” collision argument. It seems people are discounting altitude. If you fly directly above birds (or look up from directly below) they appear very similar to a swarm of bots. I’ve always been impressed with the flocking aesthetics the devs put into bot swarms.

Also, I’m assuming that bots and chests are purpose-built to accept fast loading / unloading. Take a look on youtube for the various quadcopter swarm experiments and you’ll see some pretty interesting flight patterns. I’ve always kinda assumed my dude was a pretty good programmer, or maybe the bots have an excellent AI.

I’m also curious if anyone has used a large quantity of personal construction bots without a fusion reactor. They come back in and swarm around you waiting for charging pads to be free. This is especially pronounced when you have portable solar, and you’re building at night.

It is horrible. I don’t want my logistics bots doing that when loading me up after I come back from using a bunch of inventory. There is nothing worse than waiting on the game to catch up to you. (I realize people have mentioned throttling only at the logistics chest level, and specifically requester chests. But it will look odd if the bots can grab a few thousand items in a swarm, but can’t deliver them with similar speed.

[Deadlock989]

I'm playing about with the Universal Paletting mod and I have to say it's really nice. Yes, it makes belts more powerful by a factor of up to 100x. But it's not a trivial exercise to set up the logistics of returning empty palettes and so forth, and not "boring" either as suggested by the dev blog. It's just barrelling for items, and carefully chosen items only. If you also banned bots from carrying palettes/stacks/whatever you call them, this would be a great way to making belts late-game relevant again.

... Maybe 100x is a bit much.

[GasBandit]

99 hour player here, created forum account just to say that the nerf to bots (moving requester chests into a much higher research tier) has basically ruined my enjoyment of the game. I hope somebody makes a mod to put it back how it used to be.

[pgriss]

I agree with all those suggesting limits on robots as it pertains to physical space.

It seems reasonable to limit how many robots can access a chest at any one time.

It seems reasonable to add a delay when a robot picks up an item from a chest.

It seems reasonable to limit how many robots can be in a tile at any one time. (Even though the robots can move in 3D, they shouldn't stack infinitely. You could allow research to add new layers in the 3D space. Call it "air traffic controller improvement" or some such.) This could be implemented literally on a tile-by-tile basis or in larger chunks based on roboport coverage.

It also seems reasonable to limit what the robots can carry but I think this would result in more problems than benefits.

All of these limits would have an "organic" feel in the sense that they would make intuitive sense.

In contrast, making the requester chest so hard to research (relative to logistic robots) seems like an artificial barrier to limit the usefulness of robots not in terms of scale but in terms of ad hoc nerf to functionality. It should be made available by the "Advanced Electronics 2" research (and maybe require a blue circuit to make).

[Yinan]

99 hour player here, created forum account just to say that the nerf to bots (moving requester chests into a much higher research tier) has basically ruined my enjoyment of the game. I hope somebody makes a mod to put it back how it used to be.

Here you go: https://mods.factorio.com/mod/logistic-research-fix

[Deadlock989]

99 hour player here, created forum account just to say that the nerf to bots (moving requester chests into a much higher research tier) has basically ruined my enjoyment of the game. I hope somebody makes a mod to put it back how it used to be.

There are like six of them.

[bobingabout]

99 hour player here, created forum account just to say that the nerf to bots (moving requester chests into a much higher research tier) has basically ruined my enjoyment of the game. I hope somebody makes a mod to put it back how it used to be.

Just for your information, My Logistics mod DOES change when requester chests are unlocked (The Logistics system research) back down to blue science.
and as of my mod version 0.16.1, bumps the requester chest down to be included on Logistic robotics, so you have it from the start. The logistics system only unlocks the active provider and buffer chests now. (Logistics Robots only unlocks Passive provider and requester chests, no storage, and Construction robots only unlocks Storage)

My mod does however add a whole bunch of other stuff too though, Bots that go all the way up to Fusion powered MK5 that don't need to recharge, higher tier, faster belts and inserters, and 2 tiers of 2 branches of trains, a version that just goes faster, and has more storage space, and an armoured version.

The one thing I don't add however, is Loaders... and given how this whole Bot vs Belt stuff is going, I can see that it is actually a hole in my mod.


Though, if you did want a mod that just changes the requirements for the logistics system... I know back in 0.15 when they increased the requirements, someone made a mod to bring it back down to science pack 3, maybe it got updated to 0.16?

[Termak]

I agree with all those suggesting limits on robots as it pertains to physical space.

It seems reasonable to limit how many robots can access a chest at any one time.

It seems reasonable to add a delay when a robot picks up an item from a chest.

It seems reasonable to limit how many robots can be in a tile at any one time. (Even though the robots can move in 3D, they shouldn't stack infinitely. You could allow research to add new layers in the 3D space. Call it "air traffic controller improvement" or some such.) This could be implemented literally on a tile-by-tile basis or in larger chunks based on roboport coverage.

It also seems reasonable to limit what the robots can carry but I think this would result in more problems than benefits.

All of these limits would have an "organic" feel in the sense that they would make intuitive sense.

In contrast, making the requester chest so hard to research (relative to logistic robots) seems like an artificial barrier to limit the usefulness of robots not in terms of scale but in terms of ad hoc nerf to functionality. It should be made available by the "Advanced Electronics 2" research (and maybe require a blue circuit to make).

I agree with everything in this post. I personally dont use bots for mass transportation, i like belts. I do use bots but for the smaller stuff, like refueling trains, catalysts etc ( i play pretty much only angelbob games, not vanilla), and for hacking small scale production.
I really find it weird that basically whole heavyduty logistics can be just simplified by massing superfast botswarms, ports and chests.

[Lastmerlin]

Or you could also ask yourself why is there so much noise with people asking to nerf bots in every imaginable way now that the devs have opened the pandora box, but almost no bot balancing issues were reported before (and I could find only one mod that did nerf the logibots that existed before this shitstorm, whereas there are literally dozens that BUFF bots in any possible way since months or years).

I wonder what would have happened if the devs had swapped bots and belts in their 224&225 FFF. Would have people overwhelmed us with ideas to nerf belts ?

This is an excellent insight, that I want to emphasize once more.

My interpretation:
The mods for better bots are for a simple reason: Better UPS, means more production and even larger train networks.

The shouts for bot nerf are far more on the psychological side. Nothing is wrong with the standard game up to the rocket. Bots are rather weak in this area, people built their belts and were happy about the challenges. No complaints. Now Twinsen posts this *boring* base, where similar problems are solved with high-end tech and equipment (like mass level3 modules) and it gets way easier. This induces a feeling of *devaluement* of the achievements attained by the spaghetti belts. Thats why the typical advice *just dont use bots* does not work: The knowledge, that more efficient means exist, seems to spoil the joy in laying belts. Perhaps its also a sort of jealousy, that plays a role here. But not like *I want it easier* but rather *I dont want that others have it easier*. I dont know how to understand the large fraction of *I dont use bots much but please nerf them for the others*-opinions otherwise.

Even more personal (and provocative) opinion:
This is an excellent example of mass manipulation. The same balance worked fine for years, but then you give people a bad feeling about it and suddenly there is a huge uproar. I see some striking similarities with behavior of populistic politicians: You create a split into subgroups (bot-lovers vs bot-haters) and put them in opposing corners. You give a subgroup a bad feeling about what they have and what they do. You incite some jealousy. You suggest that restricting the other subgroup will help your own. You present selected points of view, leaving out other crucial information (like the fact, that you have to play through the whole main game of belt mess to get to the point, where you can afford such a base).
Concerning the last question of KOUB: It would not work with a *nerf Belts* outcry. As in politics, this works best on those, who do not know too much about the other subgroup. (Like: You want to incite fear towards foreigners? Best targets are those, who never saw a foreigner in their life!). Everybody knows belts. I bet every mega-baser had built several larger pure-belt factories as well. That why it does not work the other way round.

In contrast to the said politicians, I would not assume, that Twinsen created this effect intentionally. I guess it comes from loading savegames instead of building the base yourself: You can jump into a terrain without really knowing it. Nevertheless, it was a not helpful and fair way to start the debate.

[bobucles]

I do think that loaders are essential for helping players achieve end game throughput on belts. But what kind of functions should a loader have? Let's look at the options here.

- Loader from chest into assembler: Absolutely not. This is why inserters exist and they do the job very well. Loaders don't need to touch assemblers.
- chest into chest: Highly exploitable. Chain link chests together for insane speed and throughput.
- chest into belt: Slightly less exploitable. Chain link a chest to a belt to a chest for exceptional speed and buffering potential. Chaining stack inserters with chests is probably faster though.
- Belt into train: Aww yiz. Train depots are always a tangled mess trying to give the best throughput. Loaders can streamline the inserter/chest/inserter process by taking items, holding them, and plopping them into a train. This isn't very meaningful for train input, but it's extremely useful for unloading trains to get the best output.
- Train to train: The current loader design is 1x2. Inserter/chest/inserter setups are always an ODD number of tiles (1,3,5). Train lines are separated by an even count of tiles (2,4,6,etc.). This simple geometry makes loaders a perfect match for dunking items between trains. It's nothing more advanced than "They simply fit". Train => train transfer is crucial for developing a mixed network where players use a mix of huge throughput trains and small delivery trains.

So there you have it. Loaders would be great for solving and simplifying train problems. Could loaders become the only option for handling train items? It wouldn't necessarily be a bad thing. After all you get a dedicated item that can use a special animation for loading trains. A gigantic crane or something could look a lot cooler than a dozen inserters jamming items into the train doors.