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

[steinio]

We need this between belts and trains: https://youtu.be/4mAVdAJPTu8

[Optera]

We need this between belts and trains: https://youtu.be/4mAVdAJPTu8

That would make a good less than yellow belt throughput option.
For train-belt connection its throughput is way too low.

[McDuff]

OK. So. I've been thinking about this and here's my thoughts on it.

Factorio is a fractal game. The massive outputs and 1000 rocket-per-minute builds all come down to the interactions of thousands of simple and individual components which are arranged in a complex branching tree structure.

The robot/logistics system flattens out that tree, which is why people think it's not as interesting. Problems become easily soluble because there aren't as many layers and stages.

There is also the issue that you can just cram a load of beacons and roboports into a small space and get huge amounts of throughput, as demonstrated in the dev diary. This feels like a hack or an exploit even if it's not.

I think the devs have been working towards a solution with the filtered storage/buffer chests, but they're halfway there.

I would suggest, rather than changing the bots, instead change the logistics network.

• Establish a tree of subnetworks within the logistic network, centred around individual roboports.
• Storage and buffer chests form the large logistics network.
• Requester and provider chests are only accessible from their nearest roboport.
• Logistic robots only travel into adjacent subnetworks to pick up or drop off items from storage or buffer chests.
• If a requester chest wants something that is not in the subnetwork or adjacent storage, the item has to be transferred in multiple "hops" across subnetworks, dropped off in a buffer/storage chest and picked up again.
• Subnetworks don't overlap. If you move roboports closer to each other it reduces the size of their coverage area instead. This introduces a trade off of small-fast, efficient subnetworks vs more hops to move items across them.

There are a few other changes from the thread that I think would also work to add gameplay that the robot changes would lose.

• Faster/non-acccessible belts. The belt system naturally "branches" - blue splits to red, red splits to yellow. But you don't really use it because you just spam the fastest belt once you're producing enough blues. Inaccessible belts work as a mechanism for firing stuff at speed across medium distances that would be far too slow with logistic system "hops", while providing a need to think about breaking them out and distributing them at the other end.
• Get rid of beacons and replace them with more tiers of buildings and modules. This a) reduces the tendency towards one boring "minmaxed" layout and b) gives you more room to play with belt systems.
• More control over belt distribution. Loaders, even if they only go into chests. Splitters which can be filtered on both sides, lane balancing/filtering etc. Long-handed versions of all the inserters. I appreciate that some people love messing about with side loading onto undergound belts and such, but to me that always feels like a hack of a game mechanic rather than a way that it makes sense to design things.

I think these changes together would go some way towards getting the game the devs want to make, while not "nerfing" bots to make them undesirable as much as changing what they're for. They should be for moving things from arbitrary A-B points and doing complex little tasks, not for bulk shifting continuous streams of iron ore. Further, this would add gameplay because you'd have to start managing your storage better, setting up buffers and even distributions of resources, and thinking about distances between roboport cells for moving things across the map.

edited to add a diagram

Untitled Diagram.png (31.12 KiB) Viewed 9613 times

Pink arrows are requests made within one robosubnet. Orange arrows are requests made across subnetworks. Each arrow represents a single robot trip. The areas aren't all square - the two rows of roboports on the right are closer together and therefore smaller. Hope this makes things clearer if they weren't already.

[McDuff]

And, yes, I'm aware that this change basically amounts to "implement TCP/IP in the logistics system" and is therefore significantly non-trivial, so I could accept the devs saying that it's too hard

[Gryzorz]

My thoughts:
- make robots collide, preventing a too high throughput on a single point.
- implement a PORTAL BELT, that basically works like an underground belt, except it teleports items.

To me:
- robots should help simplify the logistics of non-high-throughput items
- they should not replace belts for high-throughput items
- belt main challenge is the crossing (planar graph challenge)
- portal belts would ease that a LOT

Hope it helps

[Engimage]

My thoughts:
- make robots collide, preventing a too high throughput on a single point.

Robot collision will consume too much processing power to even be considered adding to the game. So it will not.

- implement a PORTAL BELT, that basically works like an underground belt, except it teleports items.

Such mechanic will remove the main selling point for belts - creating puzzle. Factorio is the game that forces you to think and optimize your design by introducing certain limitations for you. If limits are removed there is no puzzle to solve.

[Pascali]

Such mechanic will remove the main selling point for belts - creating puzzle. Factorio is the game that forces you to think and optimize your design by introducing certain limitations for you. If limits are removed there is no puzzle to solve.

Yes! Factorio is the new puzzle. The game has the chance to get all the offline mechanical puzzlers in front of a screen and puzzle there(having more fun). But bots are getting them away...

[Aeternus]

Without a botnet, the only viable way for a megaplant to function would be to compartmentalise - mass produce every subset of product in a seperate location, and connect the whole thing up by a hyperbelt, better known as a rail grid. Logistics Bots have their time and place, but at present their ability to roam free across the entire map is what makes them overpowered. Their main drawback used to be high power consumption per unit moved, but with nuclear power that's pretty much a goner.

The issue with the LogBots, the way I see it, is their range. Interconnecting the roboports to form wider zones makes for great flexibility, but it kills the complexity.

Potential solution: Allow Logistic Bots to only operate in the zone of the roboport they are docked in, and have them be unilaterally tied to that port - only the Construction Bots would get free reign.

This would give all forms of transports pro's and cons:
- Belts(/Pipes) offer moderate throughput and speed without power across a fixed path
- Trains offer huge volumes across large distances, but require infrastructure to build their tracks, and to load/unload them.
- Logistic Bots offer high flexibility and throughput but low range.

[Pascali]

Logistics Bots have their time and place, but at present their ability to roam free across the entire map is what makes them overpowered.

Overpowered AND BORING - cutting off the puzzle-fun. The thing, why everybode starts and loves factorio. But a lot of forget...

[Deadlock989]

Are we still flogging this dead horse?

"Bots are boring". Go and do something else then. No-one's holding a gun to your head. This is one of the best and most open-ended sandbox games ever made.

Bots are fine. Belts are very slightly underpowered. Loaders would have helped a lot, but people made a song and dance. Look at Space Engineers for a game that went south, fast, because the developers listened to the players. Players, en masse, are absolutely useless game designers. Just look at the some of the Harry Potter suggestions here. Belt portals. Quantum belts. Tesseract type deals that would make even a Minecraft modder says, ooh, dunno, bit dodgy.

They've said they're going to sit on it and think about it and I reckon it's likely we'll get a rework of loaders and maybe working with stacks. Job done.

[ratchetfreak]

Are we still flogging this dead horse?

"Bots are boring". Go and do something else then. No-one's holding a gun to your head. This is one of the best and most open-ended sandbox games ever made.

Bots are fine. Belts are very slightly underpowered. Loaders would have helped a lot, but people made a song and dance. Look at Space Engineers for a game that went south, fast, because the developers listened to the players. Players, en masse, are absolutely useless game designers. Just look at the some of the Harry Potter suggestions here. Belt portals. Quantum belts. Tesseract type deals that would make even a Minecraft modder says, ooh, dunno, bit dodgy.

They've said they're going to sit on it and think about it and I reckon it's likely we'll get a rework of loaders and maybe working with stacks. Job done.

bots are also fast and easy to setup.

That wins big time over big belt systems that is either tedious or blueprint stamped for bots to build.

[Deadlock989]

That wins big time over big belt systems that is either tedious or blueprint stamped for bots to build.

So what? Boo hoo. You're bored. Your problem.

[Punyaan]

If you want to nerf bots just add some kind of routing table to roboports, where player MUST choose how many bots must serve in every manualy created route.
Player would be able to solve new logistic puzzle, like managing spagetti belt lines, or managing railroad lines. That would move bots from dumb tool(more roboports and bots for the UPS god) to smart tool where player can choose what he wants bots to do.

[Lubricus]

If you want to nerf bots just add some kind of routing table to roboports, where player MUST choose how many bots must serve in every manualy created route.
Player would be able to solve new logistic puzzle, like managing spagetti belt lines, or managing railroad lines. That would move bots from dumb tool(more roboports and bots for the UPS god) to smart tool where player can choose what he wants bots to do.

That would make the bot's much more powerful and at the same time very tedious to set up. So that would be bad.

[Pascali]

Players, en masse, are absolutely useless game designers.

The factorio-crew has another opinion to this.

[Deadlock989]

The factorio-crew has another opinion to this.

That is literally the problem.

[Pascali]

...and that´s said by the game designer? Or aren´t you a player too?

[Aeternus]

Players, en masse, are absolutely useless game designers.

The factorio-crew has another opinion to this.

So does this game's modding community. Wouldn't call Angel and Bobing "useless" on this forum... the pitchfork-and-torch wielding crew would be upon me in a hurry

[Deadlock989]

Which bit of "en masse" was hard to understand? I wasn't talking about individual players or modders. I'm talking about mass brain farts like this thread. A game developer would be insane to take much notice of it.

[McDuff]

That wins big time over big belt systems that is either tedious or blueprint stamped for bots to build.

So what? Boo hoo. You're bored. Your problem.

It is quite literally a problem that the devs themselves have raised.

Maybe it's not really a problem or maybe it is. Maybe the solutions are a few tweaks to bots or maybe it's a sign that something needs to be adjusted on a fundamental level. Either way the conversation was started and people have found it interesting to expand their thoughts on it. If you don't like it, might I suggest not reading it? Perhaps instead find a nice book and sit outside in the sunshine?