Logistics Robot Balance

[Tairon96]

(*still-reading-thread-FFF225@p12*)

As far as I have read most arguments vs a bot nerf are summarised as N.I.M.B.Y.

Although I have read a lot of suggestions to handle this problem, I have not read an opinion of the Devs about any of these suggestions ... but I am only at p 12 so far.
___

Additionally I don't think, the bots need a nerf or the belts need a buff to compete with each other. The problem is situated at the interface between bots/other systems and the interface belts/other systems. This is the bottleneck that can be evaded. If bots are always even or better in this competition, belts will always lose.

(*continuing-to-read-the-other-39-pages*)

Edit:

A few pages later, I have found this smart post which sums it up ...
Shados' summary on 'Bots vs Belts'

[5thHorseman]

do you realise how complex the train path finding system is?

How many trains does the typical base have compared to logistic bots?

[bobingabout]

do you realise how complex the train path finding system is?

How many trains does the typical base have compared to logistic bots?

Lets say, a dozen trains vs a couple of hundred logibots (and that's being generous, mine has about 10 thousand)
Trains can afford to have a path finder, robots cannot.

[5thHorseman]

do you realise how complex the train path finding system is?

How many trains does the typical base have compared to logistic bots?

Lets say, a dozen trains vs a couple of hundred logibots (and that's being generous, mine has about 10 thousand)
Trains can afford to have a path finder, robots cannot.

Aha! I misunderstood your post. I thought you were saying trains were complicated so why not bots.

[bobingabout]

do you realise how complex the train path finding system is?

How many trains does the typical base have compared to logistic bots?

Lets say, a dozen trains vs a couple of hundred logibots (and that's being generous, mine has about 10 thousand)
Trains can afford to have a path finder, robots cannot.

Aha! I misunderstood your post. I thought you were saying trains were complicated so why not bots.

robots following a track would need to be as complex as a train. so unless you did some strange magic like, making them effectively be belts, it's not a good idea...
wait... now I've just reduced bots to belts, making them pointless.

...
LEAVE ROBOTS ALONE!!!

[mystik]

I would like bots to be less intelligent and not work so much without any setting from player.

Few ideas:

1) Need to configure logistic lines between chests. Robots will not transport anything from any chest to any chest. Instead you have to manually set up logistic line between chests if you want items transported between them. Robots will only fly along these logistic paths. This will also reduce CPU load required by bots because they will only fly along user defined roads so no need to find roads themselves. These will also give player better insight to what logistic network do and what increase use of storage and buffer chests to simplify logistic lines (if you have 5 provides chests and 5 requester chests now there is 25 available lines, with this change player would probably create intermediate buffer chest and only make 5 lines from providers to buffer and 5 lines from buffer to requesters)
2) Previous version can also be made even more complex by assigning logistic robot to specific line. So robot will not fly just on any line where is needed but will just follow assigned logistic line.

[Zavian]

... Robots will only fly along these logistic paths. This will also reduce CPU load required by bots because they will only fly along user defined roads so no need to find roads themselves.

What you propose will increase cpu usage, not reduce it. Atm the bots do not do any pathfinding. They do not "find roads". They just fly straight towards their destination. The only realistic way to make them more cpu efficient would to allow them to carry more per trip, and hence to make less trips overall.

[mystik]

They fly straight line between chests but they have to decide a) to which chest go for pickup b) to which check go to drop off c) where to fly next. Now every bot have to choose between all possible combinations of source-destination chests. User defined logistic paths will reduce CPU load needed to compute and compare all possible travel paths between chests bot have to do every time its finished delivering item. It is not about pathfinding in sense which way to go between chests but it will simplify pathainding in sense of deciding order of chests travelled.

[voddan]

A User:
- The game is a bit unbalanced and weird to play in some scenarios

The Devs:
- We already know about it! We get this comment like twice a week, so shut up already!
If we wanted to fix it, we would, but we didn't, therefore we don't.
We have explained why won't ever change it. Nothing have changed since then, there is no new argument you could come up with, so no reason to have this discussion. Go read some comments on the forum.
Even if it would make sense from the playing perspective to re-balance this part, we couldn't possibly do this because PERFORMANCE! No, we didn't actually measure the performance impact or try to optimize, why do you ask?
Actually, if you have any problem with the game, here is a mod to change it in the gameplay. Basically, mods create separate games that look like Factorio but are totally different in any way you want. You have very strong opinions, maybe you should play another game then, Fortnite or something.
You can even write a mod yourself! Go write a mod!!! Your opinion does not matter until you write a mod. Yes, mods are in Lua, no, they can't access the core game mechanics or be as fast as a C++ implementation.
TL; DR: This is a feature, not a bug. Everybody likes it the way it is. If you hate it so much we suggest you ignore this part of the game.

[weaknespase]

Devs have a vision of their game and building it according, they also consult with community on dubious moments and fix bugs. They even extend modding API regularly, so we can implement our dreams better and with better performance, what are you else asking them for?

The bots are okay as they are because they are different from other logistic options in the game, and i'd hate laying paths for them manually, because that doesn't make them any better or more fun than belts.

[McDuff]

A User:
- The game is a bit unbalanced and weird to play in some scenarios

The Devs:
- We already know about it! We get this comment like twice a week, so shut up already!
If we wanted to fix it, we would, but we didn't, therefore we don't.
We have explained why won't ever change it. Nothing have changed since then, there is no new argument you could come up with, so no reason to have this discussion. Go read some comments on the forum.
Even if it would make sense from the playing perspective to re-balance this part, we couldn't possibly do this because PERFORMANCE! No, we didn't actually measure the performance impact or try to optimize, why do you ask?
Actually, if you have any problem with the game, here is a mod to change it in the gameplay. Basically, mods create separate games that look like Factorio but are totally different in any way you want. You have very strong opinions, maybe you should play another game then, Fortnite or something.
You can even write a mod yourself! Go write a mod!!! Your opinion does not matter until you write a mod. Yes, mods are in Lua, no, they can't access the core game mechanics or be as fast as a C++ implementation.
TL; DR: This is a feature, not a bug. Everybody likes it the way it is. If you hate it so much we suggest you ignore this part of the game.

This is such a weird comment given that the mods listen to and implement ideas all the time. Part of the issue is that I think people don't realise "the community" has a lot of people in it who either don't care about the thing they care about, or do care but in the exact opposite direction, i.e. really loving the feature you think sucks or vice versa.

[GrumpyJoe]

Another idea I had was the robots being ground based and going along a player set track. This wouldn't need the same pathfinding.

you mean, like.... belts?

[McDuff]

I know this has been done to death but I always thought the idea of subnetting bots deserved a better look. Wasn't even mentioned in that massive overview post! *insert ironic crying emoji here*.

For what it's worth, the pitch that got me into Factorio from another player was "you build a factory to make robots and get the robots to build you a bigger factory," so they're a key part of the game for me and I love them, while thinking they could definitely be better.

The basic idea was to make bots:

• Good at very short range sorting, organising etc
• Bad, but still capable, of shifting items across medium (belt scale) distances

Overview Diagram

Factorio Subnets.png (47.94 KiB) Viewed 4071 times

What I came up with was to break the system down into subnets. The diagram explains it better but basically each Roboport operates as a sub network. If there is a Request, the system looks first in its local subnet and then in the bigger network. In order to move between networks, each roboport has a built in "buffer" (not to be confused with the existing buffer chest). Robots first deliver to this, which is the only place from which items can be delivered to another roboport.

Network to network

Factorio Subnets-Page-2.png (20.42 KiB) Viewed 4071 times

There would need to be some kind of IP-style routing between nodes, but you still don't have to deal with collision detectors or corners. Your "bandwidth" is limited by how many robots can operate in one subnet at a time.

Interestingly, the "optimisation" play here is in making the coverage area of individual subnets smaller so that you can pack more of them in, keeping your robots super short range.

The way I thought of doing that would be to get each subnet to shrink any neighbours, like so:

Clashing subnets

Factorio Subnets-Page-4.png (13.11 KiB) Viewed 4071 times

There could be more complex ways of doing it, such as having something which defined the edges rather than the centre of the network:

irregular subnets

Factorio Subnets-Page-4b.png (41.88 KiB) Viewed 4071 times

That kind of thing is a "whatever works best" thing though.

Even though you could still probably saturate the connections between roboports, it should make using robots to transport things any distance longer than a couple of chunks a much more arduous task, while still keeping them really good for what is, in my opinion, their perfect use case: "last mile" transport and high density sorting and storage.

[Darinth]

Please see my previous post about expecting that in order to have any vague chance of being implemented, the changes to drones would have to be relatively speaking 'tweaks'. I doubt we're going to get a complete redesign of how drones or their networks work. We might be lucky enough to get something like a limit on how quickly drones can get stuff to/from chests. Any kind of a more extensive overhaul is probably either A: going to fall upon mods, or B: we might get lucky and see a 'drone redisign' as one of the major post 1.0 patches... if Wube actually decides that drones are in fact OP and need nerfs... which I wouldn't hold my breath for. As I said before, there's a lot of people out there who either like drones just like they are or even want them to be more powerful.

[bobingabout]

Devs have a vision of their game and building it according, they also consult with community on dubious moments and fix bugs. They even extend modding API regularly, so we can implement our dreams better and with better performance, what are you else asking them for?

The bots are okay as they are because they are different from other logistic options in the game, and i'd hate laying paths for them manually, because that doesn't make them any better or more fun than belts.

It's amazing just how much influence I've had on 0.17. for the most part recently, I've just been reporting bugs
browsing through the game data and source code and just asking "Why is this here?" at least one of those has resulted in the devs doing exactly what I suggested.
And I've even written modded features, like the ability for entities other than the steam engine/turbine entity to be powered by fluids, without scripting to simulate it.

Okay, my level of access and influence is higher than the average user, but my point is, they do accept input.

[McDuff]

Bob, do you have a more detailed idea of how your "split the components up" plan would work?

Would it be something like this?

roboport pieces

Would it be something like this:

Factorio Subnets-Page-3.png (62.05 KiB) Viewed 4023 times

[McDuff]

Please see my previous post about expecting that in order to have any vague chance of being implemented, the changes to drones would have to be relatively speaking 'tweaks'. I doubt we're going to get a complete redesign of how drones or their networks work. We might be lucky enough to get something like a limit on how quickly drones can get stuff to/from chests. Any kind of a more extensive overhaul is probably either A: going to fall upon mods, or B: we might get lucky and see a 'drone redisign' as one of the major post 1.0 patches... if Wube actually decides that drones are in fact OP and need nerfs... which I wouldn't hold my breath for. As I said before, there's a lot of people out there who either like drones just like they are or even want them to be more powerful.

I know and I'm sure that we won't see any major changes to the underlying bot programming for 1.0. I'm just offering this up as I think it's a way of making bot gameplay better.