(Oversimplified, ignore please) Make chain signals the norm, and rail signals for experts

[IronCartographer]

TL;DR

Reverse the sorting order of the two signals in the crafting/filter menus.

signals.png (12.32 KiB) Viewed 5384 times

What ?

Introduce chain signals for the entire network (forcing trains to reserve the entire path) as the simple, guaranteed deadlock-free (train won't move unless it can go the whole distance) signaling method, with rail signals terminating the chain as an advanced optimization which can introduce deadlocks with the benefit of additional throughput.

Failing that, it makes sense at least that chain signals look ahead to the next signal, and putting them before rail signals would fit this outlook (L-to-R reading bias notwithstanding).

I would like to see a version of the rail tutorials approaching signals from this perspective, introducing chains first, then rail signals as the more advanced (and potentially dangerous) topic.

Why ?

Factorio players learning the rail system for the first time often build systems that work and then suddenly stop working when trains get stuck facing each other in a deadlock. Furthermore, the chain signal seems mysterious and confusing, and it's not clear where or why they would be useful.

If the chain signal reserving an entire path were the norm, then rail systems would at least work with any number of trains (as long as there were enough free paths and stations). Throughput would decay with time, incentivizing the learning and use of rail signals to mark safe-blocks.

[Ajedi32]

That's actually a really interesting idea. At first glance the behavior of chain signals does seem more complicated than regular signals, but after thinking things through I can't really see any obvious issues with your proposed approach.

Thinking about things from that perspective, here's a summary of how I'd explain things to new players: chain signals divide rails into segments, regular signals do the same but also, additionally allow trains to enter the marked segment even if they might not be able to leave that segment immediately. This can improve efficiency, but also can cause deadlocks. Be sure to leave enough room in segments marked with regular signals so that trains waiting at those signals won't block paths that another train heading to a different destination might want to pass through (such as intersections).

[MEOWMI]

This could be a really nice thing to have in the tutorial! A bit of variety helps you check your understanding.

[NotRexButCaesar]

This actually makes a lot of sense! The only reason rail signals are useful at all is to increase throughput, and a brand new player probably doesn't really care about that.

[ssilk]

Indeed quite interesting idea. It really would lower the learning curve.

But I’m not 100% sure. Because when you see chain signals on junctions, this picture changes. It needs to be tested, if that would make it really easier to understand.

[McDuff]

Over long distances, this would crush throughput, because a train could "reserve" a junction five minutes away from itself, meaning another train which could make a trip through the junction 30 might have to sit at a station for ages while that one completed its trip.

Try setting up a system with all chain signals! It's not good.

[Sad_Brother]

Try setting up a system with all chain signals! It's not good.

I wander how chain signals work without rail signals at all?
It seems they also look at the stations.
Does anyone know?

[Qon]

Have you tried a chain signal only rail network with with trains with just locomotives pointing in a single direction? Does that actually work?

[posila]

Does that actually work?

It doesn't (for example if you have 2 trains that go between two same stations, they'll block each other forever)

[coppercoil]

OT
How about railway pathfinder v2.0 that would be deadlock-resistant? Please don't say it would cost many CPU and RAM, because it would work only on small factories (say, until chain signal technology researched )

[Qon]

Does that actually work?

It doesn't (for example if you have 2 trains that go between two same stations, they'll block each other forever)

Yeah, thought so. Op might have to rethink his suggestion a bit.

OT
How about railway pathfinder v2.0 that would be deadlock-resistant? Please don't say it would cost many CPU and RAM, because it would work only on small factories (say, until chain signal technology researched )

So a new complicated feature that stops working after tops 5 more minutes of research? And that teaches new players the WRONG way about how the signals work? Seems not only like an incredibly expensive waste of a lot of dev time, but also something I would consider paying them not to do (not needed with Wube ofc, they know their stuff) because of how wrong that would be.

[coppercoil]

So a new complicated feature that stops working after tops 5 more minutes of research? And that teaches new players the WRONG way about how the signals work?

Ok, how we can improve this? I already have new ideas, how about you?

[lois lane balancer]

OT
How about railway pathfinder v2.0 that would be deadlock-resistant? Please don't say it would cost many CPU and RAM, because it would work only on small factories (say, until chain signal technology researched )

Already implemented, direct player control of trains is very deadlock-resistant and works only on small factories!

[Sad_Brother]

OT
How about railway pathfinder v2.0 that would be deadlock-resistant? Please don't say it would cost many CPU and RAM, because it would work only on small factories (say, until chain signal technology researched )

Already implemented, direct player control of trains is very deadlock-resistant and works only on small factories!

Exactly. Seems no other way.

[Qon]

Ok, how we can improve this? I already have new ideas, how about you?

I don't know how to improve your suggestion, no ideas at all. It seems like a suggestion that is fundamentally unimprovable as long as it stays with the identifying feature of working before some research. Completely new ideas starting from scratch, scrapping your idea, possible. Improving that one of yours, no.

It's not clear from your message though if you are talking about improving how rails are taught or improving your idea about changing how signals work before some signals are researched.

[coppercoil]

I think we solve more general question: can we make railway signalling easier?
And derivative: can we make it easier for new players?
And the second derivative: can we make it withough significant redesing? I started here, but we can go deeper higher.

How about some third king of signal, kind of "signal for dummies"? It would have limitations/incompabilities to force learning for more complex signals. Some players don't want to solve signalling puzzles at all, they just need a simple railway (though not a single bidirectional line).
How about some early rails without any signalling? I know there are mods, but how about deadlocks?
These ideas are incomplete, just a try to move on.

[steinio]

Haha yeah some don't understand 2 signals so you want to add a third.

In my eyes it's a slight level of retardness if anyone doesn't understand rail signals or mostly they don't want to understand it and
everyone else should solve their problems.

[IronCartographer]

Revision is indeed in order. Admittedly I went too "pure" with the idea intending it to be simple.

In testing the concept I originally used rail signals only for the start of a train stop area. Later I found it worked with entirely chain signals (working with new players and demonstrating the behavior to them), but did not add enough trains to encounter this blatant flaw in oversimplifying. Entirely chains = deadlocks if you create as many trains as train stops for a given route.

Even so: It would be educational for people to learn what happens in the case where rail signals are the exception, causing a different focus and highlighting the overall pattern by contrast.

Perhaps some value may yet come of the concept; it is still simpler to set up basic networks with entirely chains, and slightly more advanced ones with rail signals marking the safe-stop locations...but given the awkwardness of switching to rails at that scale I can see it is a trap in a real factory's development unless the rail is / signals are to be replaced.

[Squelch]

I've been following this thread with interest, and every time I visit it, the same thought strikes me - Why make chain signals the norm?

Rail signals are simple, and prevent a train from entering the block ahead if it is occupied or has been reserved by a faster/closer train. There's no confusion brought by junctions and crossings, so whilst a train may stop and start along its journey to allow the blocks ahead to clear, the possibilities of deadlocks are pretty much dealt with except for complex junctions.

Chain signals are not quite so simple. They extend the block ahead by as many chain signals in the chain plus the block ahead of the last. They also allow conditional branching at junctions. This last bit seems to be the sticking point for new players. The conditions for switching at junctions are not always obvious, and rely not only on the occupation of a block potentially some distance ahead, but also any reserved blocks between the train's current position and the final block in the chain at the time it enters the block guarded by the next chain signal.

The simplest, or even more complicated layouts can be made with rail signals alone. Trying the same with only chain signals brings a whole lot of pain. Trains do not need, nor should be ascribed with any kind of "intelligence". They simply follow stop and go instructions provided by the signals in the most simplest of terms. Yellow signals instruct the train to slow because the block ahead has been reserved by a faster train (this usually allows one train to pass ahead another at crossings). Blue signals do not indicate a choice to the train, they merely indicate that one of the possible paths is blocked, and the train will take the alternative route based on the signals ahead on that branch.

I respectfully propose that it should be "standard" rail signals that should be the first introduction to signalling, and used exclusively by new players as the norm until they wish to progress into more complex signalling and track arrangements. Chain signals and complex junctions are the "expert mode". I really do feel that chains signals are at the root of much confusion. It might be the language and terms used when describing them, or how commonly other players demonstrate their creations that lead new players into thinking that a peppering of chain signals cures all signalling ills. Whatever it is, it's a perennial problem, and won't be solved by trying to add yet more programmatic complexity on top.

[Koub]

[Koub] Seeing OP requalified the thread as "please ignore", I'm moving this to "Won't implement"