Why no faster trains?

[JackTheSpades]

Have the devs ever stated that they didn't want to have faster trains in the base game?

Obviously there are mods abundant that can do this but given the suspicious lack of base game means to make the trains faster.
In 1.1 Factorio there used to be a number of things that you get early and never upgraded, such as mining drills and research labs. Instead they would be upgraded through secondary means like research and modules that you unlock later.

Likewise, trains are in the same category.
Once unlocked they improve through research, better fuel and faster inserters for loading and unloading.

However...
With all the new additions that Spage Age has brought, such as higher tier mining drills and research labs I was hoping to see some changes for trains as well.
Except, no cigar. (Except the increased capacity of fluid wagons).

Sure, you can use higher quality inserters/pumps to speed up loading and unloading and even use higher quality fuel for some speed and acceleration bonuses.
But trains themselves? Nothing.
No speed bonus, no weight reduction, just the base line health increase which, frankly, I don't think is worth bothering with.
So many items got insane bonuses from quality with the whole tagline being "quality over quantity" yet trains get absolutely nothing?

And circling back to fuel for a bit.
For nuclear fuel (which your end game factory is most likely running on) you get a decent 250% acceleration bonus and an OK-ish 115% speed bonus.
Now for legendary quality nuclear fuel, this increases to an amazing 475% acceleration bonus and an still OK-ish 137.5% speed bonus.

With the new elevated rails and flying mech armor managing intersections and getting run over isn't nearly as much of a problem as it used to be with fast trains.
In the same vein, Breaking Force seems like an easy choice for infinite research yet never made the cut.

My point sorta being:
Faster trains feel like such an obvious late/post game upgrade yet have never manifested.
As such, the devs must have clearly thought about it at some point and decided against it... but why?

[coffee-factorio]

I can't speak for the devs. A few things come to mind... first off, what does a fast train that has to break do to trip planning? Over a short enough distance, what would a really fast train actually look like?

How big of a space should players work in (i.e. should they be encouraged to move towards the hard coded map limit)?

[mmmPI]

Have the devs ever stated that they didn't want to have faster trains in the base game?

Not something i ever heard of ! And i do read and hear about factorio a lot !

That being said i'd be more surprised to have missed something about a plan for this than the opposite.

My point sorta being:
Faster trains feel like such an obvious late/post game upgrade yet have never manifested.
As such, the devs must have clearly thought about it at some point and decided against it... but why?

Yes for the obvious x) !

But !

then i can only share the product of my imagination there, and try to explain what could be the reasons for going the not obvious way, albeit i have no idea if those have any weight for any actual decision x)

1)One of the "problem" with trains is that they require "pathfinding" which from what i understand is a "costly-one-time-request-that-need-to-finish-in-less-than-16-ms" to the CPU and that scale with the complexity of the network, because every time a train initiate a travel, it ask the game to analyze the "whole" network. But it also require such thing everytime it has to stop to a signal , or brake because a train is in front, and several other situation ....

2)This is then multiplied by the number of trains, because the more "trains" you have, the more "request to pathfinding" will occur.

3)And lastly, this would be multiplied by the speed of the trains, in several ways. Because "per-trains" request would be more frequent.

Those 3 points are counpounding into "negative reason" regarding perfomance. That's one way for me to "understand" why it wasn't done that quality train are "faster".

From another perspective, if a player has "super duper fast trains" , once those moves, they need to "reserve" a lots of rails ahead, because they would require a braking distance proportionnal to the square of their speeed, increaseing speed of trains would cause them to "occupy" more rails when travelling, and potentially reducing the number of them running at the same time because most the network would be "yellow" block, reserved for "super fast train".

Roughly reformulated : faster train wouldn't necessarily translate into "better logistic capacity" for the train network ( unlike increased braking force which has 0 drawback ).

Which makes me think : It is possible that the situation is one where there is NO trade-off where "it cost performance" for "player ease of use", there it may happens that due to the nature of the physical equation for train travels that it would "cost performance" AND ALSO be "more trouble for players" at the same time.

That is another way to explain, if it's not good for gameplay [ 1°] ( you research but your network capacity gets worse ) AND not good for performance [2°]( you research too much and then you start having little freeze when train moves) , then devs would probably not consider it something that should be done.

I have made some "simple" desmos graph that try to modelize which would be the "optimal" speed of trains that guarantees the "best" throughput. It look like this : https://www.desmos.com/calculator/qhgvzk6kip Hopefully it can illustrate how the "top speed of the train" is not the thing to max out ,rather it's the "density of trains", if you have 2 trains at 100 km/h, or 4 trains at 50 km/h, the same amount of "stuff" is being carried, but the later occupy much less rails, because of that braking distance being square of the speed. So maybe you could use 5 or 6 trains at 50km/h on the same network, instead of 2 at 100km/h. This is why the "best speed" for trains when trying to maximize network capacity is not the "max speed for trains". The same is true for cars in the highway !

To a lesser extend, the same is true for "quality fuel", so i maybe wrong about wether or not those are important considerations, the "faster train" & "better fuel" would multiply to create a counpounding effect too, maybe devs thought 1 way is "enough" and the 2 would be detrimental. But i'm pretty sure about faster trains decrease network capacity due to higher braking distance, i've seen other more precise graphs with more precise physics to explain more precisely the same thing there : 110072

[JackTheSpades]

1)One of the "problem" with trains is that they require "pathfinding" which from what i understand is a "costly-one-time-request-that-need-to-finish-in-less-than-16-ms" to the CPU and that scale with the complexity of the network

I don't think trains need to find the optimal path every frame.
At the very least I've seen trains take some questionable routes in a large train-network before to the point I had to investigate if my tracks were broken somewhere since they didn't take the "obvious" route.
It largely comes down to how things are actually implemented but at the very least you'd only need to update train routes every "now-and-then". The trade off then becomes that a train might initiate breaking when approaching a chain-signal before updating its routing and having to re-accelerate.

If you wanted to be super user friendly you could even make the train-update-frequency a user setting so that users could throttle it on mega bases.

Roughly reformulated : faster train wouldn't necessarily translate into "better logistic capacity" for the train network ( unlike increased braking force which has 0 drawback ).

Which is why I didn't just mention trains being faster but also the potential of researching more breaking force.
Or the idea that higher quality trains/wagons would have less weight, therefor getting more bang-per-buck out of their breaking force.

It look like this : https://www.desmos.com/calculator/qhgvzk6kip Hopefully it can illustrate how the "top speed of the train" is not the thing to max out ,rather it's the "density of trains", if you have 2 trains at 100 km/h, or 4 trains at 50 km/h, the same amount of "stuff" is being carried, but the later occupy much less rails, because of that braking distance being square of the speed. So maybe you could use 5 or 6 trains at 50km/h on the same network, instead of 2 at 100km/h.

Throttling speed of trains intentionally isn't really possible though.
You could use weaker fuel but then you'd sacrifice the acceleration bonus.

It also largely depends on what kind of network you actually have.
For example, if you have an ant-farm style grid where each cell gets its supplies via train, you probably don't need higher speed since any one train isn't likely to reach max speed anyway before having to break again.
But if you have long distance trains to get you to large and rich resources with not really much of a "network" so much as it has long straight lanes then getting higher speed would probably be worth it again.

[coffee-factorio]

1)One of the "problem" with trains is that they require "pathfinding" which from what i understand is a "costly-one-time-request-that-need-to-finish-in-less-than-16-ms" to the CPU and that scale with the complexity of the network

I don't think trains need to find the optimal path every frame.
At the very least I've seen trains take some questionable routes in a large train-network before to the point I had to investigate if my tracks were broken somewhere since they didn't take the "obvious" route.
It largely comes down to how things are actually implemented but at the very least you'd only need to update train routes every "now-and-then". The trade off then becomes that a train might initiate breaking when approaching a chain-signal before updating its routing and having to re-accelerate.

If you wanted to be super user friendly you could even make the train-update-frequency a user setting so that users could throttle it on mega bases.

Even if you are minimizing observation of a train to "start and stop", if you have about 40 trains at different stations you have to observe for a train to start, and do pathfinding. Once you go above a certain number of start and stop points planning where you are going, the number of possible routes you have to explore explodes. Once you're working with a large graph of interconnected rails that is not necessarily something you can optimize around, even though optimal play will make it so that once a train decides on a route technically the only thing it needs to be doing is looking for signals.

The issue is well known, getting it to run as well as it does in this game is considered to be extremely hard.

[FunMaker]

I would like to add, that increased breaking strength is not as straight forward imho:

• If a train approaches a crossing or something similar, where the reason for the stop is another train blocking the crossing that might leave the crossing in the next few seconds, the train with fewer breaking power might not reduce it's speed as far as a train with higher breaking power
• A train that approaches a station where it will stop, higher breaking power is better no matter what

I would introduce an endless research which increses acceleration in the same way like breaking power does, then the problem i described above would be gone.

[coffee-factorio]

A compromise might be better?

If a train goes (edit: accelerates) fast enough, it will "jerk" because it moves faster than the 1/60 fps can handle (or a tick in game terms). So there might be a sweet spot with research where it stops and delivers something pleasing. An infinite research might work as a way to find that point, for testing purposes. The train would also have to slow down and speed up, but in between it might be fun to mess with the idea.

[mmmPI]

I don't think trains need to find the optimal path every frame.
At the very least I've seen trains take some questionable routes in a large train-network before to the point I had to investigate if my tracks were broken somewhere since they didn't take the "obvious" route.
It largely comes down to how things are actually implemented but at the very least you'd only need to update train routes every "now-and-then". The trade off then becomes that a train might initiate breaking when approaching a chain-signal before updating its routing and having to re-accelerate.
If you wanted to be super user friendly you could even make the train-update-frequency a user setting so that users could throttle it on mega bases.

You can use the debug menu for train to "show repath" in game and see for yourself how often and when trains do repath. It shows some flying text next to the loco. The rules for repathing are written in the wiki otherwise. You're right it doesn't need to do so every frame, and it doesn't currently, but if trains were going faster and faster it would increase that frequency several fold which is adverse to perfomance.

It same wiki page also indicate the "penalty" for pathfinding which may explain why you see trains taking the "not obvious" route. When it's written "500 penalty" it means the pathfinding consider there are 500 more rails in distance, a red signal, or a train stop or a parked train or a broken train, those have different "penalty", the same distance on the map is not counted the same then, and the train takes the "shortest path" while considering those invisible rails as extra distance.

It look like this : https://www.desmos.com/calculator/qhgvzk6kip Hopefully it can illustrate how the "top speed of the train" is not the thing to max out ,rather it's the "density of trains", if you have 2 trains at 100 km/h, or 4 trains at 50 km/h, the same amount of "stuff" is being carried, but the later occupy much less rails, because of that braking distance being square of the speed. So maybe you could use 5 or 6 trains at 50km/h on the same network, instead of 2 at 100km/h.

Throttling speed of trains intentionally isn't really possible though.
You could use weaker fuel but then you'd sacrifice the acceleration bonus.

It isn't "really" possible, true, but you can do it "indirectly" or it can happen "by accident" or "as a result".

You can try to indirectly but intentionally throttle train speed by managing density of trains, if you create a simple loop of rails, and you put 1 train on it, it will run at it maximum speed. But if you add more trains, they will have to "share" and will eventually spread out evenly if the signals are also somewhat spread out evenly. You can even calculate which is the speed at which they will balance out because it will be based on their braking distance, and the amount of space left in the loop in-between the trains. I found it funny to do in editor and did many many tests x).

That's more a theory thing, in practice, it just "happens" sometimes not in controlled way that trains do not have time to accelerate fully and are constantly having short brakes which end up leaving them with a total average speed from point A to point B which could be very different than their "max speed" , even if they never actually had a full complete stop.

I would like to add, that increased breaking strength is not as straight forward imho:

[*]If a train approaches a crossing or something similar, where the reason for the stop is another train blocking the crossing that might leave the crossing in the next few seconds, the train with fewer breaking power might not reduce it's speed as far as a train with higher breaking power

I would introduce an endless research which increses acceleration in the same way like breaking power does, then the problem i described above would be gone.

I agree ! It's a phenomenon that can cause train that brakes much faster than they accelerate to sometimes underperform compared to trains with worse stats but more balanced ratio, i believe this isn't "too" significant in the current game, but that's also because "braking force research" isn't currently an infinite research.

If both acceleration and braking force were inceasing in the same way, it would be better to mitigate that particular risk i also agree.

That wouldn't unfortunaly help for the performance cost of increase frequency for pathfinding request, because each train will be allowed to potentially clear more junctions in the same amount of time, on average if they are faster. ( the better this works, to make train faster on average effectively in a game, the worse it will increase the frequency of request for pathfinding ).

I'm not saying i have the correct explanations for "why no faster trains", i'm stating what i think are tradeoff that are at play.

It also largely depends on what kind of network you actually have.

Indeed it 100% depend on your network, if you have isolated loops with single train on them none of this matter x)

I think with the "galaxy of fame" now it's possible to have an idea of the structure of the train network "players that have finished the game" uses, just picking randomly stars and looking at them, i haven't made any stats just being curious. There are many ways to do, but also many ways end up in a "normal distribution" sort of, where there are many similar in the middle and some extreme on each side. Mostly considering "complexity" and "span". The "grid-world" constituting one of the not so many very different structure. That also is the same from what i saw when joining multiplayer server. Not sure what's used otherwise to consider "what players do.

There are also minor reasons, like "faster trains are more dangerous"