The logistic System is a gigantic mess

[KingHalford]

ORIGINAL: Nemo84

Just because you haven't needed it doesn't mean it's not an issue for others. "You're playing it wrong" is the lamest excuse for a poor design decision possible.

I'm in my first beginner-difficulty game and without constant tweaking of these traffic signs my empire simply grinds to a halt. Especially with the AI's road spaghetti.

There is absolutely no reason why a lvl 1 farm should draw 5000 trucks from the resource pool while an entire army 10 hexes over is running out of supplies.

I very much support a setting that just draws the minimal logistics points needed by the connected assets.

The Traffic Light System exists to fix issues like this. It's not something you need to do every turn, you do it once, it takes three seconds max, and then it's done.

If you're having to go and change it again every couple of turns, I think you probably need to build more truck stations, as I said.

[springel]

ORIGINAL: AttuWatcher

I think a demand based system is more elegant in the end.

Does anyone have an actual good argument against it?

Yes, it is very hard to plan correctly. Think Soviet style failures where they planned everything from the top, without sufficient computer software.

The system is this game is much like the structure of transport here in NL until the late 20st century: transport was carried out by many small companies, each servicing their own route. This system grew out of a system of cargo boats in the 19th century and before. These systems of companies cooperated with big central cargo stations in the city, where they were all parked until they departed at a certain time. I went there every Friday with my mother to deliver packages to all the villages in the neighbouring provinces, as my parents had a printing office that made flyers for groceries of a large organisation.

Only after the 1970's this system disappeared and was replaced by larger transport companies who could use large computer systems and distribution centres to handle everything, but the old system worked fine without requiring a huge planning organisation. Each line was a single company with a smaller or larger or multiple trucks, adapted to the load of the average traffic.

The system was more flexible in a way than the big integrated system, as you could deliver your package at the station just before departure, and the package would be delivered an hour later at the destination. The big systems that replaced it don't offer that kind of speed without a large premium.

[DeltaV112]

I'd agree that points spreading arbitrarily isn't very good design. It's manageable by as people suggested building more truck stations but at the same time that wastes resources meaning the player is incentivized to spend lots of time fiddling with the traffic signs to maximize the throughput for their logistical network, especially around major trunk resource lines that require lots of points to function properly.

Having said that I can see how it might be necessary- an optimal solution is non-trivial, because it might be necessary to route resources in weird ways in order to get as much as possible out of all the logistical assets. I don't think it would actually take up all that much turn processing time- modern comptuters can crunch a huge volume of data, and also this family of problems(transportation problems) is well-characterized enough that efficient optimal or near-optimal solutions exist. The issue would be the complexity involved in creating a solver that was reliable and fast.

My initial tack for how a solution would work based on my knowledge is something like this: take all the logistical centers on the map. Establish their base cost as the inverse of the number of points they have(our goal is to minimize the occurrence of a logistical center being used to capacity, so we try to use the smallest fraction possible). Then, calculate the cost to reach the logistical center with one point from the SHQ and add it on to the cost of that logistical center, along with the route used. This will change all the costs, but because the number of logistical centers is small, we can iterate it several times to get a near-optimal solution. From this point, for each logistical stage you can calculate the optimal routing to minimize the total network usage, and then check to see where if/where you are hitting the ceiling and limit throughput on those routes.

Complicated, not insoluble. Worth it? Maybe. That's for Vic to decide.

[Dampfnudel]

ORIGINAL: KingHalford

ORIGINAL: Nemo84

Just because you haven't needed it doesn't mean it's not an issue for others. "You're playing it wrong" is the lamest excuse for a poor design decision possible.

I'm in my first beginner-difficulty game and without constant tweaking of these traffic signs my empire simply grinds to a halt. Especially with the AI's road spaghetti.

There is absolutely no reason why a lvl 1 farm should draw 5000 trucks from the resource pool while an entire army 10 hexes over is running out of supplies.

I very much support a setting that just draws the minimal logistics points needed by the connected assets.

The Traffic Light System exists to fix issues like this. It's not something you need to do every turn, you do it once, it takes three seconds max, and then it's done.

If you're having to go and change it again every couple of turns, I think you probably need to build more truck stations, as I said.

You also need to change the traffic signs every time asset upgrades as it now requires more Logistical Points.
Now imagine having an empire with hundreds of assets upgrading all the time all connecting by hundreds of roads and intersections.

[KingHalford]

Transportation problems are very well documented, and they're documented to be extremely cost heavy. I like this possible solution ^^ but imagine that running for 60+ factions each turn?

[ramnblam]

I think a cool solution to these issues would be the ability to tell your economic council to optimize your road and logistic network based on demand at a cost of BPs. So you can do some fine tuning and manual adjustments as needed but if it becomes a mess the ability to get your underlings to sort it and deliver a good but perhaps imperfect logistic network.

[Nemo84]

ORIGINAL: KingHalford


The Traffic Light System exists to fix issues like this. It's not something you need to do every turn, you do it once, it takes three seconds max, and then it's done.

If you're having to go and change it again every couple of turns, I think you probably need to build more truck stations, as I said.

No, it means I'm constantly adding logistics capacity and conquering more of the AI's road spaghetti.

An asset needs a certain minimum of logistic points to function (100/level). If each turn the logistics capacity of the main road increases, the fraction needed in the branch to operate the asset decreases. Because I'm forced by terrain to support a large invasion force through a single supply line with branches to vital resource assets and hex bonuses, I'm constantly having to readjust the signals to make sure my strained logistics capacity is not strangled by diverting trucks to assets that don't need that many.

Likewise the region around the newly conquered AI cities at the end of that supply line is tangled with roads going nowhere. This dilutes the already tiny logisitics capability in the city so the surrounding countryside assets cannot properly supply it. This significantly lengthens the unrest in the city, increasing the build time on the city's logistics improvements and reducing supply to the units protecting the city from its former owners.

Again, just because you haven't needed it so far doesn't mean others won't. Yes, brute-forcing the system by doing nothing but spam truck stations the next 15 turns will probably solve my problem as the AI is too weak on beginner difficulty to properly counter-attack. But better control over the traffic system will solve my problem far quicker, more efficiëntly and in a far more fun manner.

EDIT:
I don't mind having to place traffic signs to direct the flow, as an intelligent logistics assignment system might be computationally problematic. But at the very least I need to be able to define a specific logistic value that can pass a sign instead of it being a percentage of a varying total. Combine that with a bottleneck overlay that clearly identifies assets with insufficient logistics to operate fully, and it's basically a good enough solution that's relatively easy to implement.

[DeltaV112]

Minors/nonaligned don't play by the logistics rules so you wouldn't have to run it for them. But OFC eventually the majors will own most of the territory and will make lots of assets and units that need to have their costs computed. The upper bound is essentially twice the number of hexes on the map, so reasonably the constraint matrix won't exceed several thousand rows in size. Likely it will stay below a thousand(imagine the time it would take you to shuffle a thousand units around the map).

Modern LP solvers running on desktop hardware can handle a LP problem of this sort of size in less than a second. Crunching lots and lots of numbers is easy.

[Cornuthaum]

ORIGINAL: Kamelpov

If you don't want to think build supply truck and max size everytime and put them at 8 hex from each other and call it a day.

who has the population/workers for that? that's the major problem I have with logistics, even with rail IV stations between all my cities, there just aren't enough workers (and I've sucked up all free folk in all my zones already)

[Emx77]

I also prefer demand based system. For example, Ageod in some of their titles (American Civil War, Revolution Under Siege...) used what they called advanced supply. It is very elegant system where each depot, city or army acts like magnet for supply produced all over a map.

[Nemo84]

I think it's not realistic to expect Vic to implement an entirely different logistics system at this point. The amount of work and testing needed for a demand-based system will likely far exceed what he's willing to invest in this issue, and will take away time from other new development such as air and naval combat.

A few minor tweaks to the current system, such as fixed-value traffic signs or whatever else we can come up with, are far more likely to actually be patched in.

[AttuWatcher]

A few minor tweaks to the current system, such as fixed-value traffic signs or whatever else we can come up with, are far more likely to actually be patched in.

Agreed. Is it currently possible to remove roads? I'd consider that a must-have feature to help smooth things out.

[Bleek]

Being able to demolish roads would fix a lot of the complaints.

I like the logistics system, but I would really like to remove obsolete roads instead of having a traffic light Christmas tree.

[IvanJoukov]

Logistic system is great, clear and easy. Some little things are missing (deleting road clearly) but I really don't want it to change, it's one of the best thing in this game.

[Kamelpov]

ORIGINAL: Cornuthaum

ORIGINAL: Kamelpov

If you don't want to think build supply truck and max size everytime and put them at 8 hex from each other and call it a day.

who has the population/workers for that? that's the major problem I have with logistics, even with rail IV stations between all my cities, there just aren't enough workers (and I've sucked up all free folk in all my zones already)

Well I'm optimizing road+rail with traffic so i never had that problem even my railroad can output 21000 on a medium map I can supply my whole empire with only 1 SHQ

[Kamelpov]

A demand based supply work well but require priority so what get the thing first zone or unit that is the question. One gonna starve while the other get their need. + The amount of calculation go up really fast with more combinaisons.

[misterblort]

Something that you put, just like a traffic light, but instead you put a priority marker that pulls supply to it first before doing the rest as normal?

[GodwinW]

I like the current system. So far. But only about 30 hours in.

[KingHalford]

This whole issue throws up a fundamental problem with complex strategy gaming, and that is that if you allow for people to micromanage things excessively, they will if they feel they need to in order to win. Good games designers will take this into account: the example I used in my review of the game was Vic's decision to punish changing BP allocation across Organisations, a genius move that still allows the player to crash fund something if necessary but at the cost of a serious penalty to loyalty across the board. This discourages excessive micromanagement, and it's something that's been used in 4X game design since the original Master of Orion.

This logistics issue is just is not a problem on the Regular difficulty setting. I've played this game well in excess of 100 hours now, maybe getting on 150 and most of those on Large maps and into the late game, and I've never felt the turn by turn Traffic Management was ever necessary, and certainly not after the first twenty or so turns.

Consider that after turn 100 or so, your turns are taking in excess of half an hour anyway especially when you've hit a Major Regime and are fighting a huge war. Whilst you certainly have to watch your logistics you never need to be constantly fiddling with the Traffic Management: if there's some odd mesh of roads then it's a matter of seconds to fix it and you'll likely never touch it again.

I understand that people who play on the hardest settings will be trying to leverage every advantage they have, but demanding that a core game mechanic is changed to facilitate that seems a bit much to me.

[Nemo84]

KingHalford, it's been explained to you several times already in this discussion that this is an actual problem for some people playing even on beginner difficulty. Can you please stop dismissing it just because you've played "over 100 hours" and never personally needed it? As if a mere 100 hours can show you every situation a game this complex can throw at you...

Let me give you a very clear example of one of my beginner difficulty games. I have a newly conquered city sitting at the edge of a 25 hex rail and covered road line from its nearest neighbour. The rail and road line is on a thin strip of land surrounded by oceans. There's a whole bunch of resource sites and bonus hexes on that strip, all branching off from that some road. Loads of truck stations and supply depots have been built or are under construction to support that strip, as are rail stations.

Near the newly conquered city I have several brigades engaged with a major power. Near that city is also a spaghetti of AI-built roads leading nowhere. The city is not yet fully pacified and construction of new transport assets in it is slow.

By using about 50 traffic signs to prevent all those side roads on the strip and near the city from eating unnecessary logistics that city receives over 3500 logistical points, my logistics lines extend up to 20 hexes from that city and I can keep the front intact and the city (barely) building its own logistics network at max speed (given the level of unrest). If I remove those traffic signs, my logistics in that city are reduced to below 2400 points, construction is bottlenecked by lack of resources and my logistic lines are reduced to 13 hexes, forcing me to abandon a good defensive position in the mountains for a battle on wide open terrain against a numerically superior opponent. All because suddenly every single lvl 1 resource site is drawing thousands of logistics instead of the 100 it needs, merely because it's located along the main logistical path.

And because my logistics capacity around that city is being increased each turn, I have to constantly fiddle with the traffic signs near it to keep all nearby assets operational without having them draw off so much capacity that it impacts the supply of my troops.