Another point in belts vs bots: inserter stack bonus

[mooklepticon]

So I finally broke down and built a bot-fed assembler. I needed a large enough amount of Big Electric Poles that an assembler was the right choice. However, my copper and steel lines weren't really conducive to this. My spaghetti was bad.

So, I made a bot fed assembler. That shit was ridiculous. In two swings of the inserter, it started running. I was confused, because it needs 10 raws (5 steel and 5 copper). For a belt, that will take 10 swings. That should take a while to fill up and start, but it was running. This was kind of a shock to me.

I didn't realize that going from chest to assembler got the stack bonus. I think this is a bit unbalanced. This makes bots way more powerful than I thought. I already thought they were a bit of a discontinuous jump from ground-belts to flying drones, but this makes it even more disparate.

I think that their behaviors should be more similar. Either 1) inserters should be able to pick up more than 1 item from belts or 2) inserters shouldn't be able to insert more than 1 item into a factory at a time. IMO, #1 is much more plausible and consistent. For example, the player can even grab more than 1 thing at a time. (Stand in a belt and press F. You'll pick up a bunch!)

Thoughts? Alternate solutions?

[ssilk]

Moved to balancing.

And you should not compare bots to belts, cause they have a completely different behavior.
See https://wiki.factorio.com/index.php?tit ... ch_case%3F

[_itg]

It doesn't really matter how fast the assembler gets started, in the long run. As long as you can get the components into the assembler in a shorter time than it takes to make the part, it will run efficiently. You may need to use multiple arms to do that with fast-building items, when pulling off a belt.

Moved to balancing.

And you should not compare bots to belts, cause they have a completely different behavior.
See https://wiki.factorio.com/index.php?tit ... ch_case%3F

They should be compared, because they do the same thing, i.e., moving things from one spot to another. What you linked is a comparison, incidentally.

[Koub]

How many bots would you need to convey continuously let's say ... 2400 items in one minute across 300 tiles ? Probably around 240. That's incidently what a single 300 tile blue belt can convey.
Over short distances, bots are the thing. Over base-ish distances, when high throughput is needed, bots are a very inefficient solution. You won't replace a 6 or 8 blue belt main bus with bots in a big base.

[bobucles]

Get rid of the logistic cargo upgrades and watch the tears roll in.

A base that worked off of 2000 bots suddenly needs 10000.

[mooklepticon]

Moved to balancing.

And you should not compare bots to belts, cause they have a completely different behavior.
See https://wiki.factorio.com/index.php?tit ... ch_case%3F

thanks. I didn't even know this forum existed O.o

[bobingabout]

Another thing to remember with the chest inserter stack size bonuses, is you can still benefit from this with Belts, you just need to use a chest as a buffer.

The easiest way to do this is with a chest for each ingredient, fed either by well-placed belts, or smart inserters, then you can have your inserters grab whole stacks at a time of each ingredient from each chest and feed everything into the assembling machine at once.

[mooklepticon]

Another thing to remember with the chest inserter stack size bonuses, is you can still benefit from this with Belts, you just need to use a chest as a buffer.

The easiest way to do this is with a chest for each ingredient, fed either by well-placed belts, or smart inserters, then you can have your inserters grab whole stacks at a time of each ingredient from each chest and feed everything into the assembling machine at once.

I've thought about this, but it feels kind of jury-rigged solution. It's not very elegant. Another example is to just use more inserters from belt to assembler, no chest. Adding the chest to that situation is nice for burst production, but the time savings are exceedingly small.

Realistically, you can feed a chest from 3 sides and draw from 1. Just have those 3 inserters feed directly from belt to assembler.

[bobingabout]

Another one I like to do is use a chest as a buffer on the output, I typically use these on furnaces.

[mooklepticon]

Another one I like to do is use a chest as a buffer on the output, I typically use these on furnaces.

I do this on multi-step assemblers. Ie, Speed 1 to Speed 2 to Speed 3

[Ghoulish]

How many bots would you need to convey continuously let's say ... 2400 items in one minute across 300 tiles ? Probably around 240. That's incidently what a single 300 tile blue belt can convey.
Over short distances, bots are the thing. Over base-ish distances, when high throughput is needed, bots are a very inefficient solution. You won't replace a 6 or 8 blue belt main bus with bots in a big base.

Inefficient perhaps. But you could technically still get higher throughput with bots than belt right even over distance? And you can replace belt with bots entirely should you wish, even on the main line, I've seen many pure bot bases, I think it was Xeteth I was watching stream a fortnight or so ago who was doing just that, pure bot base and going for 1 rocket per min. It was going to be a huge base, to give an idea he had 300GJ stored in accumulators, connected what he had built so far, and that 300GJ vanished in well under 1 min (though the end game plan was to use steam). A big, power hungry base indeed, and not fully built when I watched.

Now imagine the additional complexity needed if that base would be belt based.

And that's the issue for me with bots, and why i feel they should be rebalanced (perhaps by stack bonus for bots being readjusted). They give you the same or better throughput than belts with none of the complexity needed to design good belt based factories. They are simple to set up. Achieve the same or better than belt systems in a smaller footprint. Expanding your base is far simpler with bots than belt, as you don't have to work out how to route the belts. And when it comes to mega base scale are more efficient than belt from a UPS angle as there is no pathing needed, they just go straight there as the crow flies. Granted mega base builders on THAT scale are quite rare, but it's still applicable when considering old hardware running Factorio. Though the UPS point isn't exactly relevant here.

Bots give you many benefits yet little complexity in their implementation (not saying it's entirely a skilless tool) All the benefits but no real downside. And I feel that should not be the case, belt should always give you a higher throughput than bots because it takes more work, and requires better design to fully utilise. Bot setups do require large amounts of power, might be inefficient, but neither of these are an issue when you can just keep adding more steam or solar and more bots.

Edit: maybe I am wrong in the main line part with Xeteths base, thinking on iit a little more I'm not sure if that was the case.

[ssilk]

But you could technically still get higher throughput with bots than belt right even over distance?

That question is easy: No.

And the reason for that is, that you need to charge the bots and need to think about returning.

Let's explain in detail!
Think to an diagram. X-axis is "Distance", Y-Axis is "Time". The graph should represent, how much time it will take in AVERAGE to move one item.

So for the belts I can draw as a simple line, beginning from zero-point and the steepness is the speed of the belt.

For the bots this more complicated.

I can draw first 300 tiles (or so, I didn't count it) the same line as for the belts, but then they need to stop for charging. In the ideal case they need to stop directly above an roboport and don't need to wait. So there is a "step" in the line. But for the worst case, they need to fly back a bit and need to wait a long time. So there is a peak (distance is reduced, cause it needs to fly back) and then a long flat step (waiting time before the the roboport and longer charging time).
So - I sure you can agree - the true time, that is needed to transport one item by bots must be somewhere between this both extremes. The details (yes, you can optimize a lot) are not so important here. And yes, I don't included the stacksize.

Most players stop here and think: Oh, this is cool, even with charging this is way much faster, than belts.

But the truth is, that you need to think about, that the bot needs also to fly back to pick up the next items! In the best case the bot can pick up an item directly from his former destination. But in the worst case, there is just nothing around and he needs to travel the whole way back. That is just a big flat line, which adds time to travel time of items and no distance is taken.

That is just the whole secret of the bots. You need to calculate in their travel back and that gets extremley expensive, if you transport just one item-type into one direction. It is by far more expensive than laying a bulk of parallel belts (which is also much more reliable).

What follows out of that?
I mean most is already described in the linked wiki-article. But as an additional thought:
- It doesn't make sense to transport bulk items, as ores, plates, circuits etc. by bots over long distances (say: more than 50 tiles). Especially, if the item stream is constant (as it is, when you are thinking to miner). Bots can be far more effective, if you give them time for charging. Which items? Just look at the production statistics and multiply with the distances they need to be transported on the map and you know immediately what to make and what not. It doesn't mean, that no such items should be transported by bots! It depends on the above rules.
- It makes much sense to transport low used items by bots only. Good examples are capsules. Maybe the research potions. But also stone, iron plates and other bulk items, if there is a low usage of them. Think to complex assembling, where you need to put 4 or more items together but you need only a view output items. But as everything with bots, it depends on distance between production and usage, what works in one game can really be a big mess in the next, if you forget about the distances.
- Already mentioned, but again: If you keep in mind the rules for the bots you can of course transport items over long distance, if you are using pretty less. Good example is production of the biggest modules, supply of trains with coal, eventually supply of turrets with magazines...
- Also the other way around: If you have short distances and not a steady stream if items. A good usage for such bulk transport via bots is a train station, where you need the bots to transport from the output chests to a storage to belts that go to your production. Make sure to limit the robot-area and don't connect with other area!

[bobucles]

And the reason for that is, that you need to charge the bots

Your math is attacking the wrong angle. The problem is not with bots covering large or short distances. Bots can be abstracted away as the "conveyor belt" part of the air network. More bots isn't very different from building more belts to move goods around. The issue is with a bot's ability to access chests.

In short, bots have UNLIMITED SPEED when accessing chests. This is akin to having an infinite speed inserter at any arbitrary point in your logistics network.

There is no way in a million years that an inserter can load a chest in less than a second, but bots do just that. You can never make an inserter fast enough, and players would call it a broken machine anyway like with the idea for loaders. You have to address the problem with how bots interact with chests. For example, change chests to now use a docking port. The docking port is limited to one bot at a time, therefore bots can only access the chest a finite number of times per minute. So now you have chests loading and unloading at mortal speeds, as though the air game was just another type of inserter that happened to be airborne.

Ta daa.

[mooklepticon]

In short, bots have UNLIMITED SPEED when accessing chests. This is akin to having an infinite speed inserter at any arbitrary point in your logistics network.

There is no way in a million years that an inserter can load a chest in less than a second, but bots do just that.

Oh, wow. I never thought of that, but you're right. I routinely have 100's of bots access my inventory. If I had an inserter do that, the maximum would be 8. I don't often do that with chests, but it can be applied just the same.

[Koub]

With infinitely many robots, you're right, you can achieve higher throughput than any number of blue blets ... but at what price ?

[bobucles]

With infinitely many robots, you're right, you can achieve higher throughput than any number of blue blets

Infinity is always going to beat out a real number of anything.

There are some cases where moving items at a very high rate isn't that big a deal. Construction and player logistics are both very convenient things to perform quickly. There is no issue with having many bots perform many individual tasks to make factory life easier. Their value is to clean up the spaghetti base after all. However when you focus bots on any SINGLE task it becomes very obvious that they will always outperform anything on the ground. You can't unload a chest in seconds with ground tools, you just can't. But for some reason a blur of bots can snake its way to the chest and float away with the entire payload.

Sure it's expensive, pile of logistics drones aren't cheap. But an uncapped air conveyor will eventually beat any conveyor on the ground, simply because it's uncapped. All the ideas to make ground conveyors faster are misplaced because the issue is that air transport is both more convenient AND ultimately faster.

[ssilk]

Your math is attacking the wrong angle. The problem is not with bots covering large or short distances.
Bots can be abstracted away as the "conveyor belt" part of the air network. More bots isn't very different from building more belts to move goods around.

No, you just cannot build endles numbers of bots, cause you need to charge them. Roboports are big. You cannot build so compressed then, you need more space to build the production.
And you can calculate how much bots can be used per area, until the whole area needs to be plastered with roboports.

The issue is with a bot's ability to access chests.

The devs work currently around this: See the current Friday Facts Blog Post and a bit back.

In short, bots have UNLIMITED SPEED when accessing chests. This is akin to having an infinite speed inserter at any arbitrary point in your logistics network.

No, they don't have. They have waiting times before picking up and deploy. That times are very short, but they are there and I admit, the time is not per item.

There is no way in a million years that an inserter can load a chest in less than a second, but bots do just that.

Yes, bots can do that, but only so long, as they need to recharge. And it depends from many more factors, if they can do this. One is space. One is needed power to drive that. One is the costs of the bots vs. belts. And if I dig I find more.

You can never make an inserter fast enough, and players would call it a broken machine anyway like with the idea for loaders. You have to address the problem with how bots interact with chests. For example, change chests to now use a docking port. The docking port is limited to one bot at a time, therefore bots can only access the chest a finite number of times per minute. So now you have chests loading and unloading at mortal speeds, as though the air game was just another type of inserter that happened to be airborne.

I don't know, why this is useful. It looks quite cool, when they do this, I'm always quite fascinated about it. And I have no problem with the bots and belts. As I tried to explain above: They are self balancing!

What I mean with this: In direct comparison with me, if you're using bots only and I use a combination of bots and belts, you will surely loose. Not because I'm eventually the better player, but because using bots only is total crap.

Ta daa.

Boing.

[bobucles]

They have waiting times before picking up and deploy.

Is the waiting time by any chance equal to one tick of game time? That's the problem. You can not find any other factory piece in the game that performs 60 actions a second. It doesn't exist.

If a chest's waiting time was doubled to 2 ticks, you automatically cut maximum throughput in half. A value of 5-10 ticks would still be extremely powerful while 15-25 ticks is a closer match to inserter power rankings.

As I tried to explain above: They are self balancing!

Many aspects of bots work out okay. Access speed is a factor all its own and it's the main reason belts vs. bots has such a strange tipping point. If bots couldn't access chests an unlimited 3600 times a minute, there would be no choice but to use a healthy mix of both systems.

[ssilk]

They have waiting times before picking up and deploy.

Is the waiting time by any chance equal to one tick of game time? That's the problem. You can not find any other factory piece in the game that performs 60 actions a second. It doesn't exist.

I didn't measure it, but it looks so, that it is some ticks. When I slow down the game to 1/6 you can clearly see, that it is more than one tick. It looks like 5 ticks, but I'm not sure.
Perhaps the times are fixed into the code, or have other reasons?

Many aspects of bots work out okay. Access speed is a factor all its own and it's the main reason belts vs. bots has such a strange tipping point. If bots couldn't access chests an unlimited 3600 times a minute, there would be no choice but to use a healthy mix of both systems.

Ok, that would mean implementing a waiting queue for chests - as for roboports.
Might be useful, cause it will give a visual feedback to the users, that there is now too much traffic for this chest. It would also mean, that the maximum transfer per chest is 18,000 items per minute. Without the queue-stuff, that will add some more time (See how roboports).

Well it is eventually a way to fine-balance the robots vs. belts problem.

[Koub]

The thing is that even with 1 bot per stack of 5 items, past a certain distance, a belt will always be able to be faster that bots to convey the content of a chest, whatever the quantity. Bots are faster over short distances, but not long distances.