Finally... It only took a week... but I read all 44 pages...
Of course, by the time I finish writing this, I expect a few more posts.
First, I'd like to say that many of you need to read the fact in the FFF, that bots are not being removed. While some people understand that, a lot of posts are ignoring that fact. I find bots useful and it would be a shame for that to happen... I'm glad it won't happen.
There are a lot of posts from people who do not want to be forced into a single way of playing. Sadly, in late game, that is already happening. Any late game base is required to go bots or suffer a severe penalty in throughput. Using bots, you have a virtually infinite amount of throughput... If you need more throughput than you currently have, all you need to do is just add more bots. At a certain point you may need to take measures such as isolating roboport networks into minimal sizes supporting the production line, and adding roboports for charging. However, this is not that difficult to do, and even with the added space of extra roboports, any attempt to replicate the same amount of throughput using belts will take exponentially more space. There is no way for belts to compete with infinite throughput regardless of any reasonable buff.
I do believe a nerf to bots is required to balance them against belts.
However, any type of power cost will not be an effective nerf. Power generation in the early game needs to be monitored, but; by the time any player gets to the late game where the number of bots has an impact on power, they should have the means to expand power easily. In late game, if your power is insufficient, you are doing something wrong. Power is unlimited... Nuclear combined with the Kovarex process should provide enough, otherwise increasing the size of solar arrays is effectively free power, and even steam power is infinite if you feed it with solid fuel... oil wells never completely empty and coal is usually available to the point of skipping many patches. Even if coal isn't that abundant. with electric furnaces, the only other need for coal is grenades/military science packs.
Similarly, changing the resource costs would not be an effective nerf either, and is potentially harmful earlier on. In late game, modules have a significant cost... but all costs are meaningless. Everything has a return on investment... the more it costs the longer it takes to recoup. And lets face it, as your base expands in both size and production capacity, your ability to get resources from far away improves as well. You only need resources for science, power, and expanding your base, of which, the cost of more bots easily is covered in the costs of expansion. More expensive robots effectively cripples them in early game, but in late game, it just shifts the amount of resources building them to factories creating the raw materials needed for more bots. (e.g. more resources would go to bots leaving slightly less resources going to science.)
The idea of not allowing bots to be in the same spot is a possibility... However, I think that the increase needed to the pathfinder logic would be a significant performance impact. While that is not reason alone to avoid this, I believe there is a much easier way to accomplish the same thing.
IMO, the best nerf would hit throughput directly.
The idea of limiting one bot to interact with one chest at a time would effectively limit the throughput. I think the time necessary for each interaction should be equivalent to the swinging speed of a fast inserter. With a maximum carrying capacity of 4, this would mean that bots can effectively feed chests that are connected to a fast inserter with a maximum stack size of 3. However, they would come up short of supplying a chest connected to a stack inserter. The way around this would require additional chests and inserters to compensate. And remember, there would need to be more provider chests necessary for bots in the pickup... essentially one provider chest for every requester chest that you need full throughput. The other benefit of this approach is that it does not harm the early game, it will not impact most assemblers until after beacons greatly increase the speeds.
Even with a bot nerf, I believe that a belt buff would still be necessary. Right now, blue belts move 40 items/second. Blue circuits require 20 green chips to create and 10 seconds. Essentially 20 assemblers can be supplied assuming a crafting speed of 1. (admittedly, without modules, assembler 2s have a crafting speed of 0.75 and assembler 3s have a speed of 1.25.) However, a speed of nearly 4 can be easily achieved with speed modules in beacons and productivity modules in the assemblers. This means a fully saturated blue belt can only supply enough green circuits to 5 assemblers. This approach still requires belt weaving (and dealing with pipes as well) due to the number of ingredients required in blue circuits. This is greatly underpowered compared to the amount of throughput provided by bots. Even with the suggested nerf to bots, a max of 5 assemblers and required belt weaving is a significant disadvantage to logistic bots.
One idea is to create a Tier 4 belt with 4x the speed of T1(yellow) which would be about 52 items per second. (I think green was suggested as a color.) I do like this idea. and it could be implemented with space science packs which would delay it until after the first rocket launch. The delay would effectively stop it from impacting the early game. This is a start, but essentially it is not enough... in the blue chip example above, you would only get one more assembler with a little left over.
Another idea is a stacked belt. There seems to be 2 different interpretations of this "stacked" belt in this thread. One is a normal belt that can hold a full stack of items instead of individual items. The other interpretation seems to suggest a second (or even a third) layer of belt immediately above the base belt.
On the idea of a belt holding full stacks of items, this seems far overpowered. With green chips, you are talking about effectively increasing the throughput of a belt 200 times. IMO, that is insane. We do not want to replace overpowered bots with overpowered belts. Even the 50x increase of ores seems like way too much. While some products do have even lower stack sizes, many intermediate products are in the 50-200 range (rocket parts being an exception.) I am personally not fond of this idea at all.
I do like the other interpretation of a stacked belt. The way I see it, you make "stackers" instead of splitters... that take two lines of input and place one over the other. I would not use any merge effect other than essentially moving all contents of one belt to the top and the other belt contents to the bottom. This can effectively double the amount of item per second (to 80/s on a blue belt) or allow for transporting 4 different items at once on the same belt... All without needing belt weaving. In the blue circuit example above, it effectively increases from 5 to 10 assembler units supplied, (combining this with Tier 4 "green" belts would increase it to 13 assemblers supplied.) This is a nice little buff without becoming severely overpowered.
The 4 separate lanes in a vertically stacked belt would be helpful as well. As it has been mentioned, you can only have one belt on either side of an assembler in a standard setup with beacons. This means more ingredients can be fed to assemblers without belt weaving.
The vertically stacked belt idea still needs a little flushing out on a few details... such as do splitters work with stacked belts? or possibly only the top or bottom? Can you go underground with stacked belts? One thing that I have considered is that maybe it takes two squares to go underground and two squares to return... this can limit the use of stacked belts with weaving. How do inserters interact with stacked belts? Personally I do not see a change needed for this with the assumption that if they can mechanically move horizontally to pick up from either side, they can probably move vertically to get from the top or bottom belt as necessary.
Additionally, a reversed stacker to separate top from bottom will be needed, and maybe a special piece, a "merger" to combine top and bottom together into a non-stacked belt.
Another possible buff might be to allow for fast long handed inserters, maybe even stack long handed inserters. This could be helpful to throughput if you ever need to have 2 belts on the same side of an assembler.
Of course, now that I finished with my 2 cents, in only a few hours, a new FFF will be started. But, I do believe that this topic will not die.