Friday Facts #161 - Infinite Research and Blueprint Library

[MrGrim]

When it comes to resource sinks, is the rocket not a large one? I've not yet managed to construct one, so I don't know exactly how demanding it is, or do people just want to give their labs something to do as well?

On the topic of potential research, perhaps add a boost to radar range, it would both be useful to the player and not have any kind of OP effects in-game. Maybe even make it an early to mid-game research (green science?) and allow the player to continue boosting it as the game progresses. Imo research that improves production or energy output should not be made infinitely upgradeable, as this could get quite rediculous given enough time.

Anyways, that's my two cents on the situation, I trust the devs will come up with something exciting and fitting in due time.

It is, indeed. I just like the idea of multiple complex chains going achieving different ends. IMO a factory producing modules, firing rockets, and performing research all at once is more interesting than a factory only launching rockets.

Also, agreed on the need for radar upgrades.

[kovarex]

When it comes to resource sinks, is the rocket not a large one? I've not yet managed to construct one, so I don't know exactly how demanding it is, or do people just want to give their labs something to do as well?

The rocket is demanding, true. It's a good sink for a number of resources.

What the rocket needs, is a circuit-connection that can allow auto-launch when the satallite is installed

Dev's: is there a plan to include circuit-network-triggered-launch?

Certainly!

[Mehve]

I'm not sure I'd necessarily agree with more powerful radar. Having used the Big Brother mod, I've discovered that revealing too much territory can play havoc with the biter expansion behaviour.

What could stand to changed is the random nature of the scanning. Right now, on an already-scanned area, it's entirely possible for a biter nest to expand towards you several times, and a radar never pick it up until it's too late, thanks to the randomly chosen scanned square each cycle. I'm honestly mystified about the reasoning behind this implementation.

[Nemoricus]

I'm not sure I'd necessarily agree with more powerful radar. Having used the Big Brother mod, I've discovered that revealing too much territory can play havoc with the biter expansion behaviour.

What could stand to changed is the random nature of the scanning. Right now, on an already-scanned area, it's entirely possible for a biter nest to expand towards you several times, and a radar never pick it up until it's too late, thanks to the randomly chosen scanned square each cycle. I'm honestly mystified about the reasoning behind this implementation.

Wait, what? I thought that radars look at the chunk that hadn't been 'looked at' the longest.

[Rseding91]

I'm not sure I'd necessarily agree with more powerful radar. Having used the Big Brother mod, I've discovered that revealing too much territory can play havoc with the biter expansion behaviour.

What could stand to changed is the random nature of the scanning. Right now, on an already-scanned area, it's entirely possible for a biter nest to expand towards you several times, and a radar never pick it up until it's too late, thanks to the randomly chosen scanned square each cycle. I'm honestly mystified about the reasoning behind this implementation.

Wait, what? I thought that radars look at the chunk that hadn't been 'looked at' the longest.

They do.

[GuiltyBystander]

How do you plan to do infinite research for normal/stack inserter bonus? This one is slightly different because every level of tech isn't the same.

Bump. I'm still curious about the answer.

[Mehve]

I'm not sure I'd necessarily agree with more powerful radar. Having used the Big Brother mod, I've discovered that revealing too much territory can play havoc with the biter expansion behaviour.

What could stand to changed is the random nature of the scanning. Right now, on an already-scanned area, it's entirely possible for a biter nest to expand towards you several times, and a radar never pick it up until it's too late, thanks to the randomly chosen scanned square each cycle. I'm honestly mystified about the reasoning behind this implementation.

Wait, what? I thought that radars look at the chunk that hadn't been 'looked at' the longest.

They do.

But it's a long time before the radar finishes cycling through all its options, and you can get a lot of ninja-biter advancement in the meantime. I'd be happy to sacrifice a lot of radar range, if it regularly updated that limited range. Right now, the solution is simply to dump tons of radars down together, and hope that at least one of them scans a critical square so you get some advance notice of a new biter nest.

[Rseding91]

But it's a long time before the radar finishes cycling through all its options, and you can get a lot of ninja-biter advancement in the meantime. I'd be happy to sacrifice a lot of radar range, if it regularly updated that limited range. Right now, the solution is simply to dump tons of radars down together, and hope that at least one of them scans a critical square so you get some advance notice of a new biter nest.

There's no reason to keep the map updated that frequently. That would waste a ton of CPU time for virtually zero benefit.

If the biters are migrating around - what does that change? You've got turrets defending things you want to keep right? If so, they can't kill it so them migrating again changes nothing.

[Hannu]

When it comes to resource sinks, is the rocket not a large one? I've not yet managed to construct one, so I don't know exactly how demanding it is, or do people just want to give their labs something to do as well?

Rockets are just as large resource sink as you want because you can have many silos launch them continuously. But it is more interesting if there are more complex logistics and material flows than just one. I think that infinite research would be a good idea to get more variety to end game (which begins when I have ability to produce and launch rockets automatically at some rate, IMO)

There are two types of players. Others want to get some predefined achievements and begin a new game after it. Others want to make huge systems and see when they work steadily. Maybe there could be setting for infinite research or some levels could give achievement "research complete" so that both groups can have fun.

[Koub]

How do you plan to do infinite research for normal/stack inserter bonus? This one is slightly different because every level of tech isn't the same.

Bump. I'm still curious about the answer.

Starting from current stack inserter bonus, one could do for infinite stack inserter research level x :
- if x modulo 2 = 0 then stack inserter bonus +2
- if x modulo 2 = 1 then stack inserter bonus +2
- if x modulo 2 = 2 then stack inserter bonus +1, regular inserter bonus +1

of just leave regular inserters as is, and provide only +1 for stack inserters every additional "infinite" level.

[Mooncat]

How do you plan to do infinite research for normal/stack inserter bonus? This one is slightly different because every level of tech isn't the same.

Bump. I'm still curious about the answer.

Starting from current stack inserter bonus, one could do for infinite stack inserter research level x :
- if x modulo 2 = 0 then stack inserter bonus +2
- if x modulo 2 = 1 then stack inserter bonus +2
- if x modulo 2 = 2 then stack inserter bonus +1, regular inserter bonus +1

of just leave regular inserters as is, and provide only +1 for stack inserters every additional "infinite" level.

Or just keep such technology to have finite levels?
Infinite research is applicable to technologies that can have non-linear growth on the values. When you keep researching on them, the growth will eventually become very very small, like +0.0000001. But the minimum value for stack count is 1, so it can only be linear growth.
And it will be hilarious to see an inserter keep getting items to its head infinitely.

[IronCartographer]

How do you plan to do infinite research for normal/stack inserter bonus? This one is slightly different because every level of tech isn't the same.

Bump. I'm still curious about the answer.

Starting from current stack inserter bonus, one could do for infinite stack inserter research level x :
- if x modulo 2 = 0 then stack inserter bonus +2
- if x modulo 2 = 1 then stack inserter bonus +2
- if x modulo 2 = 2 then stack inserter bonus +1, regular inserter bonus +1

of just leave regular inserters as is, and provide only +1 for stack inserters every additional "infinite" level.

Or just keep such technology to have finite levels?
Infinite research is applicable to technologies that can have non-linear growth on the values. When you keep researching on them, the growth will eventually become very very small, like +0.0000001. But the minimum value for stack count is 1, so it can only be linear growth.
And it will be hilarious to see an inserter keep getting items to its head infinitely.

While it is possible to rely on non-linear (decaying) growth values, using exponential costs will eventually have the same bounding effect. Stack bonuses are a fine candidate for a (hugely exponential) infinite research tech.

[Unoxxis]

If stack bonus becomes an infinite research, I would however strongly advocate for a setting in the entity menu to set the actually used stack bonus per inserter. Make it default to "MAX" (current behavior), but players who wish to change the value may do so (added bonus: per circuit condition, but manually would be a good start). This would stop a ton of setups from breaking if you happen to arrive at an odd (or worse prime) number for the inserter stack bonus...

[RobertTerwilliger]

+1 for removing "classic" blueprints from the game. They became obsolete after blueprint book have came up, and now when library is on it's way to us, it is possible to get rid or blueprints as items completely. Honestly, that 1 slot that book occupies already have broken the balancing of "to have blueprint, or not to have?", so just free that last slot.

Another idea: how about allowing us to create and change blueprints in special GUI, not just building things for the only purpose "scan and delete".
Currently if your factory gets big - you have to request items to set up a blueprint, and after setting it up you trash away all the items bots brought to you. Waste, isn't it?
This can (and I believe should) allow creating blueprints from items you don't currently have (say, creative window with side menu with all placeable entities, wires and modules).

[mrvn]

Here are some more ideas for infinite research (and maybe for some basic research too):

- robot speed
- robot carrying capacity
- robot enery capacity (travel distance) -- Why is there no tech for this yet?
- roboport logistics/construction range
- roboport recharge capacity / rate
- train/car/tank speed
- train/car/tank acceleration (I want my train to stop on a dime and give me 9 cents change
- exoskeleton MK<x> (walking speed)

[Nemoricus]

On infinite research...I'm generally against the idea. It's true that it's a resource sink, and the game does need more of those, but anything like bullet damage and such has effects on combat balance, potentially making it entirely trivial after a while.

What I'd prefer to see are more scoring items like rockets. Perhaps habitats and such? Preparing the world for colonization is ostensibly a goal of the game, and it would be nice to see a nod in that direction.

[bobingabout]

anything like bullet damage and such has effects on combat balance, potentially making it entirely trivial after a while.

Agreed.

It's okay for some things, like, Robot follower count, you still have to build and launch them. A few others would be reasonable too, but weapon damage should probably be one of those with a cap on it.

[mdqp]

I would love to have extended marginal capacity for logistic/construction robot speed, and logistics capacity. Additionally, I would love increased logistics slots, and even increased armour-bonus slots for inventory.

More effective shields, faster walkers, bigger equipment grids (MKIII armour?) Personal robo-ports with more robot slots, bigger area, more efficient recharge.

The limits I run in to in large factories are:
- not enough logistics slots to maintian an inventory
- not enough inventory slots to hold all the things I want/need
- logistics robot system is too slow over a large area - faster robots with more capacity
- personal roboport range is not enough for large blueprints
- follower robot count at 134 is "not enough"
- bullet damage/speed is not enough for turret defense against behemoth biters
- turret range is too short (laser-turret is better, but I would like to research an increased gun-turret range).
- cluster-grenade shooting-speed ?

All the above could be given "infinite" research, and I would happily sink a large factory's output for those.

You say: "perhaps ones that can be used for a more tangible benefit" - I consider things like the above to be very tangible.

I think we won't see eye to eye on this, but again, it's not like people can agree on everything. I don't believe it's meant to really expand your factory's capacity in a significant way beyond the current limits, without investing hundreds of hours, and if that's the case, I believe you can achieve similar results even without those incremental improvements (what I mean is that, almost inevitably, adding a turret or improving the design of the outer walls will always be much better than investing in damage research after the short period). Effectively, you'll be running on the hamster wheel for days, doing nothing but menial work, at which point, I'd prefer to have, I don't know, a "mega research project" to lift the limit on the robot count, or doing improvements on the turrets themselves (maybe modules, whatever), rather than the research, so you have the added challenge of wanting to keep the upgraded turrets to survive attacks, or just have other weapons like nets which spring when enemies pass on them slowing them down, and other automated weapons with effects that synergize together (kind of like in tower defense games).

To sum it up, I think there are more interesting ways to tackle the issues some people experience with the late game, which are more definitive. I honestly don't see the appeal of spending hundreds or thousands of hours to increase my capacity doing repetitive tasks, I'd rather get a rather hard one time "remove limits" for robots and other hard caps, or perhaps inventive solutions like the possibility to teleport items to and from the inventory and storage, but which requires "teleporting stations to be present between the player and said storage, and other such constructions. The infinite research is a simple solution, but not one that appeals to me.

Again, it's not like I am saying "kill it with fire", so long as some people are happy with it, it doesn't really detract from my enjoyment, I just won't spend much time on the infinite research myself.

[Anson]

- if x modulo 2 = 2 then stack inserter bonus +1, regular inserter bonus +1

small detail: x mod 2 will never be 2 :-)

Or just keep such technology to have finite levels?
Infinite research is applicable to technologies that can have non-linear growth on the values. When you keep researching on them, the growth will eventually become very very small, like +0.0000001. But the minimum value for stack count is 1, so it can only be linear growth.
And it will be hilarious to see an inserter keep getting items to its head infinitely. :lol:

While it is possible to rely on non-linear (decaying) growth values, using exponential costs will eventually have the same bounding effect. Stack bonuses are a fine candidate for a (hugely exponential) infinite research tech.

that's exactly what i was thinking immediately :
Wheat and chessboard problem

Moral story
The problem warns of the dangers of treating large but finite resources as infinite, i.e., of ignoring distant but absolute and inevitable constraints. As Carl Sagan wrote when referring to the fable, "Exponentials can't go on forever, because they will gobble up everything." Similarly, The Limits to Growth uses the story to present the unintended consequences of exponential growth: "Exponential growth never can go on very long in a finite space with finite resources."

even when you add +1 bonus to stack inserters for each additional level, and when using only 2^level for the exponential function (the formula given by the devs above, and not really "hugely exponential"), you would end up in using insane amounts of resources and time to get minimal improvements. rough estimation on the magnitude of required research:

• for bonus 10 : 1000 research - 1000 blue science packs = 20k iron ore, 16k copper ore, 1k coal, 6k gas, 7k water
  let's assume 1 minute only to do that: roughly 150 assemblers, 620 smelters and almost 500 mining drills would be needed
• for bonus 20 : 1000000 research - 1000000 blue science packs = 20M iron ore, 16M copper ore, 1M coal, 6M gas, 7M water, this would already take 1000 minutes (16 hours), or with a MUCH bigger factory still require only 1 minute, but 150000 assemblers, 620000 smelters and 500000 mining drills (how many trains and belts to transport these amounts in one minute? and how much time to set up so many mining fields? :-)
• for bonus 30 : 1000000000 research - 1000000000 blue science packs = 20G iron ore, 16G copper ore, 1G coal, 6G gas, 7G water, this would either take 700 days, or 1 minute with 1200 million (1.2 billion) machines (assemblers, smelters, miners)
• for the Nth +1 bonus, it would require all the resources and time that the first N-1 levels have required
• for bonus 60 : a factory with 12 billion tiles (12 million chunks; ignoring additional space for trains, belts and power generation) would be 110*110km in size, its savefile might be around 12M*5k=60GB, and this megafactory still would need almost 2 years to get bonus level 60 (if it can keep up max FPS/UPS)
• for bonus 70 : a factory with the size of the earth (500000 km^2), savefile 210 Terabyte, would need 50 years
• for bonus 98 : this earthlike factory would need 13 billion years (the age of the universe)
• to get to bonus 100, a better computer than earth would be needed, if it isn't destroyed by vogons before.

i think that there will be no danger to get overpowered inserters with infinite bonus, even if the research is "unlimited" (no more hardcap) ... :-) LOL

[rolfl]

.....

[*]to get to bonus 100, a better computer than earth would be needed, if it isn't destroyed by vogons before.[/list]
i think that there will be no danger to get overpowered inserters with infinite bonus, even if the research is "unlimited" (no more hardcap) ... LOL

meh, I will just quickly research logistics-robot-speed-bonus 30 and logicistics capacity 30, then I won't need all those pesky rails to support the research for inserter-bonus

On the plus-side, at least with my factory that size there will not be enough space left for any biters in the 2Million x 2million factorio-map. I will need those +60 exoskeleton walkers!