I miss building multiple factories

quote: Plopping a 800m factory, then building two 200m copies, then reclaiming the original for 400m will double your buildpower for no net metal cost.

Similarly, you'll always want to build a second factory for 200m ASAP as cheap insurance against losing your factory, since rebuilding your only factory from nothing costs 800m vs. rebuilding either one of your two factories costs only 200m.

I personally don't think the potential benefits of any change along these lines are worth the additional arbitrary complication/inconsistency. (As a corollary making the game more arbitrarily complex is likely to make the game harder to learn, thereby losing any benefits to the new player experience this offers.)

quote: I don't like this plan very much, but i don't hate it as much as i would expect. Identically looking things are going to have different properties: a 800m factory is going to look exactly the same as the 200m factory These things will behave differently: full factory will leave a wreckage, while a discount factory will leave a partial wreckage (at least that one is going to have a reclaim healthbar) The 200m factory is going to have full health of a 800m factory, which will make it repair four times faster/cheaper. The 200m factory is going to have the full health of a factory, as opposed to a Caretaker. This method of boosting production is going to be much less raidable. Plopping a 800m factory, then building two 200m copies, then reclaiming the original for 400m will double your buildpower for no net metal cost. Adding some kind of an indicator that a factory is a discount will address the cluster of the first three "looks identical" issues, i guess.

I agree.

quote: I'll note, regarding your motivation, that even with this change new players making multiple factories is still a mistake because they typically vastly overbuild production capacity relative to their incomes. The fact that their overproduction takes the shape of factories rather than caretakers (as it "should") is almost incidental. New players don't understand caretakers so they make multiple factories. New players ALSO don't understand the economy so they make multiple factories.

I disagree. I see new players notice that they could be producing more units and decide to increase their production capacity. The big mistake that they make is spending 620 more metal than they need to by making a factory. The cases where they overbuild BP are not even that bad as excessing BP is usually fine.

quote: I'll also note that while I like this system, and it makes sense given the goals we're trying to realize, it's lacking "justification". The player can quickly learn that additional factories are cheaper than first factories, and that they're cheaper when they're closer to the existing factories. But then they might ask "Okay, but why?" The best answer we could provide is "... because it makes for better gameplay." Whereas the Master/Slave factory concept (which I don't particularly like) provides some kind of justification for the difference. Rather than factories mysteriously being cheaper when they're close to each other, instead there's expensive factories and cheap factories, but you have to have an expensive factory before you can make the cheaper factories, and the cheaper factories have to be close to the expensive factories. This makes a certain amount of sense. Players can easily reason about it.

I agree.

quote: I like the simplicity of the proximity discount system, but I feel like it needs something more than tooltips to make it not just discoverable but also understandable. Perhaps when placing the nanoframe, if an allied identical factory already exists, not only do you get the tooltip displaying the discount, but the name of the building changes as well. "Master/Slave" probably isn't appropriate, since after being built the factory is no longer dependent on the previous factory. Maybe the second and subsequent factories could be labeled "Duplicate"? Or "Auxilliary"? Or "Supplemental"? Or "Remote"?

I disagree. The distance discount exists in part to make the discount discoverable. There is a very tight feedback loop between moving your mouse around, noticing a line change length, and noticing a big number go up and down. You haven't said how you are teaching your auxiliary factory. Changing a name in a tooltip is fairly invisible. I think you need to put a lot more work in to make this good.

quote: Similarly, you'll always want to build a second factory for 200m ASAP as cheap insurance against losing your factory, since rebuilding your only factory from nothing costs 800m vs. rebuilding either one of your two factories costs only 200m.

I agree.

quote: My suggestion is this: the metal spent on unlocking tech (600m) or on building in a new location (up to 400m) is lost as soon as it's spent. So a completed factory, no matter how much was spent on it, is only worth 200m (the value of its buildpower) for things like reclaim and repair and so forth. That addresses concerns like "The 200m factory is going to have full health of a 800m factory, which will make it repair four times faster/cheaper" and "Plopping a 800m factory, then building two 200m copies, then reclaiming the original for 400m will double your buildpower for no net metal cost".

Essentially constructors spend 600 metal on the nanoframe researching the ability to build the factory, then build the factory. This adds a global tech system as I don't see it working with proximity. Also, metal spent in parallel on two factory nanoframes should go to a shared research progress pool. This is relatively simple to do (it is like the prebuild cost on terraform) but takes some work to display. A healthbars that says "researching" feels inadequate.

quote: Another suggestion: as players on an allyteam plop their initial factories, go ahead and apply the discounts for duplication and proximity, and any amount discounted from the nominal 800m factory cost is refunded and split across the players in the allyteam. This way, if two players plop the same fac next to each other they haven't wasted a potential 600m. Which means players won't feel pressured to choose a fac they don't want to play just to not waste the potential to unlock new tech for the team.

I disagree. Players already waste their tech unlock by duplicating plops in teamgames and the situation seems fine as is. There is currently very little drama around factory choice. With your refund system I guarantee that people would plop duplicate factories just to get the metal to rush a Strider. In general people would always feel pressured to shift towards plopping a monoculture of factories just for the extra metal. Your system basically makes "what is powerful" equate to "what is boring to play". Diversity is interesting, which is why it is currently encouraged.

quote: I personally don't think the potential benefits of any change along these lines are worth the additional arbitrary complication/inconsistency. (As a corollary making the game more arbitrarily complex is likely to make the game harder to learn, thereby losing any benefits to the new player experience this offers.)

I have concerns along these lines, but I also believe that in a properly executed design it is possible for an increase in complexity to better guide players towards understanding the system and using it correctly.

SmokeDragon new thread.

These ideas are solutions to "making duplicate factories close together is always terrible" but they still both amount to design cruft when solving this. I don't see duplicate factories having many mechanical advantages over other forms of buildpower. Someone who is paying attention to the design of Zero-K while learning it may become confused as to why a location-limited cheap factories exist.

Imagine that we design some sort of cheap factory system that somehow prohibits cheap proxy factories. If these factories cost the same as a caretaker, what is their advantage?

Cheap nearby factories:

• Healthier than Caretakers.
• Can be used to spam Flea while making Crab (or similar).
• Can be used to produce units on multiple sides of blocking terrain (you could plop up a cliff and put a factory at the bottom after you walk there).

Caretakers:

• Can help constructors produce economy, defenses, factories, more Caretakers etc... in the base.
• Can help produce units of other techs given the nearby factory.
• Improve factory production in series rather than parallel. This results in less metal sitting wasted in unbuilt nanoframes.
• Take up less space.
• Can be buried or put behind walls.
• Can terraform.
• Can repair units.

As for constructors, cheap factories have better BP than mobile constructors but are still worse in all the ways that cheap factories are worse than Caretakers.

With all these downsides how much would a cheap factory have to cost? 120 metal? Lower? The question is then whether we want expanded specific unit construction to be that cheap. There are some mechanical advantages to economically encouraging cheap factories over caretakers:

• Parallel construction buffs cheap units (I like buffing cheap units).
• The option to invest in cheap but highly specific BP is an additional strategic choice.

Currently there is little choice in how to ramp up factory production: just make Caretakers with the assurance that they can switch to whatever tech you make in the future. It could be good to reward people who make deeper plans about unit production with a bit of extra efficiency. Of course, both of these 'advantages make the game harder overall. Is higher skill cap an advantage? Is this the kind of economic thought and planning that people who say ZK lacks in that area want?

Here is a radical third option, just teach the game through widgets:

• Factories cost 800 metal. In fact, the entire game is the same mechanically.
• When placing a factory, the duplicates of that factory than you own have a circle of radius 450 drawn around them. A line is drawn between the cursor and closest duplicate.
• When placing a factory, if the cursor is within 450 elmos of a duplicate then the active command is set to "build Caretaker". Moving the cursor away from a duplicate turns the command back into "construct factory".
• Note that Caretakers have 500 build range.
• Profit.

Edit: Also... unlock Caretaker in mission 0.

I like the toggleable assist widget idea GoogleFrog. We can make and use as many as we want and it we don't have to worry about doing something weird with balance.

I made the tech: https://github.com/ZeroK-RTS/Zero-K/commit/a8e6953c511139cc2605286e51fca1dfc417397f

quote: Global research is the simplest, most straight-forward way to implement the things that I think are being asked for in this thread: viable multi-fac production while keeping factory choice == faction choice. It makes proxy facs the same cost as closeby facs - cheap if already unlocked, expensive if not yet unlocked; I don't know whether that matters.

Global research is my least favorite proposal because it adds and entire new domain to the game.

quote: Note that your UI won't allow for someone who knows what they're doing to deliberately place a second master factory within range of a first; maybe there needs to be a way to override the placement so you can say "no, I actually do want another expensive factory here, not a cheap one".

If you really want to you can drag a row of factories. I don't know many cases where you want to do this. Also, the widget could be put in the menu and toggling it could be bindable to a hotkey.

quote: I think your radical third option to teach the game through widgets will only serve to frustrate new players. "Why won't this stupid thing let me build another factory?!? @#%$@!!@@#$!!" Watching new players I've seen a number of things that are mysterious and frustrating to them; let's not add another one.

I don't think so. Players are not set on making a new factory, they want to make more units and see making a new factory as the easiest way to do this. The UI is essentially saying that when you put factories close enough together you instead produce a modular extension of the factory (a Caretaker). This is the same UI that I would implement for the more involved modular factory pad system. The only difference is the Caretaker model and tooltip (which I also renamed from "Static Constructor" to "Construction Assistant").

You could be right, but you haven't convinced me that it is not worth trying.

Maybe because it's something lots of people have been thinking about but not annoyed enough to mention? Multiple factories is a staple of the entire RTS genre, so maybe it's always bothered people a little bit, dunno.



Anyways, I still think a lot of the discussion is unnecessarily complex. Having variable costs seems to produce all sorts of complexities and exploits when it comes to reclaim, HP etc.


I still think the Primary / Secondary mechanic is the simplest to understand. Or maybe a tech structure mechanic, which might be familiar to other RTS players:


- Make a single new model & structure. It's called "Tech Centre" or "Tech Core" something like that. A small boxy datacentre looking sort of structure. It has some cables on one side, to visually attach to a factory.

- Factories cost 200. Centres cost 600. Centres have low HP.

- Your initial free factory automatically has a Centre attached.

- Otherwise, a Centre must be built attached to a factory. You can use the code for mexes for this - mexes autosnap to the nearest mex point, and do not let you construct if there are no mex points? (This feature should pop up an error message when this happens - "Must be built on metal." "Must be built next to factory." etc)

- After your first free factory, you can build factories of any kind independently from tech centres. To do a tech switch, you build a factory of the new type first, then a tech centre attached to it.

- A factory with a Centre attached, unlocks the tech of that factory for your team. If there are no Centres, all the build options for factories of that tech are greyed out, clicking on them will result in a message, "Requires Tech Centre."

- (If you really must have the distance mechanic, then the influence of the Centre has a radius, beyond which factories do not benefit from the tech anymore, and need a new factory+centre built nearby. But I think we should just start with global tech and see if there's any problems that need this distance mechanic.)

oh and btw tech-trees are also a staple of the genre.

AdmiralZeech that idea seems like it has the same effect as Global Research but with some differences in the feel. I'll summarise it.

Tech Center:

• Factories cost 200
• There is a structure called a Tech Center that costs 600.
• If a factory on your allyTeam has a Tech Center associated with it then all factories of that type are active. Otherwise, all factories of that type are disabled.
• A Tech Center can only be associated with one factory. This is achieved by making the Tech Center confer activation only for factories at a particular location adjacent to it. The UI may help people find this point.

As far as I'm concerned this replaces Global Research. The Tech Center is not a hidden global variable and it can be destroyed. It has some unsolved problems:

• When plop automatically places a Tech Center you have to ensure that players place the factory in such a way that the Tech Center is not blocked.
• More generally, when you place your first factory of a type there needs to be a way to show that the Tech Center is not blocked.
• Duplicate Tech Center plops are very useless under this system. The duplicate plopper would prefer to place the factory and Tech Center independently to allow it to be used on a new tech later.
• I don't yet know how players are meant to learn that a Tech Center is required to produce units. Without some UI, perhaps auto-queuing Tech Centers for new factories, I can see people getting confused by factories that don't produce units.
• It will be difficult to make a model that fits and seems to feed into the various factory models.

I think there needs to be a blueprint that is Factory + Tech Center and that has an appropriately large footprint. This solves the unclear UI and some issues with factory plop. Alternately we could figure out how to draw two blueprints at once. This involves reimplementing engine behaviour as it currently draws the blueprints and checks for valid placement.