Hacker News new | past | comments | ask | show | jobs | submit login

The fastest "filter inserter" in modern Factorio meta is the splitter, which operates at full speed of the belt.

There's almost no reason to use inserters for belt-to-belt transfers for modern meta, aside from a few speed-running stats where you use red-inserters while skipping logistics2 or something.

But if you're trying to sort items on a belt by placing them onto another belt, the answer is a splitter. Item A splits off to another belt, while all other items loop back.




> Item A splits off to another belt, while all other items loop back.

This doesn't address the parent comment's concern:

> IIRC splitters only accept a single filter, and so you'd need many of them at each junction

I haven't played Factorio in years, but IIRC the splitter maintains state (direction for next item) per item type, so I guess it can be set up to filter as many types as you like? I remember you had to prime it for the desired item type, but I forgot the specifics of how it does the rest of the filtering "logic".


> I haven't played Factorio in years

Oh geez, your comment reminds me of like 8 years ago. You've really been out of the loop haven't ya? Yeah, what you say used to be true, but that's not what I'm talking about.

All splitters today can split items off. You can just click on a modern splitter and say "Left side Iron ore", and all iron-ore leaves the left side of the splitter, and all other items go out the right side. This operates at full speed, no glitches.

> so I guess it can be set up to filter as many types as you like?

So use a splitter per item, and then merge the belts back together later.

If you have 5 items to sort, create 5 filters, and then run the belts to route them where those 5 items need to go.

For "Meta" builds, the key requirement isn't size. Its throughput. When you have 5 filters inserters, you barely will have ~10 items/second throughput (and that will glitch out depending on how successful your inserters are at picking up items, corners can pose issues for example)

When you have 5 filter-splitters (on 5 different items), you easily prove that every decision point operates at the full 15/30/45 items/second (yellow/red/blue belts respectively).


It's more complicated when you have many types of item and not many of each item, because a filter inserter can do five types and a splitter can do one.

And if you add some wires, you can have each inserter automatically grab whatever is directly in front of it that isn't on a blacklist. At that point a max-throughput build with inserters is a big but roughly fixed size, while a build for splitters is proportional to the number of items.


Who said anything about throughput ?

Splitters can still only filter out a single item type, so if your goal is to do comparisons using a single entity (for clarity reasons for instance), they won't cut it.

----

Also, Factorio speedrunning has no metagame, since it's not a PvP game (well, aside from the less played PvP mode) : speedrunners don't have to adapt to changes in tactics of other speedrunners, they only have to learn new tricks that other speedrunners might discover, which is part of discovering the game itself.


Meta doesn't mean metagame.

"The Meta" is how most other people play. The "standard" set of designs that experienced players have all discovered (and rediscovered) as you play the game. Ya know, 3-wire assembly machines feed into 2-green circuits kinda things.

There's patterns of play between players. We all got our own style, but some designs are universally deployed across all experienced players, because those designs are just so good.


You've just described what "metagame" is; "meta" means metagame.


There's no "metagame" in Factorio because its not a PvP game.

The "metagame" in Magic the Gathering or Starcraft revolves around the game before the game is played. For example, if I see that my opponent in Starcraft is a Terrain player (and I'm a Zerg player), I can study their games and see that they prefer to open 8-rax 3 marines very early harassments.

I then decide to practice 9-pool / 6-ling and make sure I'm good at that before the game even starts, so that I'm well practiced against what they like to do. That's the "metagame", decisions you make before the game even starts.

Or in MTG, its knowing how good Rakdos decks are (or whatever is popular today) and finding counter-cards. You play the game before you even start the game, because you have to prepare your deck.

--------------

So there's no real "metagame" in Factorio. There is a "meta" however. (IE: what most players tend to do).


The game being PvP is not relevant. PvP games have a goal, beat the other player(s) or team(s). Factorio has a goal, launch as many rockets as you can. Not to mention, within the community or your own imagination there are an infinite amount of niche game styles encompassing goals of their own. For example: speed runs, smallest footprint, highest efficiency, most advanced automation, most chaotic pipes, CPU design, actual PvP mode (yes it does have PvP, but in the spirit of my argument I'll ignore this)...the list goes on and on, and permutations of all of the above.

Regardless of the goal you choose, you're competing, expanding knowledge, and defining standards with yourself and/or the community in some way, be it by playing co-op mode or sharing your results / optimizations / innovations on forums. Similar to PvP games, reaching that goal requires extremely nuanced strategic planning and making tactical decisions, even before starting the game, drawing from personal and/or community experience, understanding and adapting to game patches that may affect strategies, etc.

Regarding "meta" vs "metagame," I still assert the former is simply shorthand for the latter.


I'm gonna disagree with you - accepting your definitions, Factorio has both a meta and a metagame. Because co-op can have a metagame, except instead of needing to know what your opponent will do in order to best hinder them, you need to know what your friend will do in order to best help them.

Point in case: try joining a deathworld server (with experienced players), and then spamming burner miners. Once you're immediately kicked, please ponder your statement that "there's no real 'metagame'".


That's still game, not metagame : burner miners are bad because polluting, that's just game knowledge, you would have trouble if you tried to do that yourself in your own Death World game.

But great point, co-op might require modifying your tactics to better work with teammates - in case of Factorio I know that you have to be WAY more careful to not make spaghetti in MP, which might also depend on the familiarity between the players - so I would say that you can potentially have a metagame as soon as you have MP, even without PvP.

But now that I think of it, you can potentially have a metagame even when playing against AI. It's just that it needs to be advanced enough to be able to play rock/paper/scissors and remember previous games against you. AFAIK these games exist, but because it's much less effort and better reward to support basic AI + MP, they tend to be vanishingly rare, especially for more complex games.


Filter inserters are what came to mind when I saw the title.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: