The Tao of Gaming

Boardgames and lesser pursuits

Posts Tagged ‘factorio

Factorio Tutorial — Simple Production Ratios

I see a fair number of questions on production ratios (on reddit) and amazingly, nothing on the wiki seems to deal with it, so here it goes.  (This is really just a draft I plan on cleaning up and putting on the wiki).

The beginner says “Aha, I need a ton of <insert something here> so I’ll build some factories for it.” Then the factories sit idle most of the time. Adding factories doesn’t work if you don’t have the necessary inputs. So, how to figure that out? The easiest way is decide how many outputs per second you want.

For simple things (such as red science) you can build a lot of factories because the inputs (copper + gears) are either fast to make or basically infinite. (OK, not really, but if you aren’t smelting enough copper you have bigger problems). But let’s consider science pack 3 (blue science). Here’s the formula (as of 0.15):

1 Advanced (Red) Circuit + 1 Engine Unit + 1 Assembling Machine 1 (Grey) and 12 seconds makes 1 blue science.

(Thanks to Dave McW’s wonderful cheat sheet for this).

So, the easiest way to is to target some # of packs per second. Let’s say you want one. Picking round numbers makes it easy.

As the old saying goes ‘You can’t make a baby in one month by getting nine women pregnant.’

But you can average a baby a month that way.

To average 1 blue/second, we need 12 factories. (I’m ignoring the speed multiplier for assembling machines. It isn’t actually 12 seconds, depending on the factories. If you have all blue assembling machines (level 2) and change everything to level 3s, the number you get will speed up, but there will be no bottlenecks or slowdowns if they are all even).

Rule of thumb — If you want one unit output per second, you need as many factories as the time it takes.(Again, ignoring speed multipliers). If you want Z per second, you need Z times the number of seconds it takes one factory.

So, we need 12 factories producing blue packs. Now the reason to pick a nice round number like “one blue a second” is that we now know that we need one red circuit per second, one assembling machine per second, and one engine unit per second. We have to have enough inputs.

So, what does an Engine Unit take to make?

1 Steel + 1 Iron + 2 Pipes and 10 seconds makes an engine unit.

In order to support one blue science a second, we need 1 engine per second, which means 10 dedicated engine assemblers. Similarly, we need six red circuit assemblers. We only need half of an assembly one assembler, since it only takes 0.5 seconds. We can make these factories dedicated for this output (which may make sense for some objects) or we may have a big production area (for, say red circuits) so that any overflow can be used elsewhere.

Let’s look at a more complex example: Production Science Packs (purple)

1 Pumpjack + 1 Electric Engine Unit + 1 Electric Furnace + 14 seconds gives us 2 production science packs.

If we want one purple per second, we need seven factories dedicated to production (since they produce two each every 14 seconds).  Every fourteen seconds our seven factories dump out their two purple each to give us one purple per second, and every fourteen seconds they each take in a pumpjack, electric engine and furnace.

So we only need 7 of each input per 14 seconds.

We need 1/2 of a pumpjack per second (or 1 pump jack per two seconds). Ditto the other two.

Electric Engines take an engine unit (sigh), 2x electric circuits and 15 units of lubricant and 10 seconds.

If we want a dedicated line we need five factories (for 1 every other second), which will also require 5 factories for engine units (again, one every other second).  (Remember our rule, we want Z per second. Z is “one-half” and the factories take 10s, so 0.5 x 10 = 5 factories.

“It’s really easy to forget the two outputs per unit time,” he wrote looking at his current factory and suddenly realizing why there’s such a backlog of electric engine units. Ah well, I can divert some of that production and my pump jack production….In fact, there are a number of practical matters. If you only use purple science packs intermittently, you can get by with lower inputs and you’ll buffer some up while researching things that don’t require purple).

So far we’ve just spoken of production speed, but if you start getting into big numbers, you may have to worry about throughput. Maximum throughput of a yellow belt is 13.33 items per second.  If you need 15 or 20 items per second, you’ll requirea faster built (or a second belt) otherwise the first factories in your line will consume all the inputs and choke off your remaining factories. For our examples, that hasn’t been a huge concern (although technically it means that our one purple per second is actually 13.33 purple per 14s, assuming we have perfectly packed yellow belts). But if we wanted to go to two per second, we’d need to upgrade our belts (red belts are exactly twice as fast). If we wanted to go to 5 a second, that would require multiple lines (or maybe having a staggered line where we produce a few intermediates, then a few purples. But that way lies madness).

You can do this to see compare if you have enough labs to consume your science production (or too many labs) although that’s tricky because the colors you need (and times) change based on what you are researching. But typically research takes 30s or 60s (some of the early game research takes less, and some military researches take 45 and braking speed takes 35!) If you produce one (of each color per science) then you can keep 30 labs working all the time, and your 61st lab is guaranteed to sit idle (assuming you never stopped), but in reality you will stop from time to time….

I haven’t done the math on launching one Rocket Per Minute, but throughput becomes a big concern.  1000 Rocket Control Units (which are required per rocket) requires just shy of 40k iron per minute, 650ish iron per second (smelting and transport). Of course those actually have a bunch of intermediate steps, but that’s 17 lanes of express belts just for iron. (In reality, its a bunch of trains to various outposts to make green circuits, etc). And rockets have a bunch of other parts.

 

 

Written by taogaming

May 2, 2017 at 10:26 pm

Posted in Artificial Opponents, Strategy

Tagged with

If you were borderline on Factorio

They’ve just released 0.15 (up to 0.15.2 for some bugfixes). I’m sitting here waiting impatiently for it to download. Nuclear power, rebalanced combat, a ‘marathon’ mode (critical items take a lot more materials to build), and prebuilt settings for various styles of play….

It is a sign of good sportsmanship on my part that I didn’t skip gaming to download it yesterday (when it released), but that’s only because I’d promised that I’d teach High Frontier.

Update — It’s slower going (7 types of science instead of 4 — and purple science is no longer “stuff you get for free by killing natives”). I’ve got a horrible spaghetti maze because I haven’t been looking ahead. I should probably punt, but my goal is to use this base as a starting point for a big base, so its ok. But I do like many things:

  • Flamethrowers are still good, but much more expensive. Bullets have been buffed, because they were terrible.
  • Infinite science (even if some of it seems to be a bad deal, like mining productivity).
  • Lots of quality of life fixes (showing oil patches underground, etc).

I need to get logistics bots for this factory. It’s not pretty. Gotta find a new copper patch, too. Have basically run out.

Written by taogaming

April 25, 2017 at 7:32 pm

What is Winning? Factorio, High Frontier and the Goals of Games

(Update — Fixed some typos, fleshed out thoughts, added links. Also, I don’t wish to imply that Mombasa or Jump Drive are bad. I like both. I use them because I played them during the same week).

During my time at the Gathering, I played 31 games. Eclipse, Colonists and 3 plays of High Frontier took roughly 21 hours. The remaining 26 games took about 20 hours.

I spent my first two games of High Frontier doing terribly. I earned 1VP apart from Heroism (aka “The Challenger Explosion card” aka “Pity Points”).

Suppose winning just involved earning VPs at a fixed rate (ignoring opponents). It takes about 40 VP to win High Frontier, so I would ‘win’ after four hundred hours of gameplay.

Let’s flip that around. Wins/hour. For those first two games I earned 0.0025 WPH.

Among the other games I played were Mombasa and Jump Drive. For Mombasa I earned about .7 WPH, since I generated enough VP to win in my game, which was about 1.5 hours. A nice rate. My Jump Drive rate was 4 WPH for the Gathering, slightly higher than my typical 3 wins per hour. But I taught the game twice.

So, for 10 hours of High Frontier (roughly 1/4 of my gaming time) I was arguably the least effective game-player in all of the gathering. I’d have to go to ten gatherings to eke out a single win. (I have some true anti-skill in that game. But I did eventually get a bit better). Apparently I did OK in my only game with second edition (although it left me cold), but we did the fast start, so I may have randomly gotten a good setup.

Of course Wins Per Hour is a somewhat ludicrous idea. Isn’t it? It seems obvious, but winning isn’t the problem. And we routinely praise games for being “tight” or “fast” or having a high decision density.

I was already pondering the similarities between High Frontier (the current hotness) and Factorio (the current hotness is not necessarily a singular). I mean, they both have Rockets, but there were other similarities. I’d been thinking about it (vaguely) even before Jeroen asked me to pitch Factorio to him over breakfast. One of the random thoughts I blurted out was its old school scoring. (It was a long pitch. In my defense, I had previously warned him that if he asked about Factorio he should set aside a day for my answer).

Anyway, old school scoring. Launching one rocket (aka “Winning”) gives you a score of …. one. Its like how the original pinball machines had scores like 1-2-3 for each bumper, but now you get a million points for simply launching the ball. But after you get on the scoreboard in Factorio, you can keep going.

Factorio is optimization, but you are free to decide what to optimize:

1) Speed (the speedrun: how fast can you win?)

2) Throughput (how many rockets per unit time can you get? This can be subdivided to a final sprint, ignoring the setup time, or over the lifetime of your game).

3) Size (How small a factory can you get a rocket out of. This is currently a challenge on Reddit. Yes, these rockets would take roughly one year of real time to be built, but they are ludicrously small factories).

There are others. People play without using robots (or just logistic robots, or just personal robots), or without lasers, or trains. Some play peaceful, some play Deathworld. Factorio has a victory condition, but people often ignore it. You can keep playing after you win. It’s a sandbox. You make what you want. I’m watching “The Belt Diva” on youtube, and its like watching Bob Ross. Happy little conveyer belts. She wants to build a mega base. I have no idea if she’ll launch a rocket. Does it matter?

Mombasa is not a sandbox. Neither is Jump Drive. You could argue that High Frontier isn’t a sandbox either, it has VP and end game conditions, but the standard criticism again Eklund’s games is: “Great simulations with arbitrary endgame/VPs attached.” (The latest edition of High Frontier’s scoring seems reasonable and at about the right time).

Ignore the victory conditions for High Frontier and just play with the system and it would still work. (Also true of American Mega Fauna, or other games like Seven Ages. You could just tell people to play in the best interest of their species/nation and not the actual victory conditions).

You could add self imposed conditions on your game. Only use solar sails, or “try to get out to Neptune with only basic cards,” or “Make the Kessel Run in 7 parsecs.” For any Eklund game, you could take a reasonable goal (“Expand your species,” “Become President of Mexico,” “Kessel Run”) and if you achieved it, who cared what the VPs said?

High Frontier feels like a sandbox (much like all Sierra Madre games).

Can you do that with a Euro? I suppose you could just say “I’m trying to maximize my red cubes and not VP” But that seems silly. The goal is to maximize VP.

Objectively, there’s no difference. You lose, the other players win. You may throw the game by being silly. But in a sandbox game it feels acceptable.

You might not even mess up the other players. (If you didn’t mention you had mentally altered your victory conditions, they might not even notice, assuming you chose something ‘reasonable’) We’ve already quietly accepted the premise that in a sandbox game the VPs are somewhat not the point when we criticize the poor victory conditions in otherwise good games.

We wouldn’t play a Euro similarly broken. Wouldn’t make sense.

I once knew someone who built a Magic deck with 58 islands, 1 mountain and 1 fireball. (Several of my readers should remember him). The point was to build a deck that could theoretically win (if you drew the mountain, the fireball, and then managed to play 20 lands). He’d pull this deck and lose and lose to people who often did not notice that their opponent never did anything except play islands. He was truly Andy Kaufman, CCG player.

Or perhaps Stanley Milgram.

When I’m losing a game like High Frontier (or Combat Commander, or Seven Ages, or Here I Stand), I may flounder and grasp for a way out of my situation, but I feel perfectly happy just exploring the system even though the time invested is well out of proportion to what I’d normally give to a game.

I’ve dismissed countless games after one play (or less), games I’ve won included. Games I won especially. Those tend to fail the test of time (unless everyone was new).  A game where I can beat experienced players probably isn’t that deep, or has too much luck.

Whereas games I lost repeatedly include Titan, 18xx, Magic Realm (counting dying as a loss), Chess, Go, Bridge.

Clearly, I’m not optimizing my wins per hour. Knizia has that quote….

When playing a game, the goal is to win, but it is the goal that is important, not the winning

Pretty Zen, but I get (and approve) the idea. Those guys who crushed that sub-sub-Mendoza line Magic deck and chortled and kept playing it again and again to rack up wins? Not great gamers.

But is the goal really to win? Usually, yes. When I played American Megafauna with non-existent or terribly wrong Victory Conditions, I don’t recall having less fun. In High Frontier I started out trying to win. Mainly I was trying to do something constructive. A Winning-adjacent goal.

Some play games to explore systems, but that requires a system worth exploring. Which is not to say that Mombasa or Jump Drive’s design is not deep. Just (relatively) transparent. You can argue about the best path to winning, but that path is well defined. With Combat Commander or Magic Realm or High Frontier, you aren’t sure what’s going to happen, and sometimes the joy is just in unlocking the secret or even seeing that rocket take flight. In fact, defining the direction of the path is surprisingly hard … the player that did best may be the one that lost, according to the rules. It’s like that old Supreme Court definition of pornography. You know who won when you played it. Who cares what the rules said?

Its like watching your factory grow, then deciding what you want to optimize next.

Written by taogaming

April 15, 2017 at 11:04 pm

Posted in Ramblings

Tagged with ,

The rule of factorio is more factorio

While I’m waiting for 0.15 to drop, I’ve been building up a mega (for me) factory.

Not a mega-base, but pretty big

That’s about 35 hours of game time. After I launched a satellite, I stopped researching (and paused my second satellite because the strain of it nearly cost me my entire petrochemical resources). I’m building up my train network and moving slowly out.

I’ve already depleted three of my iron deposits and multiple copper deposits (those kinks in my train network were adhoc stations, now removed).

My train network is (sadly) spaghetti near the base and I originally built track with 1-2-1  trains (one locomotive, two cargos, a reversed locomotive). This lets you run automatically on a single track, but it doesn’t scale well and switching to run-on-right tracks (like highways, one lane going one direction and the other reversed) matters when you start wanting to have a lot of traffic. (I guess 6 trains counts as a lot for me). I’m in the process of switching to 1-2 double tracks, but you can still see where I haven’t finished (right near CopperLoading-2). I also have to hook up my iron and copper/oil networks together, so that I can move Iron from the NW to the SE loading station.

I now know to not put a station for ore loading on a main train line, just branch. And to keep straight lines (horizontal or vertical) for the most part, because you can blueprint a few patterns and be done. Ad-hoc is bad. Power isn’t a big concern. I’m mostly solar, which costs lots of real estate and resources, but since the point was to experiment with mega-bases, that’s ok. I do look forward to Nuclear power in 0.15.

Some of my solar arrays …. Don’t hand-build more than one….

I think I’ve got trains down …. I still miss a signal from time to time, but I’ve blueprinted them and rarely spend more than a minute before spotting a problem. A significant fraction of my power goes towards laser turrets, for good reason.

The natives are restless

Anyway, scaling up a mega base means more-more-more. More ore. More factories. More throughput. I’m building way too many modules, which is also way too few.

Magic (main) bus,carrying parts to all the assembly machines

Still too much spaghetti in the station (despite having a main bus) and I’m using logistics bots, but mainly as my valets (and a few odd things like my satellite construction and disposing of lumber by feeding trains). And I (again) found a lot of things I would do differently next time.

Still enjoying this.

Update — And then I ended up staying up another few hours to fix my train network, add two mining operations, re-jigger my oil network, expand my power supply by three mega-arrays (thankfully I noticed a few minutes before a catastrophic shortage). Next is ore processing throughput/switching to advanced ovens. And there’s four hours gone.

Written by taogaming

April 1, 2017 at 8:26 pm

More Factorio, all the time

  1. I watched Anti-Elitz’s new WR speedrun live. Another “Well, I can’t get any geekier” moment come and gone. (I actually went and had lunch during some of the early restarts, but I saw the last 1.5 hours of the WR run). As I said to Mrs. Tao — The difference between this and the rest of the nation’s sports viewing today is that they’ll never actually play college basketball. (Some of A.E.s new designs are clever, although he’s optimizing for fewest keystrokes/mouseclicks instead of other aspects. But optimization is optimization).
  2. I wrote an actual strategy post for Factorio and posted it to reddit. It concerns the use of negative feedback to promote factory resilience. And I say (with all due modesty) that a) that’s a better title and b) more use of control theory than my Master’s Thesis. That would have really depressed me 25 years ago.

Written by taogaming

March 26, 2017 at 7:49 pm

Posted in Uncategorized

Tagged with

Factorio Train Debugging

(This post is so that people can help debug my train layout. Nothing to see here unless you are into that).

Update — There is a single rail missing right above the station in the first picture. But you can’t see it because the toolbar blocks it, and the locomotive obscures the view. Thanks to everyone who helped.)

Here’s the layout. The train can go to OilLoading-1 or OilLoading-2 but not back to OilUnloading.

Map Layout

 

Here’s a look at the OilUnloading Station. I think it says NoPath because … honestly I’m not sure. If I hit “Go to OilLoading-1” it does. Is it trying to go to Unloading even though its there?

Unloading station

Walking to the NW, we can see a junction. The most likely culprit.

Main intersection

 

Continuing, its a long stretch with a few links just to break things up (for future traffic).

Next links

More links

 

Finally we get towards the outpost.

Join into loading station

loading station

 

Ideas appreciated.

Written by taogaming

March 25, 2017 at 5:40 pm

Posted in Uncategorized

Tagged with

All Factorio All The Time

And not much boardgaming. I actually took the TaoLing to game night this week and played …. one game of Jump Drive. Just wasn’t in the mood to play much. It happens. On the Factorio front I’m building a mega-ish-base (ish because once 0.15 drops I’ll probably upgrade and restart). An interesting learning experience (quite different from speed runs). Also learning how to deal with big alien bases. (I do hope they do something interesting with Laser Turrets in the new expansion, because there are lots of inferior weapon types, and Laser Turrets.

A large up front cost to build, but no ammo cost (assuming your power is fine).

I’ve actually just been watching a lot of Factorio as well. Watching the TaoLing play, and some youtube channels.

On the actual game front, been playing Sentinels and finally convinced the TaoLing to play Advanced villains, so its a challenge. Also played a few games of Fairy Tale.

Written by taogaming

March 22, 2017 at 8:44 pm