"Algorithms Discover Build Order From Hell"

#1 Posted by Bov11 (72 posts) -
#2 Posted by Orangefreak92 (60 posts) -

And this is why I will never play SC2 online.

#3 Posted by Brendan (7511 posts) -

This kind of stuff is the one problem I have with this game.  "Cheese".  The lines between what an affective strategy are and what a "cheap" strategy are seem to depend on whether someone won with it or lost with it, masked by how much "skill" someone needs to pull it off. 
#4 Posted by Diamond (8634 posts) -

I generally get turned off by any game where the ultimate result of any conflict can be decided within the first 1/10th of play time.  If what I read in that link is true, this cuts that time even shorter.

#5 Posted by Arkanti (254 posts) -

As  said in the comments, some well played regular Protoss defences at least have been shown to beat this, plus if anyone attacks whilst the zerg player is rushing for those roaches they're screwed.

#6 Edited by Three0neFive (2274 posts) -
@Brendan said:

" This kind of stuff is the one problem I have with this game.  "Cheese".  The lines between what an affective strategy are and what a "cheap" strategy are seem to depend on whether someone won with it or lost with it, masked by how much "skill" someone needs to pull it off.  "

Seriously. It seems depending on who you ask, everything is cheese. "derp you beat me so your strategy is cheese. fuck you you're a noob faggot nigger"
#7 Posted by keyhunter (3207 posts) -

Like he says. He loses to ling rushes and void rushes. Which is something every zerg and toss do 1 v 1 anyways. Also this strategy can be scouted and countered.

#8 Posted by sixghost (1679 posts) -
@Arkanti said:
" As  said in the comments, some well played regular Protoss defences at least have been shown to beat this, plus if anyone attacks whilst the zerg player is rushing for those roaches they're screwed. "
Good point. It annoys me when people see things like this and assume starcraft is just Build Order Memorization, The Game. You have to understand a couple things. 1: Something like this would never be useful past the first 4ish minutes of a game. Think of it like someone planning out a chess opening. If you just open this program and crank out the most efficient way to get X units, it's only of use to you if everything goes perfectly up until that point. Something like this would never be useful for a regular strategy because of the infinite amount of permutations a game can take even in 7-8 minutes. 2: This program will never come up with a build order that breaks the game because it still operates in reality. The people crying about the 7 roach rush would most likely be crying about this, or some other, rush build whether or not this Build Order machine exists. All it does is iron out some inefficiencies in the build to reduce the timing of the attack by a few seconds. Nothing proper scouting and play can't stop.
#9 Posted by endless_void (706 posts) -

protoss players could probably fend this off easier (forcefields would definitely help). terran will have problems though :|

#10 Posted by nintendoeats (5975 posts) -

I only seem to enjoy starcraft when both sides turtle and you get massive battles. This is really cool, but christing-racist-hell do i have no interest in coming up with or combating these build orders.

#11 Posted by Feanor (1377 posts) -

If this build is unbeatable then, we would have seen every Zerg at MLG use it. Its just one of those builds that is going to rape lower end players for a while. Just like the 6 pool or 2 gate Zealot Rush.

#12 Posted by sixghost (1679 posts) -
@endless_void said:
" protoss players could probably fend this off easier (forcefields would definitely help). terran will have problems though :| "
scout extremely early roach warren, build bunker, place marauder in bunker. Rush stopped.
#13 Edited by Ares42 (2443 posts) -
@Brendan said:

" This kind of stuff is the one problem I have with this game.  "Cheese".  The lines between what an affective strategy are and what a "cheap" strategy are seem to depend on whether someone won with it or lost with it, masked by how much "skill" someone needs to pull it off.  "

Think you have missed what "cheese" usually refers to. There's a pretty common concept when it comes to strategy , nicknamed "all in", where a player does a move that's either gonna win him the game or lose horribly. If someone uses the same concept from the very get-go of a game (which makes it very hard for the opponent to react to it), it's called "cheese". It's generally frowned upon because it makes the game into a coin-flip.
#14 Posted by pause422 (6153 posts) -

This along with anything else in existence can be scouted/countered by just paying attention to what your opponent is doing. Its that simple, really. If you generally don't like the game, thats fine, don't play it. People acting like things like cheese in general, and this, are some absolute things that will work and are reasons not to play, however,  just stop. Its bullshit, and it makes you look dumb.

#15 Posted by StarvingGamer (7559 posts) -

Random gave me Zerg 3x in a row so Fate was basically forcing me to test out this build.  Destroyed two Terrans and lost to a Zerg that timing pushed right when I started my Roach Warren with a billion Lings and a few Drones to build Sunks.

#16 Posted by Rayeth (1000 posts) -

Here's the thing about this.  This is a total all-in.   If it is executed wrong, counter-cheese'd, or stopped the player is totally screwed.
 
You stop producing drones very early and if your 7 roach rush fails you will be totally boned on economy and will be overrun by the other player.
 
So yes it's a very powerful build, especially vs Protoss, but will not be the end-all be-all of the matchup.

#17 Posted by Mr_Skeleton (5117 posts) -

I hate the Stracraft player's logic, if it works it sucks.

#18 Posted by Ares42 (2443 posts) -
@Mr_Skeleton: @Mr_Skeleton said:
" I hate the Stracraft player's logic, if it works it sucks. "
It's actually "if it's not reliable, it sucks". You can do this and you will win alot, but you will also lose alot once people adjust. It's basically just a crapshoot. If you lose a game and never ever made a mistake according to your plan, how was that plan good ?
#19 Posted by SamDrugbringer (1099 posts) -
@Ares42 said:
" @Mr_Skeleton: @Mr_Skeleton said:
" I hate the Stracraft player's logic, if it works it sucks. "
It's actually "if it's not reliable, it sucks". You can do this and you will win alot, but you will also lose alot once people adjust. It's basically just a crapshoot. If you lose a game and never ever made a mistake according to your plan, how was that plan good ? "
The problem is with Ladder, while a build like this CAN be countered, the fact is they'll wrack up a lot of wins and do it, beacuse many people will simply be unprepared. 
 
Anyway, I just think it's cool they can do this.
#20 Posted by Skytylz (4015 posts) -
@Rayeth said:
" Here's the thing about this.  This is a total all-in.   If it is executed wrong, counter-cheese'd, or stopped the player is totally screwed.  You stop producing drones very early and if your 7 roach rush fails you will be totally boned on economy and will be overrun by the other player.  So yes it's a very powerful build, especially vs Protoss, but will not be the end-all be-all of the matchup. "
If he doesn't pull his drones it's not that all in, sure a rush would destroy him but most games he'll be fine.  If the enemy has proper defenses then just pull back and he has a very strong early game army that he can just transition out of.
#21 Posted by Ares42 (2443 posts) -
@SamDrugbringer said:
" @Ares42 said:
" @Mr_Skeleton: @Mr_Skeleton said:
" I hate the Stracraft player's logic, if it works it sucks. "
It's actually "if it's not reliable, it sucks". You can do this and you will win alot, but you will also lose alot once people adjust. It's basically just a crapshoot. If you lose a game and never ever made a mistake according to your plan, how was that plan good ? "
The problem is with Ladder, while a build like this CAN be countered, the fact is they'll wrack up a lot of wins and do it, beacuse many people will simply be unprepared.  Anyway, I just think it's cool they can do this. "
Sure, they 'll rack up alot of wins for a week or so, and then either Blizzard deems it too powerful and it gets nerfed or people discover new starting builds that can deal with it. It's not like stuff like this only happens in Starcraft though, just take something like cross-map nading in MW2 as an example. People always find new "tricks" to games that are unintended faults, and no matter what game it is it's always considered cheap to exploit them.
#22 Posted by Rayeth (1000 posts) -
@Skytylz: I don't really agree with you there.  If you are cutting workers very early game like this, even if you don't pull them to attack, you are going to be at a severe economic disadvantage.  
 
I guess its a difference of opinion but really cutting so many workers early will cripple you later.  Following this BO will basically make you dead to a counter attack later if your rush doesn't do severe damage.   I would consider anything that doesn't have the ability to defend against a counter attack an all-in because you will die if you don't succeed same as if you pulled the drones.  Heck you might as well pull the drones as far as I'm concerned.  Not a big deal, people have different opinions.  =)
 
I don't really like builds like this 7RR and the 5RR because I suck at micro, so my roaches just end up dying to microed marauders anyhow.  lol.
#23 Posted by StarvingGamer (7559 posts) -
@Rayeth With the 7RR I was able to simultaneously attack a Terran player while defending myself from a 3 proxy rax marine rush.

Also vs a Zerg player who went for a fast expand I was able to keep enough pressure on him despite the fact that he scouted me and transitioned into a double expo muta roach ling macro game

Finally vs a Terran player that scouted me right when I started my Roach Warren, I was able to bust his base and take out his Starport before he finished his first Banshee

Bottom line, this is a very buff build. I haven't been able to try it out vs Toss because of the way Random has been working out for me, but I imagine that's the matchup where its going to be Super Effective!
#24 Posted by AngryRedPlumber (211 posts) -

Interesting.  Luckily I'm not nearly good enough of a SCII player for this to affect me.

#25 Posted by ryanwho (12082 posts) -

When you're copying algorythyms you're not even playing the game. This is basically an autism test.

#26 Posted by Zero_ (1970 posts) -
@ryanwho said:
" When you're copying algorythyms you're not even playing the game. This is basically an autism test. "
It's no different to any other build order - a build order is just the most effective/efficient route to get to a particular path you want to go, only this algorithm optimises it even further than first thought.
#27 Posted by ryanwho (12082 posts) -
@Zero_ said:
" @ryanwho said:
" When you're copying algorythyms you're not even playing the game. This is basically an autism test. "
It's no different to any other build order - a build order is just the most effective/efficient route to get to a particular path you want to go, only this algorithm optimises it even further than first thought. "
But you're not playing the game. You're copying an algorithm because you wanna be above Platinum. If anything upsets your precious build order you'll probably rage quit. This happened a lot while I was still playing, people quitting because their carbon copy technique hit a hiccup. There's no joy to be found here and like I said, you may have Autism if this is how you choose to play a game.
#28 Posted by Chubbaluphigous (571 posts) -

I am going to do what I really shouldn't, and use this for my 5 placement matches.  I also have a build that is similar to this that will get me 7 roaches and 4 mutas with an expansion in about 7:30 in case I scout and find a heavy cannon wall.  I just want to see what will happen.  I find the algorithm program fascinating from a hypothetical data stand point.  "What happens if I do X".  I am really interested to what a Terran and Protoss one will figure out.

#29 Posted by Hailinel (22704 posts) -

I just read the topic title as "Al Gore Rhythms Discover Build Order From Hell."
 
Mother of god.

#30 Posted by Carbon (23 posts) -

 I've encountered this several times on the ladder now, and I haven't lost yet. Zerg cannot stop you from getting a scout in on a two player map with such as short distance between bases like the Steppes of War map in the video. For some reason, that Terran didn't scout into the base even though he built an engineering bay at the natural expansion of the Zerg. That's meant to block and delay a fast expanding Zerg, but it actually puts the Terran player far behind if the Zerg isn't even going to expand. He went straight to tanks without even knowing what the Zerg was doing. Both players were on auto pilot and one player's build beat the other, which is not how Starcraft should be played.
 
If he had poked into the main of the Zerg, he would have seen a fast spawning pool along with gas. The spawning pool tells you that he's either rushing, fast teching, or just safe expanding by getting his pool for defense against a marine rush before building the hatchery. The gas gives warning that a fast Zergling rush with speed (100 gas) is coming, or a fast one base lair (100 gas) tech rush, or a roach rush. If it's a roach rush, the roach warren will be built very soon, so just have your SCV hang around.

With the standard Terran opening (orbital on 15 with gas at 13), the Terran will be right at the moment where he needs to decide between a fast expansion by saving 400 minerals for a command center, or just building a second barrack to defend a rush. After seeing the roach warren, just throw down the second barracks and get a tech lab on your current one. Make a Marauder (6 range attack) and send it out right away to kill the queen (3 range attack, moves very slow off creep, easy kill for the marauder), which will be about in the middle of no where with no chance of living vs one marauder, then send it straight back home. Your SCV will see the queen leaving the base, so it's not like you're randomly sending out a marauder and hoping for a queen kill.

If the Zerg player wants to make Zerglings to kill the marauder, not only will he need to use two or three larva on them, which is less roaches for the rush, but he'll need to build them preemptively as an escort for the queen or they won't hatch in time to save it. Since you're playing against Zerg, you should have walled in. You can have your marauders attack the roaches over your wall in while you repair the supply depots. Just pull all of your SCVs except two or three on gas, and keep dropping mules. Even if he breaks your wall in, the drones will meet your SCVs and the marauders will win you the fight. You should have at least four marauders when the rush hits you, even on the fastest rush maps like Steppes of War. Four marauders two hit a roach, and the roaches can't get within range of the marauders to attack back.

The roaches will likely kill a lot of your SCVs that are repairing, but he pulled all of his drones (if he didn't, you win easy with concussive shells) with the attack, so even if you lose all of your SCVs, you're ahead because of mules. If he pulls his drones back, then he ditched the rush and you'll still have plenty of SCVs as well. Just expand and out muscle him economically later in the game. If he loses all his drones during the attack, just put down more barracks a and just make marines, continuing marauder production from your first two barracks (no reactors, they take too long to build for a counter attack). You want a lot of marines for the counter attack since he'll likely just make Zerglings to counter your marauders. Rebuild your wall if you need to and attack quickly.

I'm not sure what Protoss does against this rush, but I think the fact that very few Zergs at a high level are bothering with it makes it obvious that it's not that effective against a good Protoss.

#31 Posted by Jeffsekai (7017 posts) -

This isn't cheese, the zerg player is not crippling himself after this. He will be able to grab an expansion during the push and then sky rocket ahead of the opponent.

#32 Posted by gunslingerNZ (1896 posts) -

Personally I think moving the queen out is about one of the dumbest things you could do. Sure you get the extra attack power but if you don't kill the scouting worker (which you won't with this build) He will take the watch tower and see your slow ass queen when she's only halfway across the map. That gives him plenty of time to bunker up and cut worker production to build extra units. 
 
@Jeffsekai said:

" This isn't cheese, the zerg player is not crippling himself after this. He will be able to grab an expansion during the push and then sky rocket ahead of the opponent. "
It's debatable whether it's cheese or not. If he sends his queen and/or workers this is definitely a cheese build. If he doesn't I'd just consider it a standard rush. Also with the amount of workers you cut to get the early 7 roach grabbing an expansion isn't really a good strategy unless you just want the extra larvae. You'd be better to build a bunch of drones while the attack is going on and then either grab the expo or rebuild defences depending on how the attack goes.
#33 Posted by Turambar (6484 posts) -
@Jeffsekai said:
" This isn't cheese, the zerg player is not crippling himself after this. He will be able to grab an expansion during the push and then sky rocket ahead of the opponent. "
An expansion without any workers isn't going to be helping anyone.
#34 Posted by Jeffsekai (7017 posts) -
@Turambar said:
" @Jeffsekai said:
" This isn't cheese, the zerg player is not crippling himself after this. He will be able to grab an expansion during the push and then sky rocket ahead of the opponent. "
An expansion without any workers isn't going to be helping anyone. "
I dont see why you have to do exactly what the guy in the video does, 7 roaches at that time in the game can still cause alot of pressure.
#35 Posted by Turambar (6484 posts) -
@Jeffsekai said:
" @Turambar said:
" @Jeffsekai said:
" This isn't cheese, the zerg player is not crippling himself after this. He will be able to grab an expansion during the push and then sky rocket ahead of the opponent. "
An expansion without any workers isn't going to be helping anyone. "
I dont see why you have to do exactly what the guy in the video does, 7 roaches at that time in the game can still cause alot of pressure. "
I know that.  I'm just saying 16 workers + however many you make during that attack isn't really going to warrant a new expansion no matter how true idea of "expand while attacking" is most of the time.  Building a queen and droning the hell up would be more ideal.
#36 Posted by Jeffsekai (7017 posts) -
@Turambar said:
" @Jeffsekai said:
" @Turambar said:
" @Jeffsekai said:
" This isn't cheese, the zerg player is not crippling himself after this. He will be able to grab an expansion during the push and then sky rocket ahead of the opponent. "
An expansion without any workers isn't going to be helping anyone. "
I dont see why you have to do exactly what the guy in the video does, 7 roaches at that time in the game can still cause alot of pressure. "
I know that.  I'm just saying 16 workers + however many you make during that attack isn't really going to warrant a new expansion no matter how true idea of "expand while attacking" is most of the time.  Building a queen and droning the hell up would be more ideal. "
He did build a queen.
#37 Posted by StarvingGamer (7559 posts) -
@Turambar If you throw down an expo as you make your first attack, you'll have been able to drone up beyond the 24 drones you needed for optimal Zerg play by the time it pops. Not only that but the sooner your second hatch is up the sooner you can start pooping more larvae the sooner you can build more drones. There's a reason why top Zerg players like to 14 hatch before they start "building a queen and droning up."

@ryanwho So you're saying that the people who are following the 5RR build they learned from TL.net are playing the game, but people following the 7RR build they learned from an algorithm aren't? Build orders exist for a reason, they give us the most
#38 Posted by StarvingGamer (7559 posts) -
@ryanwho: Sorry I was posting form my iPad which for some reason decided to cut off the bottom half of my post.  It should have read as follows:
 
So you're saying that the people who are following the 5RR build they learned from TL.net are playing the game, but people following the 7RR build they learned from an algorithm aren't? Build orders exist for a reason, they give us the most <blank> way to reach a certain goal, be it fastest or safest, defensive or aggressive.  The only difference in using an algorithm is that it doesn't take players months or years to discover.
 
I play random and, between all the races, I have about 10 build orders that I use in my regular rotation.  Adding this one to my roster doesn't mean I'm not playing the game.  If anything, I'd say build orders help us skip all the trial and error so we can start playing the real game.
 
If algorithms discover the cure for cancer, does that mean every doctor that uses that cure isn't really practicing medicine?
#39 Posted by RsistncE (4496 posts) -

Is anyone else more impressed by the fact that someone made an algorithm that will mash out the best build (instead of being impressed by the build order itself) or is it just me?

#40 Edited by Entus (253 posts) -
@RsistncE said:

" Is anyone else more impressed by the fact that someone made an algorithm that will mash out the best build (instead of being impressed by the build order itself) or is it just me? "

I'm not too surprised, especially with things like map analyzers that calculate map balance out there.

This edit will also create new pages on Giant Bomb for:

Beware, you are proposing to add brand new pages to the wiki along with your edits. Make sure this is what you intended. This will likely increase the time it takes for your changes to go live.

Comment and Save

Until you earn 1000 points all your submissions need to be vetted by other Giant Bomb users. This process takes no more than a few hours and we'll send you an email once approved.