As someone here for the AI, and only roughly familiar with SCII, this more beginner friendly dissection is very much appreciated. Watched a few matches before this but I got a bit lost with some of the micro/macro discussions and more nuanced plays without that bit of extra explanation.
Artosis is doing a great job here with the explanations. He has a ton of experience with casting, actually belonging to the best caster duo (Tasteless + Artosis) we have in SC2. I can fully agree with you that the "normal" cast of pro games is very unfriendly for beginners, the game is just too complex and you have to take a lot of things for granted to focus on the things that actually make the difference of winning and loosing (for example the detailed explanation of what a unit does and which army compositions are usually seen, you won't find that in a normal commentary). Too bad there isn't much like this on youtube.
Lowko provides a small trickle of background information about unit abilities, costs, etc. in many casts. I've played a bunch of SC:BW and SC2 on the order of ~5 games and I'm entertained by his casts. (FWIW.)
At this point i've probably followed this game for 3-5 years and I remember this map being one of the first ones I ever played in ranked. I'm unsure if pirates cove was the same map pool along with some other map I don't know the name of. Either way, it's very funny to see how beginner friendly Artosis can be despite having talked about starcraft since 1998 and starcraft 2 since 2010, maybe earlier since the beta and al that stuff. Either way it speaks a lot about him here make it so beginner friendly seemingly for people with little to no experience with the game. I smiled ear to ear hearing him not just explain the production tab but also the adept ability 'shade'. That is so fundamental I almost smiled off my jaw. High respect for helping beginners though. I think I probably was pretty adept at the game by the point this video came out but I think I got into it though videos like these, explaining basic ideas, concepts and information from the overlay(s).
Incredible how you make a complicated game I've never seen before accessible, thanks. Just shows how great commentary is more than half the joy of watching matches.
@Rick morty I can't believe how ass-pained Starcraft players still are about League, even down to its casting. You can hate the game for all I care, but calling the casting bad is the most ignorant and ridiculous shit I have ever heard from a starcraft elitist. League definitely has the most sophisticated commentating by this point, the casters being able to predict picks and bans in advance regularly and following map-movement before it happens. They work well with replays and the interface (like drawing on the map) and are knowledgable about players, the history and the meta. Just check out the casters notes by Froskurinn for *one* match: medium.com/@lolFroskurinn/group-c-funplus-phoenix-splyce-jteam-gam-esports-dcf2ef11e903 It really pisses me off, because I like both games. But Starcraft fanboys are such huge bitches with an inferiority complex towards a game, that they can't even get to appreciate any efforts made. Riot has stepped up with the professionalization of their events year after year after year. Their casters are top-notch and have helped to create some of the most memorable memories for me in gaming.
I am a veteran StarCraft player but listening to you break everything down so even the most inexperienced can understand it is incredibly appealing to listen to. It reminds me of how much stuff I take for granted.
I completely agree. This is a great breakdown. I realized how many things we take for granted when I tried to explain the game to a new player recently. It was surprisingly hard to do.
This is very very good for the beginners, which I believe is what he is targeting with this. The researchers who aren't playing the game are very interested.
@@szymonsabat8272 you are right. I guess that's really what I didnt like - the pauses. If he had been able to cram or shortened the explanations without pausing the game it would have been ideal.
I dont play SC2 at all but I know what all the abilities are and can deduce the reasoning behind tactics while i watch. Its all pretty straight forward. I agree this would be very educational to someone new but i wish this exact series wasnt the one used as an educational vehicle. There are millions of people like me who purely watch SC2 and dont play. It is fast paced excitement for me so yes there are millions of disappointed people who cannot enjoy this extremely important series.
@@michaelmao6180, true) And if he was super good in scout, tactics, strategy he can beat Serral too. But instead we see perfect execution of buildorder, dumb wall and super probe micro.
What an incredible performance by Alphastar. Finally we see what those disrupters can actually do if used properly. So refreshing to see a protoss beat a zerg, that too the best zerg in the world.
That's a really interesting point, because it adresses why StarCraft was interesting for DeepMind in the first place: SC is a game of strategy, and strategy is a big part of all the Alpha-projects. Go, chess, and now StarCraft. Like the commentator said; most pros focus on making a lot of those units that are strongest and deal the most damage, but Alpha does not because it focused on the strengths of the other units as well and found something better. In short, it found the best strategy.
@@bard6184 I disagree. AlphaStar does not find the best strategy, it find a strategy with the highest correlation with victory and gets to try it against an endless stream of pro players on ladder that are mostly unaware they are facing the AI. If players are allowed to face the AI repeatedly in that fashion, I suspect literally every strategy in SCII is exploitable (or near it, and any that is not is a clear red flag for game balance), so we would see more AlphaStar losses and DeepMind would be forced to stop pretending they're even better than they really are. Seriously, they've done a good job with AI, but their overhyped claims go beyond their real achievements, not only with AlphaStar, but with AlphaFold as well. I'll leave the Chess and Go assessments to those more familiar with DeepMind's claims vs. achievements in those areas, only speaking on what I've taken the time to look at critically myself.
@@zvxcvxcz I'm not so sure. Alpha gets stronger the more games it gets to play. Leela Chess, which is basically the same AI, the only difference being that it has played a lot more, is far stronger than Alpha ever was. But that is chess and this is SC, and I don't know much about SC. These facts may not apply to SC as it has in chess.
@@ayurge This. I remember in one of AS's early stages, it didn't have a limit on its apm and it was reaching upwards of a couple of thousand apm almost continuously. Granted most of its actions it was performing at the time weren't exactly good or efficient actions because it was still learning the game, but still. I think limiting the apm was one of the first things they had to do if they wanted to obtain any valuable data.
@@Enkarashaddam I am curios whether harder to control unit composition, not as popular among pro players as explained, is due to a superior ability to micro? There has to be a limit of how many different classes of units player can effectively control. Spellcasters or disruptors require a little more APM.
just a thought: what if alphastar's wall-ins are intentionally bad since it doenst really loose probes to harassment ? maybe it is trying to bait the oponents early units in so he can kill him and doenst have to fear an early cheese/allin ?
This is actually something I learned from these Alphastar games, if you warp in a hostile unit in the mineral line the zerglings can't a-move the probes because they prioritize hostile units or cannons, so even without the wall as long as you have anything hostile in the mineral line it's actually hard to get an efficient harassment with just zerglings. I still think not making the wall is a mistake though, I think because Alphastar perceives the map through a grid, the grid resolution makes it think it's walled on that map when it actually isn't.
Agreed. It's amazing how people say the restrictions are "human level" when it can target and pull back multiple probes, from a tight ball, while flicking between screens and unit groups and harassing on the other side of the map. There are just so many aspects that won't be on the same level until it's forced to play with a mouse and keyboard.
@@TheBbaughman there is an entire paper on why Alphastar´s "wits" cannot be properly gauged in starcraft 2 due to how much of the game´s outcome is determined by a superior control of the game. Many of the human limitations dont apply to alphastar (atleast yet), so any accomplishment in regards to its neural network, is overshadowed by its superior control of the game. Even on just 180 apm (which is just 3 commands per second), alphastar is gonna have better precision on issuing these 3 commands per second... perfectly hitting the exact pixel where it wants specific units to go. We have known for a long time that robots are better suited for precision work. The automobile industry, the chip industry... they all have been using robots to do precision work in lightning speed (and here physical limitations akin to humans somewhat still apply). In a digital setting, not even these physical limitations apply, thus the robot obviously becomes even more efficient in precision work (which ordering around units basically is). But "wits" is what alphastar wants to demonstrate, yet it cannot really do that without all other degrees of freedoms having been removed. And the most upsetting degree of freedom obviously is lightning fast, mistakeless, precise control. Any "control" mistakes it does, are not due to its hand slipping, or its eyes getting tired, or its fingers being too slow or losing precision due to an adrenaline rush during an engagement, they happen due to alphastar not knowing that they are mistakes. Those mistakes are purposefully executed. Whereas a human can do a mistake and KNOW that it is a mistake, but realize that he simply wasnt able to control it more efficiently for better results.
@@gh0un That's not true though. Alphastar will do misclicks and mistargeting, occasionally will attack its own units for one second. In this case, its that it can see each unit clearly in the ball, but it is still capable of misclicking actions.
@@DaRealDJ Not it does not misclick. If it does that, it does so because it thinks it's useful to do so (even if it isn't). Each click is purposefully executed, there are no control mistakes. It's just a bad decision to attack your own unit, but alphastar doesn't always know that at all stages of the game. It might have figured out that it isn't beneficial to do so in the early game, doesn't mean it knows not to do that in the late game. That's not a case of bad control, it's a case of bad decision making.
@@gh0un I agree with ur comments yet part of the goal of training an AI is to beat human. Being able to achieve this goal itself is something worth noting. Machines are designed to be more precise and mistakeless than human just as tigers are born stronger and faster than human. One thing limiting the power of machines/robots previously is that it can never be programmed perfectly. Now AI provided an answer to this and showed that the programming itself has evolved to a point where even if it's still not perfect, it is enough to bring the overall power of machines over human - which is impressive and never achieved before.
I really really appreciate how you take your time to explain what's going on. I enjoy watching starcraft video's but don't play myself. I tend to get stressed at the high level of management required but still enjoy the mechanics and gameplay. Most commentators assume those watching understand what's going on and cater to those people. And I totally understand and am not complaining. However I strongly felt you deserve praise for taking the time. Thank you a bunch!!
Yeah, I wonder if they couldn't find some way to create an input layer between the API and the graphical output that the AI is forced to work with so that it simulates actual mouse selection?
@@Zeuts85 the issue is that mouse group selection goes through the display system (rectangle creates a 3D cone-like collider that is based on the camera and the are of the viewport that has been selected that is intersected with unit colliders); but to let the machine "see" the game they (most likely) provide it with the game state itself, that is more like a super-detailed mini-map with all the data. It is a weird point because the whole point of AI is to make it Superhuman, as it is already in Go and Chess, if it can act fast and issue commands perfectly, ofcorse it will look perfect and superhuman. In addition, in a sense, when AlphaStar selects a group of units it (most likely) selects them one-by-one with a shift or ctrl (never played Starcraft, don't know the bindings), so you could argue that the humans have more tools
@@danielb270 During the matches against TLO and MaNa, AlphaStar interacted with the StarCraft game engine directly via its raw interface, meaning that it could observe the attributes of its own and its opponent’s visible units on the map directly, without having to move the camera - effectively playing with a zoomed out view of the game. In contrast, human players must explicitly manage an "economy of attention" to decide where to focus the camera. However, analysis of AlphaStar’s games suggests that it manages an implicit focus of attention. On average, agents “switched context” about 30 times per minute, similar to MaNa or TLO. Additionally, and subsequent to the matches, we developed a second version of AlphaStar. Like human players, this version of AlphaStar chooses when and where to move the camera, its perception is restricted to on-screen information, and action locations are restricted to its viewable region.
Man I'm so glad you made these videos- I understand RTS's but I'm not much of a star craft player, so back when Alpha Star played the first game against that pro last year, I really didn't understand what was going on, and I really wanted to. But now I love that someone is explaining what the hell is happening in these games for us non-starcraft people! Thank you!
First time I've ever watched SC2. I was just randomly searching about AI and came across Alphastar and Google Deepmind. Great video, this game looks fun.
"fun" he says. Hahaha... Yes, from a certain point of view the cocktail of adrenaline, stress, utter disappointment, and a few sprinkles of victorious joy could be called "fun", but I think you have to be a little crazy to make it much past diamond level. It's not for me, but I do love watching it.
I just realized. Alpha star wall is deceptively "bad". He wants to appear flawed to let the lings in. That's easy enough to discover but Alpha star is also taking into account a third resource when making this decisions. That is APM, if the zerg is harassing with lings, he is bound to have less apm to manage units at his base.
That would be really surprising if it can go that far, but on the other hand, what's more surprising would be such a crucial mistake on a basic strategy which is walling. Well spotted then
@@jonnysith It's not that scary because the AI learns optimum strategy without "understanding" why it works. It could've simply been that over tens of thousands of games, human players showed that the APM they sink into the harassment is statistically a net positive for the AI. So the AI doesn't understand that it's laying a trap, it just understands that it gets better results with this type of wall structure.
Yes but extremely slow with so many unnecessary pauses. I haven't played in almost 10 years and even understand how the new units work after seeing them in action. I'm sure the guys watching this for the AI can figure out intuitively how these characters work without making this video twice as long.
An interesting strategy from AlphaStar. If you watch the first push by Serral, Alpha waits to put down his force fields until after the biles have been used by Serral. I wonder if that is intentional so that Serral won't have the bile to break his force fields?
I'm new to Starcraft 2 but have played Starcraft 1, Warcraft 3 and other RTS as in the comment section players have mentioned your explanations break everything down so clearly for players new to the game. I have found your videos enjoyable and engaging. Keep up the good work.
Seems like one of the things we can learn from Alphastar here is that keeping those Oracles alive is crucial. A lot of the time we see players harassing with them in the early game and then mostly forgetting about them, but they were really crucial to Alphastar's strategy here.
I wonder if letting the lings through at the same time as the Oracle reaches Serral's base is actually a good play. Rather than being allowed to focus first on the oracle defense, then later on focus on using the lings, he was forced to do both at the same time, losing lots of drones and nearly wasting the lings.
That's very possible. The interesting thing about machine learning is that ONLY KNOWS WHAT WORKS, it doesn't know why. So maybe it did that a few times and got an advantage and won, and just kept doing it the same way.
@@ClayJitsu And to add to that, it might associate putting many buildings in the front with a full wall-in which is usually not at all what you want. Or making the gap too small for some units, etc.
ClayJitsu you also have to take into consideration that it took alot of things from human players, which, sometimes make absolutely no sence at all. For exmaple in some earlier games, we could see Alphastar attacking the rocks below the main ramp, which makes absolutely no sense for an AI to do. So this could have just been one of those things implemented in its plays.
Artosis is the perfect spokesdude for SC2!! So knowledgable and articulate about the finer details while also willing to break things down for outsiders to the game. Really entertaining and informative analysis.
Depends. In a lot of ways alpha star is very impressive. In other ways it highlights just how far AI still has to go. The attempt at a wall off is a good example. You'll never see it in a pro game because it makes no sense. It's like it built a wall for wall's sake without understanding what it's for.
@@heroclix0rz I wonder if it does not wall off so the imo & disruptors can go in and out .... maybe it never realized that you can drop a pylon to wall off during the early game, then destroy and replace it. Just guessing, because it seems strange, I make better walls than alpha star, and I'm still really far from beating serral on a lucky day.
@@heroclix0rz can we be sure it makes no sense at this stage? If it consistently does it 'wrong' as is claimed, that strongly hints it's no accident. The partial wall obviously helps it in some way, otherwise it wouldn't have learnt to do it. Would be interesting in a A* vs A* matchup, to force one to create a proper wall artificially (the game played to completion for both possible outcomes), then measure the difference in outcome over multiple trials.
@@Lee-pf6od The probe micro is ridiculous, so it was no reason to correct bad walling behavior. It clearly can make walls and chokepoints properly where it expects there to be more army fighting
@@Lee-pf6od You could also specifically teach it proper wall placement, inject those agents into the process, and see if they're validated or not. Basically, if the AI can consistently beat human players, we can feed it specific strategies and test them against itself in optimal conditions.
Man, we came here to look SC2 vids - pro games, pro casting, all the stuff. Do we really need those kidergarden basics? I would like to listen more about macro and players mistakes rather than "this is medivac - flying unit that can carry other land units, or apparently heal them"
@AmplifiedSilence to play devils advocate I dont see the logic in using the Serral/AS game as the vehicle to educate and explain the basics of the games. Why not a random master or GM ladder match? Furthermore if someone clicked on this video with 0 prior knowledge of SC2 then they wouldnt even know what AS is in the first place. This video is important to the community and seems like an inappropriate educational vehicle. Like it or not the constant and prolonged pauses greatly reduce the entertainment of the video. Its a shame.
This level of complete explanation of everything really does kill a lot of potential enjoyment. If I watch a chess match of the best player in the world versus a computer, I don't want to listen to people explain what the pieces do. I thought it went well with the main alphastar video, but the main appeal of more alphastar work is what it comes up with that top starcraft players don't do, the appeal is not the basics
Really Alphastar's brilliance comes from it's unit control. Like Artosis said multiple times, it doesn't really have the decision making of Serral (no wall off, recalling the oracles for 1 zergling, attacking with half its army), but what it doesn't have in strategic intelligence, it really knows how to make the most of its units (the probe micro, the warp prism juggling, its choice in unit comp going toward micro intensivity). This is why I take alphastar as less strategically brilliant AI, but more of an awareness of how to very precisely to use units. This is certainly different but I think less impressive than what AlphaGo has done. Starcraft pros are lauded for their unit control, but of course a computer will be better than them.
maybe the cost of units destroyed instead of the number of units destroyed would be a more appropriate indicator considering the differences between zerg and protoss
I mean it would be interesting but there is plenty of videos of having AI unlimited like the one where they have zerglings against tanks and avoiding splash damage. That one is nutty.
Speaking from DeepMind's experience, it has managed to do some things people have never seen before and I reckon when AlphaStar reaches maturity, Starcraft is about to get a ton more interesting.
At about 13:00 you point out AS overreaction but the point is that AS did not know what resources Serral had in that area. You could look at it as a probability decision. It's similar to war when you make a decision based on incomplete information.
About all those dumb "the pauses kill me" idiots - PLEASE ignore them. You're doing this in the most perfect possible fashion. Keep it up, especially WITH your pausing and pointing out all the little details!
The commentator is wrong about recalling the 2 oracles. The moment the AI did it I knew what it was for, defending against the upcoming zergling harasses. And even though it looks bad now, the two unblocked passsages into the main (instead of one and using a unit to block the enemy) may turn out to be better for the AI in the long run considering protoss buildings can't move. YOU may think it's bad, but the AI prefers to have over the top number of workers rather than going to just the limit, probably calculating that this is better from occasional worker losses from attacks, meaning zerglings getting in may not pose that much of a problem for them. To the AI, you losing units for not enough worker kills may actually be a win for them. Also, for the warp prisms, the AI may actually be correct. Using it as a defense means you cover a much wider area with less units, meaning your attack force may actually be larger. This is like the sequel to the AI playing GO. Top players worldwide may see "weird" behaviour that turns out humans have been doing wrong all these years.
Best SC2 slide-show I've seen in a while! - Could use more pauses and more explaining what's going on for all of us who just randomly ended up here without any prior knowledge about what SC2 is but chose to stay.........
I enjoyed the cast but I personally would enjoy it more if there were much less pauses throughout the cast. While I did enjoy the explanation and the breakdown of the skirmish at the 3rd base ( 25:00 ? ) I found myself click-forwarding through a lot of the breakdowns.
Cannons: its all about the total DPS. So with 200 supply you can only get a limited amount of power projection. Cannons are a way of raising this number.
I love how you explain this so in depth as if a brand new player were watching! My wife 9ver hears your videos when I watch them and I've noticed even she is starting to understand the gameplay (shes never played ever....didn't know it existed till she met me)
It's nice how you take time to explain the things down to the smallest detail it makes people who havent played, starcraft or some who just dont understand the smallest things, I honestly don't care for it. Not the type of presentation i would want to watch on a regular basis. I enjoy the way lowko, comment takes games, but i see the appeal, this is the first game video i've seen by artosisTV.
@@davidliu4134 Makes me wonder if the GPUs installed on gaming rigs are completely saturated and for those that want an extra challenge could utilize local GPUs for enhanced AI matches. Of course, only some GPUs could support this and it'd have to be tested. Maybe not ultra graphics settings possibly, IDK.
Coming here from AlphaStar's matches against TLO and MaNa, and hearing "you want 22 workers per base" made me smile so hard after hearing them say "you don't want more than 18 workers" in the vid from Jan 2019, I'm not sure if that means that AlphaStar really did change the human meta, but it sounds to me like people have started building more workers than they did before!
Obviously it's all speculation, as these things are a kind of black box. But I think that Artosis was referring to the total amount of workers including gas miners so that would mean still 16 is optimal. I think AlphaStar is overproducing probes so that when it expands it can transfer the workers and get its economy up faster on the expansion. I was surprised that they didn't mention this at all in the January cast. This was actually pretty common practice in the early days of SC2 and there was a lot of discussion about whether it was more or less efficient/worth doing, for some reason I think it fell out of favour, but I think it might actually be more effective. It also gives you a bit of a buffer in case you lose some workers, you don't lose any mining time while replacing them and maybe it overproduces on purpose because it expects to lose at least a few workers in any game in general. I think we can actually learn from the AI here. In short, I don't think it is oversaturating because it is more efficient to mine with that many workers. It is oversaturating for other reasons. (worker transfer and worker loss replacement)
thanks for the explanations. I used to play like ten years ago but quit while still a newbie, now im back as a newbie all over again with all the changes. Glad SC is still around.
AlphaStar is a very difficult opponent because of its consistent performance.. It doesn't get tired or nervous, doesn't doubt itself, has no trouble with its stamina. In any situation, it plays just like how it's programmed to be!! Indeed, sc2 is about being consistent, not about showing a few godlike micros occassionally.
No human can currently, no, but I don't think that's an example of superhuman mechanics. Just of mechanics that are theoretically possible for a human but unreasonable to expect of one. It's like having a bot with Serral micro in 2011 tournaments
@@Battleschnodder I agree but the whole concept of the study was to show the world that ai can strategically beat humans in a complex game that you have to take quick decisions in real time without relying on superior "mechanics" like the tlo and mana agents had. I do not believe that they proved that all.
Yeah, Zerglings really aren’t that much more powerful than workers early on and the precision micro takes away from that already relatively slim advantage. Definitely undermines the idea that it’s competing tactically as opposed to mechanically.
I think that moment was very reasonable. You can see on Alphastar's cam that it is almost exclusively looking at its probes during that defense. The oracle had been sent and is doing its job alone. Sure that's good probe micro, but Alphastar is exclusively focused on this probe micro and any pro gamer could do the same. As for the probe sent to expand, it does it after the zerglings have been dealt with. What's impressive is the decision making in what to micro. Most pro would have tried to switch their attention between harassing with the oracle and defending with the probes. Alphastar instead chose to trust the oracle to do its job and focus on defense. When I say most pro, that's evident with Seral himself. I've seen the same game from his perspective and at that very same moment he is switching his attention between attacking with the zerglings and defending the oracle, doing a rather poor job at both. The choice to expand at that time is still impressive because most players in this situation would still consider themselves on the defensive.
same as alphago, alphastar is trained to fight much stronger opponents, so it's playing it "safe". the goal of the game is to win with certainty, not to smash the opponent with a high risk strategy.
At 8:07 serral runs lings up the ramp, and the two adepts seem to give up and go back to guarding the front, only to turn around again once the probes are being attacked. I wonder if serral could have run the lings off to the side and morphed some banes. A human would stop it, but alphastar might be like "oh he left, problem solved".
I don't know, it seems impossible but having someone from deepmind or an AI pro to discuss how alphastar learned (specifically, how it managed to learn these standard ideas) would really help to show off how amazing this achievement is.
The problem with machine learning from what I know (and I'm definitely not an expert in any shape or form) is that you can't really say how it got to the conclusions it got to
They must have told it some stuff about how to play. There's no way human strategies are so optimal that an advanced machine just happened to come up with the same ideas through random experimenting. If there was, then why does the meta change so much?
@@ClayJitsu I was really surprised by that aswell, but the human copying might have come from playing vs humans and losing from them, the only thing it could do is learn from the players, and as artosis says, it does alot of weird stuff
im three minutes in wondering how artosis is really going to keep up with a Serral game while also providing the basics commentaries like explaining chronoboost :o
In general, it seems like Alpha uses a wider array (compared to the pros) of the tools the game offers it. In short, a more complex approach than the brute force most pros applies. And what if Alpha walls poorly on purpose, to let more units in, and thus kill more? Anyway, I don't know squat about SC, so I could be completely wrong.
We all know that without perfect probe micro, Alphastar would have lost. It is very apparent that Alphastart agent learned to imitate the placement of humans, but did not learn that those building placements are to create a full wall.
It wasn't imitating humans at all. Alphastar became so good by playing against itself millions and millions of times, and came to the conclusion that walls are a good thing to do. It literally learned nothing from humans. I do think the reason it doesn't full wall though is because it doesn't need to. It probably realizes that if Serral wants to sacrifice 20 lings to kill 3 workers, that's a good trade when he isn't going to push out anyways and has no other way of doing damage. imo the lack of full wall off is a bait for people to make mistakes running past.
@@Figgy20000 I think you do not know how alphastar is trained. It is true that is learned a lot by playing against itself. But with such a huge search dimension, it is often very hard for these agents to find any solution close to optimal. Instead they are often initialized or "bootstrapped" with a policy. There are few methods for bootstrapping but one of the more complicated and successfull one is "imitation learning". So alphastar agents are all bootstrapped with imitation learning, from player replays. Then they play for years between themselvesi to refine their behavior. A similar thing was done in their Go agent, AlphaGo, which also first observed human play and then went to refine itself. Afterwards, they also created another agent called AlphaGoZero, which learned without any bootstrapping this time. They have no such agent for sc2 yet, nor do I expect anything similar in the near future. Alphastar is literally walling of because it watched people wall off, but did not know why it was. It just knows that placing buildings in that area helps increase the winrate. As for the "it does not need to" argument, it definitely needs to wall off. An agent not walling of vs an agent that walls of will always favour the one that walls of. Alphastart has limited APM, the more it wastes APM to the perfect probe micro, the less it has for other tasks.
At 5:34 game time 12:33 video time there was recall of oracles, stalkers warp-in and it was commented as a waste, but 30 seconds later there was an attack at the 3rd base, so may be it's just a safe play when AlphaStar saw an opportunity for Serral?
I am very much a novice regarding Starcraft, AI and everything else connected to this video. I just read about and got interested in DeepMind's AI a few days ago. But watching this match (with very insightful commentary) and knowing it's an AI playing a human is completely captivating. Thank you so much for putting up this video!
thanks, you are doing very calm smooth replays ... can watch it good with a confident understanding on what the players reason is to loose against the meta-bot
How was AS able to move its units to 3rd expansion at 21:00 and build photo cannons without having any vision on Serral's units? Is that destructive rock within AS vision? So AS saw the rocks hp decreasing and hence relocated its units for battle?
It’s crazy to think this same ai is now being used to create swarms of drones to be used in actual warfare. It’s possible it used some of the same learning it gained here into the real world
I'm wondering,for eq for the probe harrass and scouting does alpha team implements "automated scenarios" in the ia program or did alphastar realized itself it is cost effective to do so like an human would? especially the lil probe mineral harrass in the very beginning
@@stkfr They release science papers on how they've made the various alpha series AIs like Alphastar, it's always been via self learning or guided self learning. But in either case it's ultimately learning how to play itself. It's in that sense very different from old style AIs that have pre-programmed reactions.
AlphaStar has a virtual mouse and keyboard, and is limited to using them at human speeds. Its mouse cursor cannot teleport from one part of the screen to the other, it has to move its mouse. However AlphaStar will never misclick, misaim, mistime, or misjudge a distance. It always issues exactly the order it intends to, in exactly the way it intends to, and this lets it perform ungodly micro.
When Serral saw that gap in the computer program's shitty wall... he should have noticed that it was a bug within the code, and he should have canceled his third base to mass lings from that point on, simply flooding the protoss base with them. Then, game over... The oracles can't "out DPS" them with limited energy, and the protoss had ZERO ground forces at the time the wall was exposed.
Alphastar sent only one zealot back to deal with 4 roaches because it estimated that the zealot will buy enough time to be able to warp in at home. Alphastar did just that while attacking Serral's base . That's incredible.
Thanks for the cast. Serral wasn't using his own keyboard or mouse (from what i heard). Really wish there would have been a proper tournament to see how good alpha star really was.
excellent commentary, perfectly suited for the newer viewes to get an explination of stategy and tactics on sc2, but also as fast pasted as a normal pro game for me to not be so bored but that. well done Artosis
@@JacobSimonsen people who are interested in deepmind but don't know much about starcraft. it doesn't really take a genius to make that connection you know.
Id argue that the deepmind AI ppl go watch other educational videos or play the free game yourself. In terms of president this video is extremely important to the community and it is a real shame it has been used as an educational video. This is like the superbowl or olympics so no it doesnt seem appropriate to pause the superbowl and explain what the lines on the grass mean or what shape the ball is.
i don't play starcraft and am solely interested in the AI, but you commentating and explaining what the fuck is going on is great appreciated thank you.
Artosis, dont' know about other games but in this game don't think that the hole in the wall is a mistake but intentional. I believe there is a reason for that. Probably the AI figured out that if it lets some zerglings inside his base his opponent will micro them and this would allow him to make a lot more kills than the zerlings. Even if the opponent has the same micro skills as him probably the AI saw that if there are zerlings in his base and both sides are microing to attack and defend workers at the same time the oracle will make more kills than the zerlings. When playing against human it is especially evident that even a little distraction to control the zerglings led to many kills of drones and less of probes. You can see that the oracle returned to attack the drones when the AI saw a lot of microing of zerglings in his base. This was not coincidence.
I'm starting to think that alphastar purposely has the hole in the wall-off so that Serral has to micro the lings AND defend against the oracle. It knows it can deal with the lings efficiently and thought it's worth the sac to take some APM from Serral.
Well to be fair, a lot of pro decisions are based on mitigating errors and units like disruptors are very hard to use well so unless you have supreme confidence in your mechanics you're probably better off focusing on immortals.
Great commentary. To the people complaining about pauses and explanations, why are you even here for? Lol. There're other vids of this match elsewhere. Sheesh.
At 13:30 couldn't alphastart just take 2 of those 5 adepts and attack the third base from the south? since 2 adept can kill one drone why attack one drone with 5 adepts. doesn't make sense...
Redundancy. Spread out adepts would die exponentially faster than 5 adepts, due to more zerglings being able to attack at once, fewer targets to focus on, and of course the extra health pool. Also, if 1 of the 2 adepts dies, it's basically impossible to get any more damage, whereas the pack of adepts is less affected by a couple of them dying: still able to secure kills.
As someone here for the AI, and only roughly familiar with SCII, this more beginner friendly dissection is very much appreciated. Watched a few matches before this but I got a bit lost with some of the micro/macro discussions and more nuanced plays without that bit of extra explanation.
Artosis is doing a great job here with the explanations. He has a ton of experience with casting, actually belonging to the best caster duo (Tasteless + Artosis) we have in SC2.
I can fully agree with you that the "normal" cast of pro games is very unfriendly for beginners, the game is just too complex and you have to take a lot of things for granted to focus on the things that actually make the difference of winning and loosing (for example the detailed explanation of what a unit does and which army compositions are usually seen, you won't find that in a normal commentary).
Too bad there isn't much like this on youtube.
Lowko provides a small trickle of background information about unit abilities, costs, etc. in many casts. I've played a bunch of SC:BW and SC2 on the order of ~5 games and I'm entertained by his casts. (FWIW.)
At this point i've probably followed this game for 3-5 years and I remember this map being one of the first ones I ever played in ranked. I'm unsure if pirates cove was the same map pool along with some other map I don't know the name of. Either way, it's very funny to see how beginner friendly Artosis can be despite having talked about starcraft since 1998 and starcraft 2 since 2010, maybe earlier since the beta and al that stuff. Either way it speaks a lot about him here make it so beginner friendly seemingly for people with little to no experience with the game. I smiled ear to ear hearing him not just explain the production tab but also the adept ability 'shade'. That is so fundamental I almost smiled off my jaw. High respect for helping beginners though. I think I probably was pretty adept at the game by the point this video came out but I think I got into it though videos like these, explaining basic ideas, concepts and information from the overlay(s).
Incredible how you make a complicated game I've never seen before accessible, thanks.
Just shows how great commentary is more than half the joy of watching matches.
@Rick morty I can't believe how ass-pained Starcraft players still are about League, even down to its casting. You can hate the game for all I care, but calling the casting bad is the most ignorant and ridiculous shit I have ever heard from a starcraft elitist. League definitely has the most sophisticated commentating by this point, the casters being able to predict picks and bans in advance regularly and following map-movement before it happens. They work well with replays and the interface (like drawing on the map) and are knowledgable about players, the history and the meta. Just check out the casters notes by Froskurinn for *one* match: medium.com/@lolFroskurinn/group-c-funplus-phoenix-splyce-jteam-gam-esports-dcf2ef11e903
It really pisses me off, because I like both games. But Starcraft fanboys are such huge bitches with an inferiority complex towards a game, that they can't even get to appreciate any efforts made. Riot has stepped up with the professionalization of their events year after year after year. Their casters are top-notch and have helped to create some of the most memorable memories for me in gaming.
to an extent made a 11 min video 33 mins long
a little bit too slow - I'm a complicate newb to starcraft but this isa gear to slow of intro nbd
league of lesbians is for kids who cant handle the real mechanics of dota. cry more noob
@@preventablesuffering6239 12 year old comments incoming...
I am a veteran StarCraft player but listening to you break everything down so even the most inexperienced can understand it is incredibly appealing to listen to. It reminds me of how much stuff I take for granted.
Funny how at the same time there's tons of comments from impatient people expecting a regular event cast.
@@genesises Hopefully there is hundreds of those out there already.
I completely agree. This is a great breakdown. I realized how many things we take for granted when I tried to explain the game to a new player recently. It was surprisingly hard to do.
@@TwoSentenceHorror-99 Knowing what a stalkers abilities are isn't bragging sir.
i do agree, although i hated the constant pausing.
I recommend playing the whole video through first and then going back and analyzing everything. I wanted to see the game lol
i can only agree this makes the action soooooooooo boring!
This is very very good for the beginners, which I believe is what he is targeting with this. The researchers who aren't playing the game are very interested.
what if just upload a separate video? one for full gameplay, one for full analysis. so it can also attract other viewers
yep. leaving because of it. Idgaf about your terrible extended narrative.
@@seanruffolo6083 No one cares if you leave.
Every game is someone’s first time with sc2. Great cast and explanations.
Too much explanation
Explanation is ok, but pausing is not :(
@@szymonsabat8272 you are right. I guess that's really what I didnt like - the pauses. If he had been able to cram or shortened the explanations without pausing the game it would have been ideal.
wrgg
To know before you watch if you are an advanced SC2 player:
game length = 11:21
video length = 33:48
many pauses and analyzes
@@qwp1026 More a ladder player who already knows what creep and bile are
I dont play SC2 at all but I know what all the abilities are and can deduce the reasoning behind tactics while i watch. Its all pretty straight forward. I agree this would be very educational to someone new but i wish this exact series wasnt the one used as an educational vehicle. There are millions of people like me who purely watch SC2 and dont play. It is fast paced excitement for me so yes there are millions of disappointed people who cannot enjoy this extremely important series.
it took so little damage from the lings, that was absurd.
overhuman micro
@@HitoryDensetsu Thats because its not human
@@rowanmonk5652, seriously??
On the other hand, if the wall off is good, it would have taken even less damage
@@michaelmao6180, true) And if he was super good in scout, tactics, strategy he can beat Serral too. But instead we see perfect execution of buildorder, dumb wall and super probe micro.
What an incredible performance by Alphastar. Finally we see what those disrupters can actually do if used properly. So refreshing to see a protoss beat a zerg, that too the best zerg in the world.
That's a really interesting point, because it adresses why StarCraft was interesting for DeepMind in the first place: SC is a game of strategy, and strategy is a big part of all the Alpha-projects. Go, chess, and now StarCraft.
Like the commentator said; most pros focus on making a lot of those units that are strongest and deal the most damage, but Alpha does not because it focused on the strengths of the other units as well and found something better.
In short, it found the best strategy.
@@bard6184 I disagree. AlphaStar does not find the best strategy, it find a strategy with the highest correlation with victory and gets to try it against an endless stream of pro players on ladder that are mostly unaware they are facing the AI. If players are allowed to face the AI repeatedly in that fashion, I suspect literally every strategy in SCII is exploitable (or near it, and any that is not is a clear red flag for game balance), so we would see more AlphaStar losses and DeepMind would be forced to stop pretending they're even better than they really are. Seriously, they've done a good job with AI, but their overhyped claims go beyond their real achievements, not only with AlphaStar, but with AlphaFold as well. I'll leave the Chess and Go assessments to those more familiar with DeepMind's claims vs. achievements in those areas, only speaking on what I've taken the time to look at critically myself.
@@zvxcvxcz I'm not so sure. Alpha gets stronger the more games it gets to play. Leela Chess, which is basically the same AI, the only difference being that it has played a lot more, is far stronger than Alpha ever was.
But that is chess and this is SC, and I don't know much about SC. These facts may not apply to SC as it has in chess.
@@bard6184 Ai also has better epm than humans due to them being ai's which is why they can use units such as disruptors a better than pro players.
Alphastar: *fends off medium sized army with five troops.*
Also alphastar: *sends large army to enemy base*
Me: this is where the fun begins.
Technically alphastar being an ai, means it would have superior multitasking ability and ought to have superior apm
@@ayurge This. I remember in one of AS's early stages, it didn't have a limit on its apm and it was reaching upwards of a couple of thousand apm almost continuously. Granted most of its actions it was performing at the time weren't exactly good or efficient actions because it was still learning the game, but still. I think limiting the apm was one of the first things they had to do if they wanted to obtain any valuable data.
@@Enkarashaddam I am curios whether harder to control unit composition, not as popular among pro players as explained, is due to a superior ability to micro? There has to be a limit of how many different classes of units player can effectively control. Spellcasters or disruptors require a little more APM.
just a thought: what if alphastar's wall-ins are intentionally bad since it doenst really loose probes to harassment ? maybe it is trying to bait the oponents early units in so he can kill him and doenst have to fear an early cheese/allin ?
yeah just like how everyone thought alphastar was an idiot to "over saturate" the bases.
Thats next level
This is actually something I learned from these Alphastar games, if you warp in a hostile unit in the mineral line the zerglings can't a-move the probes because they prioritize hostile units or cannons, so even without the wall as long as you have anything hostile in the mineral line it's actually hard to get an efficient harassment with just zerglings.
I still think not making the wall is a mistake though, I think because Alphastar perceives the map through a grid, the grid resolution makes it think it's walled on that map when it actually isn't.
Hold position micro beats the hostile unit trick
Take a look at the AlphaGo documentary. It does similar things there.
that probe micro against the zerglings though
Agreed. It's amazing how people say the restrictions are "human level" when it can target and pull back multiple probes, from a tight ball, while flicking between screens and unit groups and harassing on the other side of the map. There are just so many aspects that won't be on the same level until it's forced to play with a mouse and keyboard.
@@TheBbaughman there is an entire paper on why Alphastar´s "wits" cannot be properly gauged in starcraft 2 due to how much of the game´s outcome is determined by a superior control of the game. Many of the human limitations dont apply to alphastar (atleast yet), so any accomplishment in regards to its neural network, is overshadowed by its superior control of the game.
Even on just 180 apm (which is just 3 commands per second), alphastar is gonna have better precision on issuing these 3 commands per second... perfectly hitting the exact pixel where it wants specific units to go. We have known for a long time that robots are better suited for precision work. The automobile industry, the chip industry... they all have been using robots to do precision work in lightning speed (and here physical limitations akin to humans somewhat still apply). In a digital setting, not even these physical limitations apply, thus the robot obviously becomes even more efficient in precision work (which ordering around units basically is).
But "wits" is what alphastar wants to demonstrate, yet it cannot really do that without all other degrees of freedoms having been removed. And the most upsetting degree of freedom obviously is lightning fast, mistakeless, precise control. Any "control" mistakes it does, are not due to its hand slipping, or its eyes getting tired, or its fingers being too slow or losing precision due to an adrenaline rush during an engagement, they happen due to alphastar not knowing that they are mistakes. Those mistakes are purposefully executed. Whereas a human can do a mistake and KNOW that it is a mistake, but realize that he simply wasnt able to control it more efficiently for better results.
@@gh0un That's not true though. Alphastar will do misclicks and mistargeting, occasionally will attack its own units for one second. In this case, its that it can see each unit clearly in the ball, but it is still capable of misclicking actions.
@@DaRealDJ Not it does not misclick. If it does that, it does so because it thinks it's useful to do so (even if it isn't). Each click is purposefully executed, there are no control mistakes. It's just a bad decision to attack your own unit, but alphastar doesn't always know that at all stages of the game. It might have figured out that it isn't beneficial to do so in the early game, doesn't mean it knows not to do that in the late game. That's not a case of bad control, it's a case of bad decision making.
@@gh0un I agree with ur comments yet part of the goal of training an AI is to beat human. Being able to achieve this goal itself is something worth noting. Machines are designed to be more precise and mistakeless than human just as tigers are born stronger and faster than human. One thing limiting the power of machines/robots previously is that it can never be programmed perfectly. Now AI provided an answer to this and showed that the programming itself has evolved to a point where even if it's still not perfect, it is enough to bring the overall power of machines over human - which is impressive and never achieved before.
I really really appreciate how you take your time to explain what's going on. I enjoy watching starcraft video's but don't play myself. I tend to get stressed at the high level of management required but still enjoy the mechanics and gameplay. Most commentators assume those watching understand what's going on and cater to those people. And I totally understand and am not complaining. However I strongly felt you deserve praise for taking the time. Thank you a bunch!!
Play desert strike in arcade mode it’s stress level is lower and it can be a great intro to unit counters if you do ever want to play.
The probe micro looked a bit superhuman to me. It was simultaneously selecting several of the weakest probes.
Yes, because AI use game API for control instead keyboard and mouse.
Yeah, I wonder if they couldn't find some way to create an input layer between the API and the graphical output that the AI is forced to work with so that it simulates actual mouse selection?
@@Zeuts85 the issue is that mouse group selection goes through the display system (rectangle creates a 3D cone-like collider that is based on the camera and the are of the viewport that has been selected that is intersected with unit colliders); but to let the machine "see" the game they (most likely) provide it with the game state itself, that is more like a super-detailed mini-map with all the data.
It is a weird point because the whole point of AI is to make it Superhuman, as it is already in Go and Chess, if it can act fast and issue commands perfectly, ofcorse it will look perfect and superhuman.
In addition, in a sense, when AlphaStar selects a group of units it (most likely) selects them one-by-one with a shift or ctrl (never played Starcraft, don't know the bindings), so you could argue that the humans have more tools
@@danielb270 During the matches against TLO and MaNa, AlphaStar interacted with the StarCraft game engine directly via its raw interface, meaning that it could observe the attributes of its own and its opponent’s visible units on the map directly, without having to move the camera - effectively playing with a zoomed out view of the game. In contrast, human players must explicitly manage an "economy of attention" to decide where to focus the camera. However, analysis of AlphaStar’s games suggests that it manages an implicit focus of attention. On average, agents “switched context” about 30 times per minute, similar to MaNa or TLO.
Additionally, and subsequent to the matches, we developed a second version of AlphaStar. Like human players, this version of AlphaStar chooses when and where to move the camera, its perception is restricted to on-screen information, and action locations are restricted to its viewable region.
Man I'm so glad you made these videos- I understand RTS's but I'm not much of a star craft player, so back when Alpha Star played the first game against that pro last year, I really didn't understand what was going on, and I really wanted to. But now I love that someone is explaining what the hell is happening in these games for us non-starcraft people! Thank you!
As someone who hasn't played in years, thank you for explaining the shade play
the comment is so patient and really taking care of all the audience. great work.
An advance AI general, in a world where drones exists, GG.
Give AlphaStar the control over a "terran" army, nothing bad can happen! #we#want#SkyNet.
Are you an effective team?
That was the plot of Ace Combat 7 actually
has john connor actually been born?
It’s actually so satisfying to see Protoss beat serral
Goblin did it at NW
@@KugasKuoska that was 1 map though. and it was a pretty weird game that began with a cannon rush.
First time I've ever watched SC2. I was just randomly searching about AI and came across Alphastar and Google Deepmind. Great video, this game looks fun.
"fun" he says. Hahaha... Yes, from a certain point of view the cocktail of adrenaline, stress, utter disappointment, and a few sprinkles of victorious joy could be called "fun", but I think you have to be a little crazy to make it much past diamond level. It's not for me, but I do love watching it.
this game is like more demanding than chess or alpha go because you need also a lot of mental speed
I just realized. Alpha star wall is deceptively "bad". He wants to appear flawed to let the lings in. That's easy enough to discover but Alpha star is also taking into account a third resource when making this decisions. That is APM, if the zerg is harassing with lings, he is bound to have less apm to manage units at his base.
That would be really surprising if it can go that far, but on the other hand, what's more surprising would be such a crucial mistake on a basic strategy which is walling. Well spotted then
the use of deception by an AI has so many implications it's almost too scary to think about.
Maybe it left the wider opening in the wall so that it can easily move wider units through it (Archons, Immortals, etc.)?
@@jonnysith It's not that scary because the AI learns optimum strategy without "understanding" why it works. It could've simply been that over tens of thousands of games, human players showed that the APM they sink into the harassment is statistically a net positive for the AI. So the AI doesn't understand that it's laying a trap, it just understands that it gets better results with this type of wall structure.
@@vladenache6937 Alphastar did not learn from humans
th-cam.com/video/jtlrWblOyP4/w-d-xo.html
Very accessible analysis, love to see these games.
Yes but extremely slow with so many unnecessary pauses. I haven't played in almost 10 years and even understand how the new units work after seeing them in action. I'm sure the guys watching this for the AI can figure out intuitively how these characters work without making this video twice as long.
@@zackj997 Not everyone plays the game, some are just here to watch the AI
Thank you, Artosis, for uploading the AlphaStar games. Your commentary is very insightful.
The beginning of sky net. Why would terminators look like humans when they can be Protoss.
An interesting strategy from AlphaStar. If you watch the first push by Serral, Alpha waits to put down his force fields until after the biles have been used by Serral. I wonder if that is intentional so that Serral won't have the bile to break his force fields?
I'm new to Starcraft 2 but have played Starcraft 1, Warcraft 3 and other RTS as in the comment section players have mentioned your explanations break everything down so clearly for players new to the game. I have found your videos enjoyable and engaging. Keep up the good work.
Seems like one of the things we can learn from Alphastar here is that keeping those Oracles alive is crucial. A lot of the time we see players harassing with them in the early game and then mostly forgetting about them, but they were really crucial to Alphastar's strategy here.
where can I find the replays?
Now show us the Florencio vs AlphaStar replays.
xmfcx this is a must see, flo screwed the AI
I wonder if letting the lings through at the same time as the Oracle reaches Serral's base is actually a good play. Rather than being allowed to focus first on the oracle defense, then later on focus on using the lings, he was forced to do both at the same time, losing lots of drones and nearly wasting the lings.
Early on, maybe, and it's a huge *MAYBE*, but later on it was just a straight up bad play.
@@veidro Agreed. Will be interesting to see what its walls look like in a future version, though ordinary full wall-ins are the most likely option.
That's very possible. The interesting thing about machine learning is that ONLY KNOWS WHAT WORKS, it doesn't know why. So maybe it did that a few times and got an advantage and won, and just kept doing it the same way.
@@ClayJitsu And to add to that, it might associate putting many buildings in the front with a full wall-in which is usually not at all what you want. Or making the gap too small for some units, etc.
ClayJitsu you also have to take into consideration that it took alot of things from human players, which, sometimes make absolutely no sence at all. For exmaple in some earlier games, we could see Alphastar attacking the rocks below the main ramp, which makes absolutely no sense for an AI to do. So this could have just been one of those things implemented in its plays.
I found you because of a link from different TH-cam! You do amazing video and explaining! SUBSCRIBED!
Artosis is the perfect spokesdude for SC2!! So knowledgable and articulate about the finer details while also willing to break things down for outsiders to the game. Really entertaining and informative analysis.
"you will never see this in a pro game" bet you will now lol
Depends. In a lot of ways alpha star is very impressive. In other ways it highlights just how far AI still has to go. The attempt at a wall off is a good example. You'll never see it in a pro game because it makes no sense. It's like it built a wall for wall's sake without understanding what it's for.
@@heroclix0rz I wonder if it does not wall off so the imo & disruptors can go in and out .... maybe it never realized that you can drop a pylon to wall off during the early game, then destroy and replace it. Just guessing, because it seems strange, I make better walls than alpha star, and I'm still really far from beating serral on a lucky day.
@@heroclix0rz can we be sure it makes no sense at this stage? If it consistently does it 'wrong' as is claimed, that strongly hints it's no accident. The partial wall obviously helps it in some way, otherwise it wouldn't have learnt to do it. Would be interesting in a A* vs A* matchup, to force one to create a proper wall artificially (the game played to completion for both possible outcomes), then measure the difference in outcome over multiple trials.
@@Lee-pf6od The probe micro is ridiculous, so it was no reason to correct bad walling behavior. It clearly can make walls and chokepoints properly where it expects there to be more army fighting
@@Lee-pf6od You could also specifically teach it proper wall placement, inject those agents into the process, and see if they're validated or not.
Basically, if the AI can consistently beat human players, we can feed it specific strategies and test them against itself in optimal conditions.
I loves how artosis explaining/casting the game like hes explaining to anyone who never know/played this game
Man, we came here to look SC2 vids - pro games, pro casting, all the stuff. Do we really need those kidergarden basics? I would like to listen more about macro and players mistakes rather than "this is medivac - flying unit that can carry other land units, or apparently heal them"
@AmplifiedSilence to play devils advocate I dont see the logic in using the Serral/AS game as the vehicle to educate and explain the basics of the games. Why not a random master or GM ladder match? Furthermore if someone clicked on this video with 0 prior knowledge of SC2 then they wouldnt even know what AS is in the first place. This video is important to the community and seems like an inappropriate educational vehicle. Like it or not the constant and prolonged pauses greatly reduce the entertainment of the video. Its a shame.
This level of complete explanation of everything really does kill a lot of potential enjoyment. If I watch a chess match of the best player in the world versus a computer, I don't want to listen to people explain what the pieces do. I thought it went well with the main alphastar video, but the main appeal of more alphastar work is what it comes up with that top starcraft players don't do, the appeal is not the basics
10 min game, 30 min video... -.-
I like the pauses, they were very useful for a newbie like myself.
Really Alphastar's brilliance comes from it's unit control. Like Artosis said multiple times, it doesn't really have the decision making of Serral (no wall off, recalling the oracles for 1 zergling, attacking with half its army), but what it doesn't have in strategic intelligence, it really knows how to make the most of its units (the probe micro, the warp prism juggling, its choice in unit comp going toward micro intensivity). This is why I take alphastar as less strategically brilliant AI, but more of an awareness of how to very precisely to use units. This is certainly different but I think less impressive than what AlphaGo has done. Starcraft pros are lauded for their unit control, but of course a computer will be better than them.
maybe the cost of units destroyed instead of the number of units destroyed would be a more appropriate indicator considering the differences between zerg and protoss
great commentary thanks for explaining for everyone but still providing indepth analysis
thanks alot! im just wondering what Alphastar is. Never played Starcraft unfortunately. Big thanks for describing what units can do.
So awesome you got the actual replays
Anyone else wanna see AlphaStar without any restrictions?
I wanna see Serral with any restrictions against alphastar :D ...
I mean it would be interesting but there is plenty of videos of having AI unlimited like the one where they have zerglings against tanks and avoiding splash damage. That one is nutty.
That's just too scary to think about lol
Speaking from DeepMind's experience, it has managed to do some things people have never seen before and I reckon when AlphaStar reaches maturity, Starcraft is about to get a ton more interesting.
how can u make general statements about alphastars game when its actually different agents?
At about 13:00 you point out AS overreaction but the point is that AS did not know what resources Serral had in that area. You could look at it as a probability decision. It's similar to war when you make a decision based on incomplete information.
About all those dumb "the pauses kill me" idiots - PLEASE ignore them. You're doing this in the most perfect possible fashion. Keep it up, especially WITH your pausing and pointing out all the little details!
9:00 "sending a probe to the 3rd base location" -- that's what she said.
The commentator is wrong about recalling the 2 oracles. The moment the AI did it I knew what it was for, defending against the upcoming zergling harasses.
And even though it looks bad now, the two unblocked passsages into the main (instead of one and using a unit to block the enemy) may turn out to be better for the AI in the long run considering protoss buildings can't move. YOU may think it's bad, but the AI prefers to have over the top number of workers rather than going to just the limit, probably calculating that this is better from occasional worker losses from attacks, meaning zerglings getting in may not pose that much of a problem for them. To the AI, you losing units for not enough worker kills may actually be a win for them.
Also, for the warp prisms, the AI may actually be correct. Using it as a defense means you cover a much wider area with less units, meaning your attack force may actually be larger.
This is like the sequel to the AI playing GO. Top players worldwide may see "weird" behaviour that turns out humans have been doing wrong all these years.
Best SC2 slide-show I've seen in a while! - Could use more pauses and more explaining what's going on for all of us who just randomly ended up here without any prior knowledge about what SC2 is but chose to stay.........
thank you for explaining. I'm an AlphaZero fan that just wants to see what his brother's hobby is
I enjoyed the cast but I personally would enjoy it more if there were much less pauses throughout the cast. While I did enjoy the explanation and the breakdown of the skirmish at the 3rd base ( 25:00 ? ) I found myself click-forwarding through a lot of the breakdowns.
Cannons: its all about the total DPS. So with 200 supply you can only get a limited amount of power projection. Cannons are a way of raising this number.
I love how you explain this so in depth as if a brand new player were watching! My wife 9ver hears your videos when I watch them and I've noticed even she is starting to understand the gameplay (shes never played ever....didn't know it existed till she met me)
It's nice how you take time to explain the things down to the smallest detail it makes people who havent played, starcraft or some who just dont understand the smallest things, I honestly don't care for it. Not the type of presentation i would want to watch on a regular basis. I enjoy the way lowko, comment takes games, but i see the appeal, this is the first game video i've seen by artosisTV.
My theory is that it doesnt fully wall because it is thinking ahead and leaves space for its larger units to pass.
I am terrified of the AI's ability to learn this quickly, yet your commentating abilities soothed me into dismissing our incipient, automated demise.
When will Blizzard include the AIs as the vs computer default?
It might not be possible. This algorithm probably takes up way too many computing resources. It wasn't designed with computational efficiency in mind.
@@davidliu4134 Training takes big resources but the trained network doesn't
@@seol1500 that actually makes sense. hopefully we get it one day.
@@davidliu4134 Makes me wonder if the GPUs installed on gaming rigs are completely saturated and for those that want an extra challenge could utilize local GPUs for enhanced AI matches. Of course, only some GPUs could support this and it'd have to be tested. Maybe not ultra graphics settings possibly, IDK.
Dont need to wall when you have perfect probe vs zergling micro t(-_-t)
Of course it doesn't. But it helps to keep focused
all part of his plan ;)
Coming here from AlphaStar's matches against TLO and MaNa, and hearing "you want 22 workers per base" made me smile so hard after hearing them say "you don't want more than 18 workers" in the vid from Jan 2019, I'm not sure if that means that AlphaStar really did change the human meta, but it sounds to me like people have started building more workers than they did before!
Obviously it's all speculation, as these things are a kind of black box. But I think that Artosis was referring to the total amount of workers including gas miners so that would mean still 16 is optimal. I think AlphaStar is overproducing probes so that when it expands it can transfer the workers and get its economy up faster on the expansion. I was surprised that they didn't mention this at all in the January cast. This was actually pretty common practice in the early days of SC2 and there was a lot of discussion about whether it was more or less efficient/worth doing, for some reason I think it fell out of favour, but I think it might actually be more effective. It also gives you a bit of a buffer in case you lose some workers, you don't lose any mining time while replacing them and maybe it overproduces on purpose because it expects to lose at least a few workers in any game in general. I think we can actually learn from the AI here.
In short, I don't think it is oversaturating because it is more efficient to mine with that many workers. It is oversaturating for other reasons. (worker transfer and worker loss replacement)
thanks for the explanations. I used to play like ten years ago but quit while still a newbie, now im back as a newbie all over again with all the changes. Glad SC is still around.
AlphaStar is a very difficult opponent because of its consistent performance.. It doesn't get tired or nervous, doesn't doubt itself, has no trouble with its stamina. In any situation, it plays just like how it's programmed to be!! Indeed, sc2 is about being consistent, not about showing a few godlike micros occassionally.
No human could micro his probes that good against 7 lings while you counter harass yourself and also que up an expand
No human can currently, no, but I don't think that's an example of superhuman mechanics. Just of mechanics that are theoretically possible for a human but unreasonable to expect of one. It's like having a bot with Serral micro in 2011 tournaments
@@Battleschnodder I agree but the whole concept of the study was to show the world that ai can strategically beat humans in a complex game that you have to take quick decisions in real time without relying on superior "mechanics" like the tlo and mana agents had. I do not believe that they proved that all.
@@nickoslekkas3336 i agree, it beat him with inhumanly precise micro. oracle micro, probe micro, and disruptor micro to be specific.
Yeah, Zerglings really aren’t that much more powerful than workers early on and the precision micro takes away from that already relatively slim advantage. Definitely undermines the idea that it’s competing tactically as opposed to mechanically.
I think that moment was very reasonable. You can see on Alphastar's cam that it is almost exclusively looking at its probes during that defense. The oracle had been sent and is doing its job alone. Sure that's good probe micro, but Alphastar is exclusively focused on this probe micro and any pro gamer could do the same. As for the probe sent to expand, it does it after the zerglings have been dealt with.
What's impressive is the decision making in what to micro. Most pro would have tried to switch their attention between harassing with the oracle and defending with the probes. Alphastar instead chose to trust the oracle to do its job and focus on defense. When I say most pro, that's evident with Seral himself. I've seen the same game from his perspective and at that very same moment he is switching his attention between attacking with the zerglings and defending the oracle, doing a rather poor job at both.
The choice to expand at that time is still impressive because most players in this situation would still consider themselves on the defensive.
The only thing that can defeat AlphaStar at this point is the Artosis curse.
same as alphago, alphastar is trained to fight much stronger opponents, so it's playing it "safe". the goal of the game is to win with certainty, not to smash the opponent with a high risk strategy.
11min game turned into a 30min video."play 1 sec, pause 3sec"
Amazing cast! Thank you!
"this is some human-like behavior" just as AI builds an nexus at precisely 400 minerals and still microing probe to delay mining.
When I see the recall of the oracle by alphastar, I really feel that we got a big move to close to the skynet.
At 8:07 serral runs lings up the ramp, and the two adepts seem to give up and go back to guarding the front, only to turn around again once the probes are being attacked. I wonder if serral could have run the lings off to the side and morphed some banes. A human would stop it, but alphastar might be like "oh he left, problem solved".
I don't know, it seems impossible but having someone from deepmind or an AI pro to discuss how alphastar learned (specifically, how it managed to learn these standard ideas) would really help to show off how amazing this achievement is.
The problem with machine learning from what I know (and I'm definitely not an expert in any shape or form) is that you can't really say how it got to the conclusions it got to
the AI cant talk to you (yet)
They must have told it some stuff about how to play. There's no way human strategies are so optimal that an advanced machine just happened to come up with the same ideas through random experimenting. If there was, then why does the meta change so much?
@@ClayJitsu I was really surprised by that aswell, but the human copying might have come from playing vs humans and losing from them, the only thing it could do is learn from the players, and as artosis says, it does alot of weird stuff
@@ClayJitsu deepmind can play unlimited amounts of games at the same time so the learning process isnt very human like.
im three minutes in wondering how artosis is really going to keep up with a Serral game while also providing the basics commentaries like explaining chronoboost :o
In general, it seems like Alpha uses a wider array (compared to the pros) of the tools the game offers it. In short, a more complex approach than the brute force most pros applies. And what if Alpha walls poorly on purpose, to let more units in, and thus kill more?
Anyway, I don't know squat about SC, so I could be completely wrong.
We all know that without perfect probe micro, Alphastar would have lost. It is very apparent that Alphastart agent learned to imitate the placement of humans, but did not learn that those building placements are to create a full wall.
It wasn't imitating humans at all. Alphastar became so good by playing against itself millions and millions of times, and came to the conclusion that walls are a good thing to do.
It literally learned nothing from humans.
I do think the reason it doesn't full wall though is because it doesn't need to. It probably realizes that if Serral wants to sacrifice 20 lings to kill 3 workers, that's a good trade when he isn't going to push out anyways and has no other way of doing damage.
imo the lack of full wall off is a bait for people to make mistakes running past.
@@Figgy20000 I think you do not know how alphastar is trained. It is true that is learned a lot by playing against itself. But with such a huge search dimension, it is often very hard for these agents to find any solution close to optimal.
Instead they are often initialized or "bootstrapped" with a policy. There are few methods for bootstrapping but one of the more complicated and successfull one is "imitation learning". So alphastar agents are all bootstrapped with imitation learning, from player replays. Then they play for years between themselvesi to refine their behavior.
A similar thing was done in their Go agent, AlphaGo, which also first observed human play and then went to refine itself. Afterwards, they also created another agent called AlphaGoZero, which learned without any bootstrapping this time. They have no such agent for sc2 yet, nor do I expect anything similar in the near future.
Alphastar is literally walling of because it watched people wall off, but did not know why it was. It just knows that placing buildings in that area helps increase the winrate.
As for the "it does not need to" argument, it definitely needs to wall off. An agent not walling of vs an agent that walls of will always favour the one that walls of. Alphastart has limited APM, the more it wastes APM to the perfect probe micro, the less it has for other tasks.
At 5:34 game time 12:33 video time there was recall of oracles, stalkers warp-in and it was commented as a waste, but 30 seconds later there was an attack at the 3rd base, so may be it's just a safe play when AlphaStar saw an opportunity for Serral?
Played a lot of Brood War back in the day, so all the new stuff was a little intimidating, but you did such a good job at filling in the gaps!
we really would like to watch a version of it without pauses and comments
I am very much a novice regarding Starcraft, AI and everything else connected to this video. I just read about and got interested in DeepMind's AI a few days ago. But watching this match (with very insightful commentary) and knowing it's an AI playing a human is completely captivating. Thank you so much for putting up this video!
isn't there alphastar vs serral game without pausing? it is really annoying
it really is
The pausing gets quite frustrating if you just want to watch the game with some commentary
hammerguy
But if you want a thorough analysis the pausing is necessary.
nice explanations
I havnet played sc2 for quite a while so those bits of extra explanations definitely helps me
thanks, you are doing very calm smooth replays ... can watch it good with a confident understanding on what the players reason is to loose against the meta-bot
How was AS able to move its units to 3rd expansion at 21:00 and build photo cannons without having any vision on Serral's units?
Is that destructive rock within AS vision? So AS saw the rocks hp decreasing and hence relocated its units for battle?
Nice analysis of the game in particular AlphaStar's moves and (possibly) motivations
It’s crazy to think this same ai is now being used to create swarms of drones to be used in actual warfare. It’s possible it used some of the same learning it gained here into the real world
I'm wondering,for eq for the probe harrass and scouting does alpha team implements "automated scenarios" in the ia program or did alphastar realized itself it is cost effective to do so like an human would? especially the lil probe mineral harrass in the very beginning
I bet they do!
There's no automated scenarios or so, it's a self learning system, so it's just doing what it picked up itself.
@@Quickshot0 i doubt about that
@@stkfr They release science papers on how they've made the various alpha series AIs like Alphastar, it's always been via self learning or guided self learning. But in either case it's ultimately learning how to play itself.
It's in that sense very different from old style AIs that have pre-programmed reactions.
ahah! a master commentator with the most recent Alphastar content in hand. Much excite.
Is AlphaStar playing with a human limitation of not able to click on several objects at once and of clicking with human speed?
AlphaStar has a virtual mouse and keyboard, and is limited to using them at human speeds. Its mouse cursor cannot teleport from one part of the screen to the other, it has to move its mouse.
However AlphaStar will never misclick, misaim, mistime, or misjudge a distance. It always issues exactly the order it intends to, in exactly the way it intends to, and this lets it perform ungodly micro.
When Serral saw that gap in the computer program's shitty wall... he should have noticed that it was a bug within the code, and he should have canceled his third base to mass lings from that point on, simply flooding the protoss base with them.
Then, game over...
The oracles can't "out DPS" them with limited energy, and the protoss had ZERO ground forces at the time the wall was exposed.
i would like to see Alphastar vs Alphstar game !!!
You can see custom AI playing each other at Probots.
th-cam.com/video/Eh9sH_L17FY/w-d-xo.html
@@simosavela2802 yes but i would like to see alphastar vs alphastar games
I just want to watch the game without pausing it. Can't find one. :(
Alphastar sent only one zealot back to deal with 4 roaches because it estimated that the zealot will buy enough time to be able to warp in at home. Alphastar did just that while attacking Serral's base . That's incredible.
Thanks for the cast. Serral wasn't using his own keyboard or mouse (from what i heard). Really wish there would have been a proper tournament to see how good alpha star really was.
excellent commentary, perfectly suited for the newer viewes to get an explination of stategy and tactics on sc2, but also as fast pasted as a normal pro game for me to not be so bored but that. well done Artosis
Who is a new viewer tho
@@JacobSimonsen people who are interested in deepmind but don't know much about starcraft. it doesn't really take a genius to make that connection you know.
@@darkphoenix00001 so you take a 11 min game and make a 33 min video
@@JacobSimonsen p much yeah
Id argue that the deepmind AI ppl go watch other educational videos or play the free game yourself. In terms of president this video is extremely important to the community and it is a real shame it has been used as an educational video. This is like the superbowl or olympics so no it doesnt seem appropriate to pause the superbowl and explain what the lines on the grass mean or what shape the ball is.
i don't play starcraft and am solely interested in the AI, but you commentating and explaining what the fuck is going on is great appreciated thank you.
Artosis, dont' know about other games but in this game don't think that the hole in the wall is a mistake but intentional. I believe there is a reason for that. Probably the AI figured out that if it lets some zerglings inside his base his opponent will micro them and this would allow him to make a lot more kills than the zerlings. Even if the opponent has the same micro skills as him probably the AI saw that if there are zerlings in his base and both sides are microing to attack and defend workers at the same time the oracle will make more kills than the zerlings. When playing against human it is especially evident that even a little distraction to control the zerglings led to many kills of drones and less of probes. You can see that the oracle returned to attack the drones when the AI saw a lot of microing of zerglings in his base. This was not coincidence.
I'm starting to think that alphastar purposely has the hole in the wall-off so that Serral has to micro the lings AND defend against the oracle. It knows it can deal with the lings efficiently and thought it's worth the sac to take some APM from Serral.
Well to be fair, a lot of pro decisions are based on mitigating errors and units like disruptors are very hard to use well so unless you have supreme confidence in your mechanics you're probably better off focusing on immortals.
You mean like the confidence of something that is able to command three units to go three different directions at the same time?
Great commentary. To the people complaining about pauses and explanations, why are you even here for? Lol. There're other vids of this match elsewhere. Sheesh.
At 13:30 couldn't alphastart just take 2 of those 5 adepts and attack the third base from the south? since 2 adept can kill one drone why attack one drone with 5 adepts. doesn't make sense...
Redundancy. Spread out adepts would die exponentially faster than 5 adepts, due to more zerglings being able to attack at once, fewer targets to focus on, and of course the extra health pool. Also, if 1 of the 2 adepts dies, it's basically impossible to get any more damage, whereas the pack of adepts is less affected by a couple of them dying: still able to secure kills.
Thank you for the newbie explanations. I love the SC2 videos!
Great explanations with amazing breakdown of all possible outcomes in every situation!