I like comparing his level to slightly lower than ShowTime’s.
Please do not try to kill this delusion.
Let atleast this one live 
Are you speaking about Mana? If NA has Neeb and Astrea that are not far apart in skill, in EU the differnce of Showtime with the second and third player (protoss) is far bigger.
My opinion.
Who the heck is mana? Oh, that guy who couldn’t beat that derpy AI?
Dude i would like to see you, it plays entirely different. You don’t know what to expect. It wasn’t APM capped and could beat mass immortal with 1500 apm perfect blink micro from 3 sides at the same time… It was literal conflip.
Yet than he devised plan how to beat it in a live match and make it retreat back into main base, each time prism threaten to harass it… If he played against it on daily basis, he wouldn’t have problem to beat it at decent win rate…
Also alphastar wasn’t that derpy, it even put couple units back to defend oracle or something sometimes and protoss ai was decent, it was scouting even and playing reactively. Sometimes still derpy, but that because there is 10 to power of 26 moves, it didn’t adapted yet, they could run it longer, but it cost money, it was done for research not to make perfect sc2 ai…
Mana isn’t that good. I used to root for him, but he always lets his fans down whenever he gets even close to a title.
Like you said, the way he always loses to zerg. Against terrans it seems like all he does is try to micro blink stalkers for 20 mins then gg’s out when he throws the game.
1 Like
Mana can not beat ai but is gm? Optoss at work.
What a loudmouth! That was not the vanilla Blizzard AI but something quite different. And by the way, that AI defeated other Pro-players and with some tweaking can win Code-S hands-down.
1 Like
Lose to ai is worse than bronze! But is gm with protoss. It speaks all to say of balance.
BabaYaga,
As far as I remember MaNa was the only person from the showmatches, that managed to beat Alpha Star.
Please ignore this pathetic person.
I don’t even want to see quotes with his text 
#Bias moment
#TeamMaNa
#Hack trolls and burn their computers
4 Likes
Some people are too dense to understand that the main AlphaStar purpose was not to beat Pro-Players.
That could easily achieved by giving AI 100 000 APM and letting Alphastar to attack from 5 directions while doing a quadruple drop microing each unit individually.
People don’t understand that we use separate control-groups because it is not humanely-possible to micro individually units when your army has 50 elements.
Simply by brute-force (virtually unlimited multitasking) Alphastar without constrains could 101:0 Serral and his uncle…
4 Likes
Wasn’t TLO offracing as protoss the one who lost on the showmatch?, later I think other players lost to it but they started tracking the AI and countering the style and the AI games ended because the developers said that they wanted people to play on the same way as against humans because the objective wasn’t creating an AI that beat humans and more an AI that could learn and they had enough games as sample.
2 Likes
Touche!
+1000000000000000000000000
Yeah that’s a problem with playstyle. If you aren’t sure what your opponent is doing or is going to do, the best thing you can do is to be aggressive. That’s why I 12 pool every time I play a random player.
Opponent is playing defensive → he has to invest in defense → you come out equal in eco but have map control.
Opponent is playing greedy → you flat out win.
Opponent is playing aggressive → you are equal.
1 Like
That was literally the point of their AI. From the opening paragraph of the AlphaStar paper:
Many real-world applications require artificial agents to compete and coordinate with other agents in complex environments. As a stepping stone to this goal, the domain of StarCraft has emerged as an important challenge for artificial intelligence research, owing to its iconic and enduring status among the most difficult professional esports and its relevance to the real world in terms of its raw complexity and multi-agent challenges. Over the course of a decade and numerous competitions1–3, the strongest agents have simplified important aspects of the game, utilized superhuman capabilities, or employed hand-crafted sub-systems4. Despite these advantages, no previous agent has come close to matching the overall skill of top StarCraft players. We chose to address the challenge of StarCraft using generalpurpose learning methods that are in principle applicable to other complex domains: a multi-agent reinforcement learning algorithm that uses data from both human and agent games within a diverse league of continually adapting strategies and counter-strategies, each represented by deep neural networks5,6. We evaluated our agent, AlphaStar, in the full game of StarCraft II, through a series of online games against human players. AlphaStar was rated at Grandmaster level for all three StarCraft races and above 99.8% of officially ranked human players
LOL. and what would happen if Alphastar was unleashed with 950 000APM?
Really?
They matched the APM to human levels, not just in the average but in the whole distribution of APM. They also matched the reaction time in a similar manner.
True, so they inserted artificial constrains to the AI. Which proves that wining was not the first imperative. Simply they wanted to see how AI would fare on Ladder with human APM and human multitasking.
Alphastar makes the AI have human-like limitations → PPP brain → AlPhA StAr ImPoSeD ArTiFIcIaL ReStrIcTiOnS ThE Ai ShOuLd Be MoRe ReaLiStIc By LETtTinG It HaVe 12721309123923109312094292319 ApM!
AlphStar isn’t biased. It was given human restrictions and with those restrictions it excelled with Protoss the most, then Terran and the least with Zerg. It’s the same as Grandmaster league which is P>T>Z in both win-rates and representation. When the stars align to say Protoss is OP, Protoss is OP.
It’s a well known fact, to anyone who has played the game and is honest, that Protoss is much easier to play. In a game which stresses multitasking, being easier has a HUGE impact on the strength of the race. The micro potential of Protoss just isn’t there, for example. What are you supposed to do with zealots except amove them! Nothing is what!
1 Like
I’m not following your argument. You are complaining that protoss players are saying protoss can’t win a GSL or major tournament. Which they haven’t, in many years.
But you are simply doing the same to try to prove that protoss is good. Yet you are using GM league, rather than GSL. You’re still cherry picking the GM league the way you say GSL’s wins are getting cherrypicked on the other side.
If anything, your argument makes less since. Because as a whole, GSL is way more skill than GM. GSL is the highest of the GM league. And as we all see, Protoss just plain cannot win any GSL’s.
1 Like
You’re saying that GML and premier tournaments are analogous and therefore deserve equal treatment, but they aren’t analogous. The situations are not only different, but the differences are massive and demand different treatment.
They’ve won super tournaments, which are a GSL event.
Cherry picking is looking at a small subset of the total data and ignoring the broader trend in the broader data. Example: Global warming is false because it’s cold in Antarctica. What I am doing is quite literally the exact opposite of cherry picking. I am saying “Stop looking at the temperature in Antarctica and instead look at the global averages”.
It’s not only justified to disregard small sample sets, statistics demands that we do. Imagine flipping a coin once. It gets heads. Is this strange or not? You can’t tell. Imagine flipping a coin 5 times and getting heads 3 times. Is this strange? You still can’t tell. Now imagine flipping a coin 10,000 times and getting heads 7,500 times (aka 75%). Is this strange? Exceptionally.
Why, though? It’s because if we assume that tails & heads are equally likely, the odds of flipping 75% heads 10,000 times is basically 0%. It can happen, but in extremely rare circumstances that are so rare that it’s basically 0% of all scenarios. We can therefore conclude that our assumption (that tails/heads are equally likely to occur) MUST be false.
The same question is at play here. If we have a 3 sided die (T, P, Z) and it rolls Zerg 14 times of 34, is that unusual? Nope, because the odds of that occurring under our assumption (that Z=T=P) is about 21%.
Now, let’s calculate the odds of Grandmaster being 45% Protoss under the same assumption. There are roughly 600 GMs with about 270 Protoss. The odds of this occurring under our assumption: 0.0001%. That is 1 to 1,000,000 odds. That means under our assumption, this scenario would occur in 1 out of every 1,000,000 scenarios. It’s too unlikely, so we must reject our assumption.
So, as you can see, statistics demands that we reject small samples, because they are too easily biased by chance (and other non-chance variables which we may not necessarily be interested in measuring, such as a player’s skill).
1 Like