I just wonder why giving protoss stargate even more "specialist" will make them more of an option.
So tempest isn't going to be a fighting unit, but instead more of a unit, that forces the opponent to attack in stand-offs. So oracle does some harrass, that I'm still not sure, if it really pays off compared to 8 marines stimming into your probeline. So void ray is somewhat obsolete then, since we have oracle? And phoenix is reduced to situational anti-muta duties, because if harrass of the oracle isn't better, they can scrap that unit.
I, as a mainly esport spectator, can really relate to Browers words, when he said: I want races to have different options, not just one way you are forced to go. I welcome the new terran units, as they give these options. The same is being done to zerg (new hydras/mutas/roaches will be a choice, viper/infestor also). But protoss will (perhaps even more) be confined into build a death ball, that will always need the colossus, since there is no other high-dmg unit available. Without colossus or real expert HT use, you can not win. This is basically the situation we are in now. The new units will change nothing in terms of protoss fighting force, which is disappointing.
On June 15 2012 06:03 KonohaFlash wrote: Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
This is actually not true at all. Modern AI is disgustingly good at micro. Where it fails is in strategic depth to direct its macro.
Look up, for example, the Berkeley Overmind. Muta micro that makes Jaedong look like a scrub. But that's ALL it does (plus some solid overlord positioning AI), plus expand, and build sunken colonies.
On June 17 2012 20:56 testthewest wrote: Without colossus or real expert HT use, you can not win. This is basically the situation we are in now. The new units will change nothing in terms of protoss fighting force, which is disappointing.
Very correct - I had high hopes for the tempest to bring back a true skytoss option. Man would that have been awsome. But as it is, it seems like it's more of a unit to break, say, zergs with max broodlord/infestor.
Protoss seems to be really limited even in theory. Before people come and claim "wait and see" - well, wait for what? That miraculously the oracle gets awsome dps? Each and everything protoss gets is "cute". But to do damage we will still rely heavily on the colossus which just sucks immensely.
Goes into detail about the reasoning behind no lan, as well as some features the new resume from replay feature has in store, as well as talks as to when to expect to see the HotS beta. Apparently if I heard correctly, you can resume the game 10 seconds before a disconnect or crash, as well as allow the A.I to take control of the other player in the replay and imitate their actions.
This could allow you to have someone perform a rush, and then allow the A.I to play it out again from a replay so you can practice against defending the rush. Sounds very cool and exciting! Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
I think bio will still be very viable, especially in combination with spider mines. None of the new protoss units are actually good vs MMMGV I think, colossus will be required to take mines out without losing units yourself.
Plus the chargelot/archon remax can't deal with mines other than sacrificing units, wich takes alot of micro to pull off in the middle of the battle when you want to focus on splitting HT, getting good storms off, warping in new units etc. If one mine gets a good hit you're probably going to lose the battle.
On June 15 2012 06:03 KonohaFlash wrote: Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
This is actually not true at all. Modern AI is disgustingly good at micro. Where it fails is in strategic depth to direct its macro.
Look up, for example, the Berkeley Overmind. Muta micro that makes Jaedong look like a scrub. But that's ALL it does (plus some solid overlord positioning AI), plus expand, and build sunken colonies.
I think that it could be programmed to be significantly better though. Maybe people are lazy, maybe they don't know how, but the technology is there. For instance scouting and being able to guess from that what the opponent is doing, giving each alternative a chance (Say DT rush 50% Blink stalkers 35% chargelots 13% citadel fake into something else 2%) upon scouting citadel, and then making it figure out the best possible response to that, etc.
I really think that it'd be possible to program a good AI but it'd be very difficult and it'd require the programmer to have extensive knowledge about pro level play as well
Ya first you hear excuses for the $$thing@lan then you get trash talk about a feature that the community "made" and now they are talking like they´ve invented that..(this is one thing that really piss me off btw e: still better to implent it then hunt it) it´s all about selling fridges to eskimos
On June 15 2012 06:03 KonohaFlash wrote: Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
This is actually not true at all. Modern AI is disgustingly good at micro. Where it fails is in strategic depth to direct its macro.
Look up, for example, the Berkeley Overmind. Muta micro that makes Jaedong look like a scrub. But that's ALL it does (plus some solid overlord positioning AI), plus expand, and build sunken colonies.
I think that it could be programmed to be significantly better though. Maybe people are lazy, maybe they don't know how, but the technology is there. For instance scouting and being able to guess from that what the opponent is doing, giving each alternative a chance (Say DT rush 50% Blink stalkers 35% chargelots 13% citadel fake into something else 2%) upon scouting citadel, and then making it figure out the best possible response to that, etc.
I really think that it'd be possible to program a good AI but it'd be very difficult and it'd require the programmer to have extensive knowledge about pro level play as well
There's definitely room for improvement, but it's also a very computationally difficult problem. Also, the Berkeley Overmind was trained against a former pro player. You don't want to program that kind of knowledge in - Finite State Machine based AIs are quite bad or horribly complex and buggy (think original Starcraft levels of oh-exploitable AI). Evolutionary algorithms and actual game play are the way to train the AI with game knowledge. That way it can evolve against the meta-game (or whatever version of a metagame it is exposed to). The micro is the part that can be programmed in.
I really liked the Chris Sigaty interview, although I can't fully understand Blizzard's decisions I'm glad that they really have been listening to the community.
On June 15 2012 06:03 KonohaFlash wrote: Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
This is actually not true at all. Modern AI is disgustingly good at micro. Where it fails is in strategic depth to direct its macro.
Look up, for example, the Berkeley Overmind. Muta micro that makes Jaedong look like a scrub. But that's ALL it does (plus some solid overlord positioning AI), plus expand, and build sunken colonies.
I think that it could be programmed to be significantly better though. Maybe people are lazy, maybe they don't know how, but the technology is there. For instance scouting and being able to guess from that what the opponent is doing, giving each alternative a chance (Say DT rush 50% Blink stalkers 35% chargelots 13% citadel fake into something else 2%) upon scouting citadel, and then making it figure out the best possible response to that, etc.
I really think that it'd be possible to program a good AI but it'd be very difficult and it'd require the programmer to have extensive knowledge about pro level play as well
There's definitely room for improvement, but it's also a very computationally difficult problem. Also, the Berkeley Overmind was trained against a former pro player. You don't want to program that kind of knowledge in - Finite State Machine based AIs are quite bad or horribly complex and buggy (think original Starcraft levels of oh-exploitable AI). Evolutionary algorithms and actual game play are the way to train the AI with game knowledge. That way it can evolve against the meta-game (or whatever version of a metagame it is exposed to). The micro is the part that can be programmed in.
It's funny, I've just been thinking about AI... + Show Spoiler +
I took 2 months off SC2 to focus on getting my grades up. Started playing while on vacation/visiting family and don't have the hardware/network to play seriously so all I can do is play vs single player AI.
For all the work Blizz has done regarding balancing, design and single player missions they've really forgotten about PvE. The AI could be so much better (GTAI comes to mind) but instead of the AI being updated with the ladder/pro metagame or doing any sort of expanding or micro (with apm caps for different difficulty) all we have is t1-2 pushes and mineral cheating.
On June 15 2012 06:03 KonohaFlash wrote: Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
This is actually not true at all. Modern AI is disgustingly good at micro. Where it fails is in strategic depth to direct its macro.
Look up, for example, the Berkeley Overmind. Muta micro that makes Jaedong look like a scrub. But that's ALL it does (plus some solid overlord positioning AI), plus expand, and build sunken colonies.
I'm just repeating what Chris Sigaty himself said. I think he knows better than anyone else.
On June 16 2012 07:41 Arush wrote: widow mines on air units is a bit over the top imo especially when you see mutalisk play disapear vs terran :S IMO make is friendly fire and only ground.
Funnily enough, I can see Mutas becoming MORE popular with the advent of the widow mine. More widow mines means fewer/later Thors, and a good zerg should be able to mitigate widow mine splash to kill only a single muta. Now, if widow mine splash effects allied units... Muta harass would become the GO TO strat vs mech Terran.
On June 15 2012 06:06 0neder wrote: .....does he know spider mines were free in brood war? try balancing them another way, or better yet make the hellion faster with moving shot and give it to that unit, or the reaper.
Keep in mind Spider Mines never attacked air, nor had a detonation time of 10 seconds.
I don't really like the idea of giving it to the Reaper because it doesn't go well with Mech, and I definitely don't like the idea of making the Hellion a Wannabe Vulture. If any unit should have this as an ability, it should be a new unit that regenerates it's stock overtime.
I had the neat idea of giving the ability to the Viking while it is in assault/landed mode. It would vary the metagame a lot and unlock an earlier tech switch timing for a meching Terran. Think of the possibilities.
I hope they do something with the Reaper cause with taking building damage and speed away and also nerfing their damage overall for combat drugs , -5 seconds buildtime and 10 HP will make Reapers completely useless even worse than they are now.
They probably should give them combat drugs passively from the start and leave speed and the current damage in. Its not like Reapers will actually kill something noticeable anyway . The building damage thats debateable i guess even though in return they should get an overall buff in damage against anything NOT a nerf .
On June 15 2012 06:03 KonohaFlash wrote: This could allow you to have someone perform a rush, and then allow the A.I to play it out again from a replay so you can practice against defending the rush. Sounds very cool and exciting! Obviously this will not be 1:1 as there is no way A.I can say micro like MKP, but this sounds like it has a lot of potential. I guess Blizzard does have the technology after all
Sorry to rain on your parade here bro but the A.I. is very well capable of making MKP's micro look like bronze league
David Kim goes over the reasoning behind the design decisions for the new units for heart of the swarm. He also discusses how Terran is being changed, so they can sit back and macro a powerful army via Mech, like how Zerg and Protoss players currently are in WoL. David also discusses how Global Rank and ladder will play out in HotS
There is no global rank and ladder. So cruel to get my hopes up, and then to read the interview which is about being able to play on other regions, and nothing to do with a global rank.
EPT
