On March 13 2016 17:38 killerdog wrote: How does the endgame normally work in GO? Do you just stop playing once neither player wants to place anymore? or do they fill the board or what
Game ends when a player resigns, or both players pass on their turn.
supposedly alpha go never goes to the end, because it'll already have the count before it gets there.
Honestly, not sure why it hasn't resigned already. There shouldn't be a 10% chance of it winning right now. I can only assume it doesn't realize the state of the groups on the left side.
Ah as I was writing that up, looks like it finally resigned.
On March 13 2016 17:38 killerdog wrote: How does the endgame normally work in GO? Do you just stop playing once neither player wants to place anymore? or do they fill the board or what
Game ends when a player resigns, or both players pass on their turn.
supposedly alpha go never goes to the end, because it'll already have the count before it gets there.
Honestly, not sure why it hasn't resigned already. There shouldn't be a 10% chance of it winning right now. I can only assume it doesn't realize the state of the groups on the left side.
Ah as I was writing that up, looks like it finally resigned.
Apparently it's definition of 10% is something like when 10% of the lines it's looked at win, which includes the ones where the opponent was just blind.
Hence why it played a ton of forcing moves in a row hoping Sedol would go full retard, because presumably in some low % of it's calucaltions, it checked variants where the forcing move was ignored.
On March 13 2016 17:48 TheEmulator wrote: Also lol at the 322 accusations. Really people?
I mean they already had accusations that Lee Sedol was paid off by Google to lose the first two matches because he didn't attempt to force kos. It's pretty stupid since you can tell by body language/reaction afterwards that he has been trying pretty hard throughout the event. Even with today's win, I don't think he is very happy due to how it happened. AlphaGo went from God to Amateur in one game. It seems to be weak with capturing races + kos. Tilted AlphaGo is very different from what it has shown in the first three games. Hopefully we see a very good game next match as the pressure to "win 1 game at least" is off lee sedol now.
On March 13 2016 17:48 TheEmulator wrote: Also lol at the 322 accusations. Really people?
I mean they already had accusations that Lee Sedol was paid off by Google to lose the first two matches because he didn't attempt to force kos. It's pretty stupid since you can tell by body language/reaction afterwards that he has been trying pretty hard throughout the event. Even with today's win, I don't think he is very happy due to how it happened. AlphaGo went from God to Amateur in one game. It seems to be weak with capturing races + kos. Tilted AlphaGo is very different from what it has shown in the first three games. Hopefully we see a very good game next match as the pressure to "win 1 game at least" is off lee sedol now.
The "tilted" part just seems to be the incredibly mechanical way in which it approaches positions. If it's at all ahead, it plays directly after the safest route to victory. If it's at all behind, the way it evaluates "win potential" of variations is so inhuman that it just looks bizarre.
The interesting part will be what caused it to think it was losing, which led to the "amateur" moves. The AGA stream mentioend something about them thinking move 79 or 81 or something looked odd, and then the alphago team tweeted that the AI got confused/"realised" it was losing around move 87, so maybe it's move at 81 was based of a variation which it hadn't calculated to enough depth?
Either way it's reaction to suddenly not winning is a pretty fascinating example of the differences between a human and an algorithm.
On March 13 2016 17:48 TheEmulator wrote: Also lol at the 322 accusations. Really people?
I mean they already had accusations that Lee Sedol was paid off by Google to lose the first two matches because he didn't attempt to force kos. It's pretty stupid since you can tell by body language/reaction afterwards that he has been trying pretty hard throughout the event. Even with today's win, I don't think he is very happy due to how it happened. AlphaGo went from God to Amateur in one game. It seems to be weak with capturing races + kos. Tilted AlphaGo is very different from what it has shown in the first three games. Hopefully we see a very good game next match as the pressure to "win 1 game at least" is off lee sedol now.
The "tilted" part just seems to be the incredibly mechanical way in which it approaches positions. If it's at all ahead, it plays directly after the safest route to victory. If it's at all behind, the way it evaluates "win potential" of variations is so inhuman that it just looks bizarre.
The interesting part will be what caused it to think it was losing, which led to the "amateur" moves. The AGA stream mentioend something about them thinking move 81 looked odd, and then the alphago team tweeted that the AI got confused around move 87, so maybe it's move at 81 was based of a variation which it hadn't calculated to enough depth?
Either way it's reaction to suddenly not winning is a pretty fascinating example of the differences between a human and an algorithm.
It misplayed move 79 for some reason, and didn't realize the mistake until move 87. Kim Myungwan in the AGA stream looked at a lot of the variations/responses there, and was able to find the correct move. I assume it's due to the complicated ko and fighting in that situation that it fell apart. Possibly it wanted to avoid the ko entirely and so it chose the incorrect move since it wanted to play solidly. The "tilt" is actually attributed to how the monte carlo engine works. You'll see it with normal bots that are currently out there for Go as well.
I'm pretty interested in how it made that move 79 mistake. I think it's the first time you can call it making a real mistake like that.
Well, I'll post my full thoughts later, but both endings to Life is Strange were pretty astoundingly unsatisfying. I didn't feel like they resolved anything at all, and they really didn't even feel like a culmination of the game up to that point. Pretty disappointed with the ending at least. On the bright side, if you can get disappointed by an ending, then there's a decent game leading up to it that makes you hope for a good ending.
On March 13 2016 19:35 GhandiEAGLE wrote: Well, I'll post my full thoughts later, but both endings to Life is Strange were pretty astoundingly unsatisfying. I didn't feel like they resolved anything at all, and they really didn't even feel like a culmination of the game up to that point. Pretty disappointed with the ending at least. On the bright side, if you can get disappointed by an ending, then there's a decent game leading up to it that makes you hope for a good ending.
Last ~hour of the game with the saving people on the streets and dream sequence reflects on how you played the game and what choices your character made, then there's a conversation with yourself where you have to think about what your motivations really are and finally you have a decision to make - do you make it so that your last week never happened outside of your head, but at least you got your one last week with your bff/bae, or do you stick with the choices you made for whatever reason (perhaps because you think everything you lived through would lose its meaning if you just reverted it). On one hand saving the bay and moving on fits thematically with the game, but the other one feels like going against the grain, to give a big fuck you to everything and actually make a decision that you want.
I also played this game very recently and I'm quite in love with it.
On March 13 2016 15:13 zer0das wrote: Ugh, my friend is making me play Dawn of War II last stand with him. My biggest complaint about DoW II (and the original to a lesser extent) is its hellishly difficult to tell how much damage anything is dealing to anything else. It's even worse in last stand, Necrons never die to anything and its unclear why other than their health regen is waaaay too high.
It's like the blueprint for not giving the user enough feedback. You have a Necron hero sitting there tanking like 100+ units and taking no damage. It's ridiculous, and there's way too many instances of this happening. You can't even check enemy units to the best of my knowledge, not their health, certainly not their current abilities. Like why would you ever need to know that information?
I mentioned this last night playing DoW with doge and sunaj - it's hard to tell if you're winning fights. I haven't done any real digging to see if there's an option for it, but something for easier feedback on enemy troop health. I know I can hover my mouse over units to see their health, but that's fairly clunky and hard to do in the middle of skirmishes
On March 13 2016 19:35 GhandiEAGLE wrote: Well, I'll post my full thoughts later, but both endings to Life is Strange were pretty astoundingly unsatisfying. I didn't feel like they resolved anything at all, and they really didn't even feel like a culmination of the game up to that point. Pretty disappointed with the ending at least. On the bright side, if you can get disappointed by an ending, then there's a decent game leading up to it that makes you hope for a good ending.
Last ~hour of the game with the saving people on the streets and dream sequence reflects on how you played the game and what choices your character made, then there's a conversation with yourself where you have to think about what your motivations really are and finally you have a decision to make - do you make it so that your last week never happened outside of your head, but at least you got your one last week with your bff/bae, or do you stick with the choices you made for whatever reason (perhaps because you think everything you lived through would lose its meaning if you just reverted it). On one hand saving the bay and moving on fits thematically with the game, but the other one feels like going against the grain, to give a big fuck you to everything and actually make a decision that you want.
I also played this game very recently and I'm quite in love with it.
From a practical standpoint, the game isn't very good or well thought-out. From the standpoint of its overarching story, I think the game is very well thought-out and uses effective, if sometimes hamfisted, symbolism and actually conveys a cohesive message, which is something a lot of games fail at.