I don't understand why Sevastopol isn't one of the best caps in the game.. Its in the +8 and allows you to easily break Russia, so it effectively blocks the two biggest bonuses on the map. Also anyone seeing you cap there will run for a mile because they'll instantly know you're a crazy person
Look Pete. I love your stuff, you got me into Risk, but let's stop the Mr Brightside slander okay? That's a UK classic (I'm aware they're not from UK), we've adopted it as a country
Maybe it's a generational thing or something, but Mr Brightside is just seen as cliche, cringe, overplayed and an absolute vibe killer for 30 and unders, I'm from the UK too lol
@@joeh5315 In this context cringe is merely a shortened version of cringeworthy, although it is still in its neologism phase so I get why you were confused
All the music talk in this reminded me of a thought I had before but always forgot to share. Would you be able to add an obs plug-in that displays the currently playing song somewhere on the screen? Would be great for twitch and would be nice for yt as well, adding some context to the dancing Pete head.
That’s a good idea. When Pete is seriously head banging to silence whilst miming, I always thought he’s listening to something heavier, like AC/DC. Nope, turns out he’s listening to Madonna 😂
As far as a machine learning model for making better AI... that's a pretty good idea, assuming they have a database of every game ever played. AND hire some expert players to say what types of moves in what positions would be considered a good/bad move. Since diplomacy is part of the game, the AI would need to behave like a player if they want it to work better. Noob slams and breaks are common noob tactics. But not every noob plays like that, and not every GM plays diplomatically (as seen by Pete's tactical attrition/hamstringing against players in many vids/streams). The problem is that it is SMG, and they don't have a great history of making AI (like, you know, for cheat detection).
Machine learning would be the best path forward for a decent RISK bot, wouldn't be very hard for SMG to train either. I'm a ML engineer that does solo contract work, think I could probably get a solid prototype done in less than a week with the right dataset - definitely better than an expert bot lol. Something that I could even implement without too much difficulty is a bot that learns the play style of a player from their early game, and if they bot out continues to play using their style/ability level. Is it a priority investment for a broken game though? Definitely not.
Hey Pete had a game that went for about 3.5hrs on the meta settings with true random dice and my opponent bot after 2hrs in the other opponent started to attack me we did multiple cap switch’s and had lots of pockets I had to grind him down but at my own risk also died a few time is there a good way to beat someone have enough to beat the bot that had 2000 troops more than me was it a lost battle or something else would love your advice
machine learning eh, i actually wanted to train a model using graph neural network concept (to learn enemy troops position and their relationship with the map and players), with transformers for learning the game state (to capture the time variable, not sure if its that useful, but its a normally an important complexity) would do this if i had the time :(
Machine learning might be their best option. It learns from data, which they have from people playing. In effect they will finally start to learn from the peteman by proxy.
I believe that « capitals » is a very misleading name for this mode. Indeed, in many games where you see the notion of capital, it is something you want to protect and therefore bury into your territory, as far as possible from the « front line ». Here SMG gave caps one main advantage, which you can only use if opponents roll it. Which is why capitals in Risk are used very differently from other strat games…
I love the first turn because not only did you break a rule when picking you cap, you tried to get another cap that broke two rules lol
I don't understand why Sevastopol isn't one of the best caps in the game.. Its in the +8 and allows you to easily break Russia, so it effectively blocks the two biggest bonuses on the map. Also anyone seeing you cap there will run for a mile because they'll instantly know you're a crazy person
Look Pete. I love your stuff, you got me into Risk, but let's stop the Mr Brightside slander okay? That's a UK classic (I'm aware they're not from UK), we've adopted it as a country
Mr Brightside is a dance floor hit at a wedding. Also Brandon Flowers is Anglicized, a man of culture.
Maybe it's a generational thing or something, but Mr Brightside is just seen as cliche, cringe, overplayed and an absolute vibe killer for 30 and unders, I'm from the UK too lol
@@manawearblack Maybe it is a generational thing. Where 30 and overs use cringe in the right context.
@@joeh5315 In this context cringe is merely a shortened version of cringeworthy, although it is still in its neologism phase so I get why you were confused
Lmao sure you're not an Aussie got some real crowded house vibes happening.
Alternate title (if it weren't too long): when players look like their teaming, but actually, they're just sharing the same braincell
Orange cats play Risk?
That was an impressive game, Pete!
Thanks Pete! Was just looking for something to watch while I try to shut my brain off.
Damn :p
You did it! Your broke the logic of everyone, by capping in Sevastapol!
Madness? THIS. IS. SEVASTOPOL CAP!
Came by the comments yet again to double down that Mr. brightside is NOT shit. Carry on.
That failed cap steal near the end 🤣
SPOILER - I think magenta wanted to play good neighbor in the 1v1.
I never thought any of the players were teaming. This is just the bad strategy we see all over the game
Magenta taking cards and then closing there cap in the first turn of the 1v1 was hilarious lmfao🤦
Where do you find these players?😂😂
Hey Pete!
Because of you I tried to make eggnog out of curiosity.
Definitely a mistake... I should never have fallen into such a delicious trap!
All the music talk in this reminded me of a thought I had before but always forgot to share. Would you be able to add an obs plug-in that displays the currently playing song somewhere on the screen? Would be great for twitch and would be nice for yt as well, adding some context to the dancing Pete head.
That’s a good idea. When Pete is seriously head banging to silence whilst miming, I always thought he’s listening to something heavier, like AC/DC. Nope, turns out he’s listening to Madonna 😂
As far as a machine learning model for making better AI... that's a pretty good idea, assuming they have a database of every game ever played. AND hire some expert players to say what types of moves in what positions would be considered a good/bad move. Since diplomacy is part of the game, the AI would need to behave like a player if they want it to work better. Noob slams and breaks are common noob tactics. But not every noob plays like that, and not every GM plays diplomatically (as seen by Pete's tactical attrition/hamstringing against players in many vids/streams).
The problem is that it is SMG, and they don't have a great history of making AI (like, you know, for cheat detection).
Machine learning would be the best path forward for a decent RISK bot, wouldn't be very hard for SMG to train either. I'm a ML engineer that does solo contract work, think I could probably get a solid prototype done in less than a week with the right dataset - definitely better than an expert bot lol. Something that I could even implement without too much difficulty is a bot that learns the play style of a player from their early game, and if they bot out continues to play using their style/ability level.
Is it a priority investment for a broken game though? Definitely not.
53:10 Moscow's failed offensive in Crimea (if only...)
There will be no killers slander Pete!
Hey Pete had a game that went for about 3.5hrs on the meta settings with true random dice and my opponent bot after 2hrs in the other opponent started to attack me we did multiple cap switch’s and had lots of pockets I had to grind him down but at my own risk also died a few time is there a good way to beat someone have enough to beat the bot that had 2000 troops more than me was it a lost battle or something else would love your advice
Further more the bot for some reason had like a 1000 troops on cap wouldn’t move them so couldn’t cap run
machine learning eh, i actually wanted to train a model using graph neural network concept (to learn enemy troops position and their relationship with the map and players), with transformers for learning the game state (to capture the time variable, not sure if its that useful, but its a normally an important complexity)
would do this if i had the time :(
How funny would that be if AI was taught risk and there was a calculable reason why Sevastopol was actually a good cap
23:42 i mean, it's the "kill pete" strategy
Advanced Spoiler Blocking Strats!
What's that capital 😂😂
Machine learning might be their best option. It learns from data, which they have from people playing. In effect they will finally start to learn from the peteman by proxy.
Kanye is the GOAT frfr
First comment?
No I’m first
Mr. Brightside has a garbage melody, a garbage beat, garbage lyrics, yet is somehow in the top 5 rocks songs on Spotify 🤦♂
I believe that « capitals » is a very misleading name for this mode.
Indeed, in many games where you see the notion of capital, it is something you want to protect and therefore bury into your territory, as far as possible from the « front line ». Here SMG gave caps one main advantage, which you can only use if opponents roll it. Which is why capitals in Risk are used very differently from other strat games…
So confusion is the only thing you can expect from noobs coming into the game. Truly tragic for them, hilarious for us though!