|
I feel like the apt analogy here would be if DoomRL only dropped like 10 weapons in the entire game. You'd be basing your build around what you found instead of taking the weapons for granted.
|
#
?
Sep 27, 2019 01:18
|
|
|
|
| # ? Jun 3, 2024 17:18 |
|
|
I couldn't keep playing ToME because it has dungeons with open edges that you just can't walk through and it drives me nuts.
|
|
#
?
Sep 27, 2019 04:03
|
|
|
Splicer posted:Can anyone recommend any roguelikes that are cute and good like sproggiwood but without the ironic genocide There's no irony.
|
|
#
?
Sep 27, 2019 04:13
|
|
|
Tuxedo Catfish posted:this isn't true of DoomRL, ToME, Caves of Qud, or frankly even Crawl (Specifically, it has 3 charges when you find it, you can recharge it once for another 3, then use the empty wand until it gives up one last wish and disintegrates. I may be wrong about how many you get from the recharge but you can only recharge wands of wishing once so you gotta do it with a blessed potion of, I think, booze. Something that doesn't have any obvious relationship to recharging wands, anyway)
|
|
#
?
Sep 27, 2019 04:47
|
|
|
IronicDongz posted:I think we're talking about different levels of adaptability here because a lot of people playing crawl will do things like, for example, ignore an early yred altar because it's not the one they planned on worshipping but taking that option would make them way more likely to survive the earlygame. or ignoring useful evocables or spells. frankly a whole lot of people just go "well, I'm a melee guy, so" and then ignore everything that's not a bigger axe in that game This works because Crawl throws enough options at you early enough that you can survive 80% of the time waiting out for a floor or two. Other roguelikes/lites aren't as consistent about this -- Slay the Spire for example you absolutely have to take nut cards early on even if you don't "want" them. A big part of it I think has to be chalked off to "gamefeel", picking Yred early on pigeonholes you into either A. a particular playstyle or B. dealing with wrath effects when you convert to another god you like better, whereas in STS or FTL you just deal with the power card/weapon you picked up until it's not as big a part of your deck anymore.
|
|
#
?
Sep 27, 2019 05:03
|
|
|
an early altar is an invitation
|
|
#
?
Sep 27, 2019 05:28
|
|
|
FTL talk: I like to compare FTL to the original Doom, in that someone coming across it nowadays might not really see what's so special about it. It's a product of its time, when there was a dearth of indie titles, kickstarter was just getting big, and Humble Indie Bundles where still awesome. And, much like OG Doom, it still holds up despite its well-documented flaws, mostly because of its fun core gameplay. It probably sold a few copies to people who didn't click with it, but that's ok. On Cogmind/FTL/Roguelikes in general and the players ability to force their own character development; I think the forced adaptability thing is totally a merit. Brogue is basically just that, right? Cogmind also has an easter egg mode where you get complete control over your item drops, it made for some very very fun deathstar builds. And for more traditional RLs like Crawl; IronicDongz posted:[...]Frankly a whole lot of people just go "well, I'm a melee guy, so" and then ignore everything that's not a bigger axe in that game
|
|
#
?
Sep 27, 2019 05:32
|
|
|
Zereth posted:Nethack has a guaranteed source of, if you have sufficient system mastery, I think 7 wishes, so if you can make it to the Castle and get the wand you can make basically anything work by brute force wishing for the required objects. the recharge is 1-3 i think, so a guaranteed 5 wishes that's usually enough to equip any character with what they need. nethack seems very scary and random until you get to the castle a couple of times and then you understand that it's really not as capricious as it seems. the post-castle stuff is a much more straightforward hack and slash dungeon crawler experience compared to the paranoid roguelike feel of the dungeon
|
|
#
?
Sep 27, 2019 05:46
|
|
|
Unormal posted:There's no irony.
|
|
#
?
Sep 27, 2019 07:14
|
|
|
Noita people are
|
|
#
?
Sep 27, 2019 08:18
|
|
|
I don't have crazy playtime in FTL and only beat the game once but I played a bunch of other wannabes and none of them came even close to it for me.
|
|
#
?
Sep 27, 2019 08:40
|
|
|
IronicDongz posted:I think we're talking about different levels of adaptability here because a lot of people playing crawl will do things like, for example, ignore an early yred altar because it's not the one they planned on worshipping but taking that option would make them way more likely to survive the earlygame. TooMuchAbstraction posted:I feel like the apt analogy here would be if DoomRL only dropped like 10 weapons in the entire game. You'd be basing your build around what you found instead of taking the weapons for granted. Doesn't always work and then you just never get that one spell that would've been really central to your build and you have to improvise around what you managed to find instead, but there's always the option of trying. That sort of freedom is important to me in a game.
|
|
#
?
Sep 27, 2019 09:59
|
|
|
Cardiovorax posted:Yeah, this really. FTL had this sense of "you take what you can get or you won't get anything" to it, while most roguelikes I enjoy have a bit more flexibility where you might need to adapt to your circumstances to start out with, but can push the game further and further into the direction you actually want to go as you go on.
|
|
#
?
Sep 27, 2019 10:16
|
|
|
I might try that sometime, then. I've been putting it off because I'm just really not into that whole deckbuilding thing, but with how much everyone seems to love that game, I think it's worth taking the plunge.
|
|
#
?
Sep 27, 2019 10:20
|
|
|
Cardiovorax posted:I might try that sometime, then. I've been putting it off because I'm just really not into that whole deckbuilding thing, but with how much everyone seems to love that game, I think it's worth taking the plunge. I think that thinking about it like a "deckbuilding game" turns some people off who would enjoy it, because it's not like you walk into every run and have to assemble a 30 card deck at the beginning and go from there. Every decision you make is very bite-sized. When you win a fight, you get three card options to pick between - by the end of a run, sure, you'll probably have like 30 cards in your deck, but you don't make those decisions all at once. Events have limited choices and the game is really good about explicitly communicating the effects of those choices to you. The game is also very well balanced in favor of fun first - the best Slay the Spire players have drat near a 100% win rate at the standard difficulty, with players pulling off win streaks of 50+ games. If you find the game too easy, the best players' win rates at the highest level of difficulty (Ascension 20) are closer to 25%, and each level of Ascension increases the difficulty in a discrete way - one step causes more Elites fights to appear on the map, another step increases the damage enemies do, while a completely different step increases hit points - in other words, even the difficulty increases are bite-sized. It's really good, and the devs have been working on continually tweaking and making improvements and adding more content.
|
|
#
?
Sep 27, 2019 10:44
|
|
|
Does anyone have any tips for Enter the Gungeon for a first time player? I just realized I bought it a long while back, but only maybe did two or three runs before getting distracted by something else. Now, I wanna dive in with both feet. So, what should I know/do/avoid?
|
|
#
?
Sep 27, 2019 11:13
|
|
|
ecavalli posted:Does anyone have any tips for Enter the Gungeon for a first time player? I just realized I bought it a long while back, but only maybe did two or three runs before getting distracted by something else. Now, I wanna dive in with both feet. So, what should I know/do/avoid? I am not great at Gungeon but one tip I can share because it was a tip I needed when I first started playing: realize the dodge roll is your oh poo poo move, not your regular avoidance maneuver. Learn to dodge patterns with normal movement as much as possible.
|
|
#
?
Sep 27, 2019 11:44
|
|
|
lordfrikk posted:I don't have crazy playtime in FTL and only beat the game once but I played a bunch of other wannabes and none of them came even close to it for me. I disagree, I played one or two FTL-alikes and they certainly matched the original for being boring and poo poo.
|
|
#
?
Sep 27, 2019 12:17
|
|
|
I played an FTL rip off that was in most respects a perfectly serviceable clone except for one difference that made it completely unplayable: it had better AI. in FTL the enemy ships all fire their weapons at random rooms whenever they're off cooldown. in this one the enemy would play like you did: saving up volleys to overwhelm defences and then focusing down your weapons and keeping them suppressed so you couldn't fight back. Turns out FTL isn't fun when the enemy disables all of your weapons on the first shot and never, ever lets you repair them.
|
|
#
?
Sep 27, 2019 12:52
|
|
|
Olesh posted:I think that thinking about it like a "deckbuilding game" turns some people off who would enjoy it, because it's not like you walk into every run and have to assemble a 30 card deck at the beginning and go from there.
|
|
#
?
Sep 27, 2019 15:08
|
|
|
I'm almost done with my first full run through of Crying Suns. I'm not disappointed with my purchase, but there's no way I play this as much as I played FTL. The plot-heaviness really impedes replayability. On the bright side, the story's pretty decent and the presentation is great. On the downside, 95% of fights are too easy and there aren't enough random events. edit: They just released a road map of how they're going to tinker the game: https://www.kickstarter.com/projects/106952529/crying-suns/posts/2636004?ref=ksr_email_backer_project_update_registered_users It addresses a lot of concerns I have with this game. If you're interested in playing this one, I'd recommend waiting until all these changes are implemented to get the best experience. Happylisk fucked around with this message at 16:18 on Sep 27, 2019 |
|
#
?
Sep 27, 2019 16:05
|
|
|
megane posted:Hm, are there any games like StS where you do build a deck in the MtG sense? Like... I dunno, at the start of each run you draft a deck, or maybe you can have decks saved and then pick one for each run or similar. I've never heard of a game like that but it could be fun.
|
|
#
?
Sep 27, 2019 16:14
|
|
|
megane posted:Hm, are there any games like StS where you do build a deck in the MtG sense? Like... I dunno, at the start of each run you draft a deck, or maybe you can have decks saved and then pick one for each run or similar. I've never heard of a game like that but it could be fun. Card Quest is kinda-sorta like that! It has you building a deck at the start of each run using discrete chunks of cards granted by equipment; You equip a Mace and get THESE five cards, then the Hide Armor and get THESE four cards, etc.
|
|
#
?
Sep 27, 2019 16:28
|
|
|
Did the most recent beta update of Nova Drift increase the damage of the spike mines? If they did I really question if that was necessary, they were already the cause of the lion's share of my deaths.
|
|
#
?
Sep 27, 2019 17:28
|
|
|
Yeah, I love Card Quest, it's definitely the closest thing I'm aware of and is probably the only good mobile game ever made. But it's not quite the same feel as building a deck one card at a time.
|
|
#
?
Sep 27, 2019 17:30
|
|
|
Card Quest is all about being good at cardplay, since you rarely want to swap cards over the course of the run. Do note there is some serious meta-progression, since all the unlockable equipment (card packs) is generally better than the starting equipment (card packs), with some pieces of equipment being straight upgrades that replace your starting equipment forever.
|
|
#
?
Sep 27, 2019 17:41
|
|
|
Olesh posted:I think that thinking about it like a "deckbuilding game" turns some people off who would enjoy it, because it's not like you walk into every run and have to assemble a 30 card deck at the beginning and go from there. Every decision you make is very bite-sized. When you win a fight, you get three card options to pick between - by the end of a run, sure, you'll probably have like 30 cards in your deck, but you don't make those decisions all at once. Events have limited choices and the game is really good about explicitly communicating the effects of those choices to you. If it's actually as good as its reputation then I'll be happy to have been wrong about that, though.
|
|
#
?
Sep 27, 2019 17:46
|
|
|
Slay The Spire is the game that made deckbuilding games start flooding the market. It isn't the first one ever created or anything but its success probably made a bunch of people say "Oh I can do that"
|
|
#
?
Sep 27, 2019 18:32
|
|
|
cock hero flux posted:I played an FTL rip off that was in most respects a perfectly serviceable clone except for one difference that made it completely unplayable: it had better AI. That's the secret of a lot of game ai, if it's competent at all it will wipe the floor with players and you have to make it stupid See: chess, XCOM 2
|
|
#
?
Sep 27, 2019 19:16
|
|
|
Top-level chess AI these days is still only a little bit better than human grandmaster players, as far as I'm aware.
|
|
#
?
Sep 27, 2019 19:19
|
|
|
And for your other example, asking if the XCOM2 AI is good enough to beat a human player in a fair fight is like asking if the rocks in the arcade game Asteriods could beat the player if they had guns. Its a nonsensical question. The AI plays within a limited framwork of being a mook, and a better AI would be welcomed by the player base.
|
|
#
?
Sep 27, 2019 19:27
|
|
|
Captain Foo posted:That's the secret of a lot of game ai, if it's competent at all it will wipe the floor with players and you have to make it stupid Seriously. I was playing an angband variant recently where whoever spun off the variant gave the enemy AI custom scripting to thwart all the usual angband tactics like anti-summoning corridors and phase door spam. Tactics which are really necessary in a game where you have to kill 10 million enemies without dying once to win and out-of-depth enemy spawns can easily kill you in a single turn if you haven't prepared properly. Enemies don't have mana limits in that game, so every unique enemy fight was an excruciating grind as unique enemies would just run away and spam heals and infinitely summon clusters of high-level monsters around the player. Cardiovorax posted:Top-level chess AI these days is still only a little bit better than human grandmaster players, as far as I'm aware. I don't understand your point. The computer can beat anyone in the world, but if you devote your entire life to chess, it will only beat you by a small margin?
|
|
#
?
Sep 27, 2019 19:30
|
|
|
It's a bit of a matter of complexity - the more moving parts and rules a game has, the worse an AI will do at playing it. Strategy games are by definition what they handle worst, because there are so many creative and off-the-cuff ways for a player to handle situations that the AI was never programmed to deal with. Now, in something like a fighting game, where superhuman reflexes, the ability to read your inputs and the ability to perform perfect inputs of your own every time is an immeasurable advantage? There, you really need to make your AI sandbag hard in order for a human player to have any kind of chance. Mzbundifund posted:I don't understand your point. The computer can beat anyone in the world, but if you devote your entire life to chess, it will only beat you by a small margin? My point is that it's not as easy as to say that a computer will always do better than a human player. Chess is structurally very simple and practically a solved game, where it is possible to write down and know literally every possible combination that the board is capable of taking. That a human can still win, even in the face of that, makes a very strong statement about the limitations of chess AI. In the of game of Go, for example, AIs still cannot outdo more than even a mid-level amateur player. Cardiovorax fucked around with this message at 19:35 on Sep 27, 2019 |
|
#
?
Sep 27, 2019 19:32
|
|
|
Serephina posted:And for your other example, asking if the XCOM2 AI is good enough to beat a human player in a fair fight is like asking if the rocks in the arcade game Asteriods could beat the player if they had guns. Its a nonsensical question. The AI plays within a limited framwork of being a mook, and a better AI would be welcomed by the player base. I meant on the lower difficulty settings there's a random chance the calculated move will be ignored in favor of deliberately making a gross blunder. There's also things like sectoids almost always using raise zombie as a first move despite it rarely being their best choice
|
|
#
?
Sep 27, 2019 19:39
|
|
|
Cardiovorax posted:In the of game of Go, for example, AIs still cannot outdo more than even a mid-level amateur player. This hasn't been true for some years.
|
|
#
?
Sep 27, 2019 19:48
|
|
|
In general the purpose of AI is to provide a fun challenge for the player to overcome. They should only "try to win" (i.e. play like a human would) to the extent that that creates a fun challenge. In particular, the greater the statistical advantages the enemy has, the stupider they have to be to continue to provide a fun challenge. Conversely, the weaker the enemy is the smarter they can be without being imbalanced...though Tucker's Kobolds can still easily be an unfun slog if they're used poorly. Nobody likes having to chase down enemies that will harry you from range and then try to run away when you get close, nor do they appreciate being forced to expend all of their ranged weaponry against such foes...especially if they duck behind cover as soon as you haul out your bow or start casting Magic Missile.
|
|
#
?
Sep 27, 2019 19:51
|
|
|
palecur posted:This hasn't been true for some years.
|
|
#
?
Sep 27, 2019 19:51
|
|
|
Cardiovorax posted:My point is that it's not as easy as to say that a computer will always do better than a human player. Chess is structurally very simple and practically a solved game, where it is possible to write down and know literally every possible combination that the board is capable of taking. This is not true. Chess's number of possible board states is greater than the number of particles in the universe. Even if you had a magical superdense storage format so efficient that you could somehow store a board state on a single electron, the hard drive needed to store all possible board states would be physically bigger than the entire universe. If you only stored "likely" board states, your superdense hard drive would still be far larger than the planet, even with the most restrictive interpretations of the what board states are likely. Cardiovorax posted:In the of game of Go, for example, AIs still cannot outdo more than even a mid-level amateur player. In 2017 the Go AI AlphaGo beat the best Go player in the world in a tournament match, and it has only gotten stronger since.
|
|
#
?
Sep 27, 2019 19:51
|
|
|
Mzbundifund posted:This is not true. Chess's number of possible board states is greater than the number of particles in the universe. Still, the point that it's a very simple game still stands. It's exactly the kind of game that AIs do best at, and at the last time I checked, it was still possible for people to beat them, if not reliably so.
|
|
#
?
Sep 27, 2019 19:55
|
|
|
|
| # ? Jun 3, 2024 17:18 |
|
|
Deep neural networks combined with reinforcement learning had gone a long ways on those sorts of problems in the last decade. These sorts of algorithms are much better at finding non-linear and off-the-cuff interactions for their Monte Carlo tree search by analyzing millions of high-level human games, and then using reinforcement learning to boost their skill. So against the very best AIs, even grandmasters struggle to beat them without finding one weird trick that was overlooked by the AI. Of course, these AIs require immense amounts of processing power to train, but once trained they can run on a reasonable high-end computer. That said, Chess AIs can become very strong even without deep neural networks by combining a lengthy opening book, endgame tablebase, Monte Carlo tree search, and player advantage scores.
|
|
#
?
Sep 27, 2019 20:08
|
|