Botting discussion

[Zampfeo]

just raw number checks.

From what I've heard and seen pvp is actually just running and maybe numberchecks if you run good enough. So bots would have to be able to move without it being too easy to trick them etc.

100% automated PVP bots seems futile without the ability to use AI learning. What would make more sense is a bot that augments a human player by drinking and using the correct moves for them while the human controls the movement. Perhaps with some pathfinding, multiboxing support could be added so the human player could control multiple of these bots. This sort of thing exists in Old School Runescape and vanilla World of Warcraft.

[Varralien]

You can already do so, without any bots. Most players progress very slowly and without focus; if you were to focus on PvP from the start you could easily grief the hell out of most players even as a hermit. Hell, by now if you've been playing well enough you can even grief most players with an unfocused character.

Bots are not a magical auto-win button. They're tools to make the things you're already doing more efficient. If you're not already no-lifing and progressing optimally, bots won't put you on the same level as no-lifers playing well without bots.

The point is that an "equivalent" player i mean someone with similar playtime/knowledge (Think me with bots vs me without bots). With bots making me way more efficient i would easily squash an equivalent in probably any measurable metric. Its obvious that someone with more dedication/hours to put in / efficiency is going to win out most of the time against another player but it feels really cheap that you can gain efficiency through setting up a few bots, and gain efficiency that isn't possible to gain if you aren't a botter. You would need to put in way more hours/effort to gain the same out of the game.

I also understand that trying to be in the competitive / top percentage of this game basically means you have to bot and i'm not suggesting people just do it of their own free will as thats dumb.
There either needs to be a blanket removal (unrealistic).
or:
A more realistic way to approach the problem would to be add official methods of efficiency (such as the study table which was brought up earlier in the thread) to keep the botting advantages to a minimum.

[MagicManICT]

100% automated PVP bots seems futile without the ability to use AI learning.

Deep blue was created simply by using deep thinking. AI has changed a lot over the last 20 years, but Haven combat is probably a lot easier to develop AI for than chess.

[Zampfeo]

100% automated PVP bots seems futile without the ability to use AI learning.

Deep blue was created simply by using deep thinking. AI has changed a lot over the last 20 years, but Haven combat is probably a lot easier to develop AI for than chess.

Deep thinking is still AI learning. My point is the AI cannot run 1,000,000s of combat simulations because the Haven server is closed source and unable to be emulated, therefore no learning can done. Contrast this to Dota 2's OpenAI which was trained with 10,000 years worth of dota 2 games against itself. Chess is also a much simpler game. It is turn-based and has much fewer variables. There are 64 squares on a chess table while there are a near infinite amount of player positions in Haven.

[MagicManICT]

The design of deep blue was with some significantly different theories and concepts. Some of the modern AI development tools are based on work IBM did in the 80s and 90s to make Deep Blue, but much has advanced beyond those design techniques... And lets face it: Haven combat, even with trying to recognize specific objects, could be distilled into a fairly simplistic set of choices rather than the breadth and depth required to play chess.

As far as being able to be "the best"... it'd take a lot of work. For beating up animals or PvP chumps like me... probably not much. (To be fair, would love some ai bots to challenge to up my game a bit, whether they were set up in Valhalla or you could meet them in the actual game at someone's village.)

[kabuto202]

just raw number checks.

From what I've heard and seen pvp is actually just running and maybe numberchecks if you run good enough. So bots would have to be able to move without it being too easy to trick them etc.

100% automated PVP bots seems futile without the ability to use AI learning. What would make more sense is a bot that augments a human player by drinking and using the correct moves for them while the human controls the movement. Perhaps with some pathfinding, multiboxing support could be added so the human player could control multiple of these bots. This sort of thing exists in Old School Runescape and vanilla World of Warcraft.

You forget that Haven is an RPG. You don't need a bot to be more skilled than a player. You just need it to be able to communicate with other bots, not get stuck on pathing, and be able to attack a target. All other issues of PvP bots can be solved by adding more bots.

[Lightning2]

100% automated PVP bots seems futile without the ability to use AI learning.

Deep blue was created simply by using deep thinking. AI has changed a lot over the last 20 years, but Haven combat is probably a lot easier to develop AI for than chess.

Deep thinking is still AI learning. My point is the AI cannot run 1,000,000s of combat simulations because the Haven server is closed source and unable to be emulated, therefore no learning can done. Contrast this to Dota 2's OpenAI which was trained with 10,000 years worth of dota 2 games against itself. Chess is also a much simpler game. It is turn-based and has much fewer variables. There are 64 squares on a chess table while there are a near infinite amount of player positions in Haven.

You have to realize you don't need the full source to develop an AI. And player positions... Uhm

[Apocoreo]

Ya'll going right over my head with the AI talk. I think it would be pretty simple to bot combat in haven. I was thinking it's possible in Python of all things. The OS module let's you scan the screen for a saved image. If you have 1 for each combat move and all 400 of the openings you could make a painfully tedious script that scans for moves and reacts based on openings. If green opening over 30, hit the button that closes green until its not.

Of course that's very small brain retard design and would be highly limited. You'd be running it on your main screen on 1 account at the time and you would still have to move and agro people yourself. Also you'd have to take over 400 screenshots and copypaste the line of code to find the image. And it would not work if they change the move images or on clients that do that.

All I'm saying is I wouldn't be surprised if people already bot combat, even if it's a simple, stupid python script that lets the player focus on drinking and running.

[czaper2]

The OS module let's you scan the screen for a saved image. If you have 1 for each combat move and all 400 of the openings you could make a painfully tedious script that scans for moves and reacts based on openings. If green opening over 30, hit the button that closes green until its not.

I mean, the entire client is open source and those values and combat maneuver information are stored somewhere - a place where more code can easily be added.

Or you could always initiate another thread with a virtual machine language interpreter that can interact with the host environment, allowing for the reading of such buff information and even the sending back of combat oriented wdgmsg's directly towards the Fightsess widget

I mean, if you're a complete psychopath that is.

[TheTylerLee]

Could someone share a bot client ?