Install Steam
login
|
language
简体中文 (Simplified Chinese)
繁體中文 (Traditional Chinese)
日本語 (Japanese)
한국어 (Korean)
ไทย (Thai)
Български (Bulgarian)
Čeština (Czech)
Dansk (Danish)
Deutsch (German)
Español - España (Spanish - Spain)
Español - Latinoamérica (Spanish - Latin America)
Ελληνικά (Greek)
Français (French)
Italiano (Italian)
Bahasa Indonesia (Indonesian)
Magyar (Hungarian)
Nederlands (Dutch)
Norsk (Norwegian)
Polski (Polish)
Português (Portuguese - Portugal)
Português - Brasil (Portuguese - Brazil)
Română (Romanian)
Русский (Russian)
Suomi (Finnish)
Svenska (Swedish)
Türkçe (Turkish)
Tiếng Việt (Vietnamese)
Українська (Ukrainian)
Report a translation problem
Discussions Rules and Guidelines
Report this post
Talisman has a whole lot of corner cases to think of if you want to make a good enough AI.
In the case of Talisman, AIs cast spells whenever they have the occasion. That's how they have been coded. They will never hold a spell. Except maybe Misfortune where they might probably wait until you make something that isn't a movement dice roll. If they have the choice of target (but only if), they target whoever they think is the strongest one.
By the way, if the caster was a spellcaster character (aka one that gets a new spell everytime somebody casts a spell), then his move was most likely the optimal move for him: the best way to play spellcasters is to perform "spell cycling", aka cast your spells as soon as possible in order to get new (potentially better) ones, and get the most return out of your ability to cast spells.
You make it sound like every game should have decent AI, and seem to have troubles understanding the challenges behind making an AI. This topic is a very interesting read on the matter.
https://www.reddit.com/r/explainlikeimfive/comments/4y2pvf/eli5_why_is_it_so_difficult_to_design_an_advanced/
I could probably also tell you to do some research before criticizing the work of others, especially if you are not familiar at all in the domain (AI programming in this case).
No, Mr. Proper Developer Worship Watchdog, I don't have to make my own AI - I pay others to do that. I don't make my own shoes either - I buy them and expect the quality to match the price. I am well within my rights to criticize a game I've paid for.
The AI in this game is baffling. It will enter into repeated auto-lose battles with the sentinel. It will poltergeist-walk back and forth between the castle and the portal of power for 30 turns because it "can't" cross the runes or black knight (even with lives to spare). It will cast a hex spell on the oasis when they're the only character in, or with the immediate chance to enter, the middle ring. It will waste all its gold on the arc sail pilot going back and forth between two squares, never encountering either because the stuff they wanted have higher encounter numbers. If will suicide against you when you have +9 strength inventions, and it certainly doesn't understand how to make high-grade inventions for itself. It will do nothing to stop ancient beasts from reaching the apex, because the only thing that matters is that the human player loses.
I see new outrageous AI decisions nearly every game, not least these DLC-related ones that tells me that any added layer of complexity is beyond the AI to cope with, because it's more important to sell stuff than to integrate it properly. There is certainly a balancing act between playing to win and potentially suiciding against the strongest player in order to cut him down. Human players are usually good at detecting this line, and at coordinating their attacks. The AI is neither. It will just annoy the human player throughout the game, without going for the kill when it's really needed. And it becomes needed because of all the bad decisions the AI make, that make it impossible to lose unless you draw a string of dragons in the opening of the game (or pop an apex beast that causes all AIs to hide in the opposing ring, leaving you alone to handle the "all players lose" scenario). I don't expect the AI to meet human standards, only to meet 21st century AI standards. This isn't even a complex game, and so little apparent effort has gone into making the AI make any sort of situational valuations.
Saying that the AI is "coded to never hold a spell" is an admission that it's too simple. It's deliberately made that way to avoid complex valuations. A random spell on turn 1 (and this is by anyone who holds a spell, whether they will draw a new one or not) is a typical start to a new game, but this is hardly when the spell is the most effective, as you always have fate, can't lose craft or strength and even rolling two 1s is not that awful since you have no items or followers to lose. I would weigh "getting rid of it to get a new one" against throwing it on someone with 0 fate with much to lose, close to my own position, and preferably not close to anyone else. Sometimes I would even throw it on myself. The AI just throws. On you - every time.
So, the dungeon DLC is free this week and I'm playing against the ♥♥♥♥♥ character. Now I have to roll the random spell twice every time it's cast. Yay.
But to relate to my previous case, I was the strongest player and made a deliberate choice to weaken my position by stealing the hag from a weaker character, and a third character counterspelled me only because I'm the human player. They will not counterspell zero-sum situations against other AIs. The AI has no understanding of the relative value of what I was stealing, or of the counterspell itself, which I certainly wouldn't waste on a zero-sum situation that weakened the strongest player. And since I was flinging spells myself, it makes even more sense to save the counterspell for an actual defensive use case. None of these things are even remotely in the AIs playbook, because most of the AI's decisions boil down to precisely those very simple coded points: "Always take the strength point". "Always look to heal at 2 lives". "Always attack and finish off a player with only 2 lives". Some of them are good to the point where I would be disappointed if they didn't do precisely that, but these very obvious cases (that typically don't require deliberation even for human players) are really the only strong decisions the AI ever make. There's nothing beyond that. It's too primitive.
That the achievements are broken is beyond the hollow dispute of apologists. They have apparently been so for 7 years, and since the developers can't fix this problem, which I've never seen for any other game (broken achievements yes, resetting ones never), I start to triangulate what's going on with the bad AI. Maybe they're just bad developers?
Demand higher quality for your money instead of expecting a pat on the head. If they truly loved you, they would "increase their craft".
Instead of living in a world of constant criticism and demand, maybe appreciate the work of others for what they are ? Or is it too difficult for your narrow mind ?
Anyway, to your initial question "How is the AI that dysfunctional" you have your answers. Now, whether you like it or not, it's up to you. But the answer is not going to change. Either shrug it up and live with it, or.. i don't know.. maybe seek "better" games out there, where their AIs is perfect? With all the strategy and board games around, surely there must be some games with perfect AIs to your liking?
But no, inexplicably, the bot reversed direction and headed directly for the inner-ring of the main board. The bot had no open quests, nor was there any other reason for this "strategy". Maybe the light from the fire-tokens attracted it, I don't know. For MANY MANY rounds, this same bot wandered aimlessly around the inner circle, until it reached a strength of 28, well over twice the strength of the boss he was supposed to be fighting!!!
Eventually, a different bot defeated the boss to end the match. The winning bot had a strength of 20+ so it was a bit shy about completing the match objective, too.
Now I don't have to be an AI programmer to know that it can't be that difficult to program these bots to at least try to win the game in a timely fashion. An instruction like "If winner-take-all and strength >= 15 proceed to top of Dungeon" would have served this bot well in this case. That can't be that many lines of code and I can't understand why such a simple fix has not been done yet.
With the large number of players that only play against the bots, it would seem that fixing the AI should take precedence over, say, designing the next set of semi-redundant characters to sell us.
As long as there will be wrong plays by the AI, there will always be people who find them "unacceptable".
I do agree though that the AI sometimes (or consistently, for certain scenarios) acts in stupid ways that break the immersion. It wouldn't be the first nor the last game to have an AI that does questionable decisions.
I'm sorry but how can you know how difficult something is if you've never done it yourself, and if you have never looked at their code?
It really depends on how their code is structured. Maybe their AI code is a mess (it's a 7-8 years old project after all) and they would need a whole rework to add what you suggest. I have had cases in projects in the past where adding a simple feature required to change so many things that in the end it wasn't worth it.
A typical example I could tell you is Diablo 1 itself. Recently devilutionX has come out, a recreation of the source code based on what was written in assembly. I wanted to fork it and do one simple thing: add a new menu item. Like a new difficulty mode, that can be selected from the main menu. Surely it can't be that hard right? Just add a few lines of code and that's it. I couldn't be more wrong. I had to spend literally hours figuring out how their menu code was working, so I could add my own menu element - and i had to do a lot of debugging as well. It took me probably 7-8 hours to do that one change that I thought would be done in 2 hours maximum.
So as frustrating as it can be to hear, the only people who can say how easy/hard something is to do, it's them. There can be technical reasons beyond logical thinking - either code organization, or just the need to do QA after the change to make sure you haven't broken any other behavior on the bot side ; you also need to justify financial wise why spending a day doing that is good for sales, maybe this day you've spent doing that could have been spent doing something else on a new game you are working on..
If all my programming projects had been "Hey I just need to add a few lines there and that does the trick", I would have probably coded 4 times more things than I did :p
Good doggie. Did you get your treat from your masters?
I can get quite annoyed with people being 'entitled' , and remember this is a small development team that continues to work on a pretty old game release, yet at the same time we are paying customers. Writing about our gripes on this discussion board is actually a good thing - 'cause usually Nomad will see it and take it on board.
Anyway, the AI is better than it used to be - it's had improvements - but yes, I'd also like to see some more work done on it... for the longevity of the game's life.
For now - I suggest to keep respawns on (for the sake of the AI) and laugh at their noob mistakes.
I'm just sick of how toxic online discourse is.
You never have this kind of thing on Reddit or on Discord. People are much more civil there.
I am personally getting more and more tired of people who blame video games as if their life depended on it, or as if it was a super important issue that is going to break the world. This, in turn, gets on my nerves, and I become sarcastic. I probably shouldn't and just let it go and not post a thing, but it's hard, especially when you know that the devs spent a lot of time on this.
"Imbecile" isn't "toxic discourse"? You're not helping.
Maybe you weren't paying attention, but I already replied, while he didn't address a single point of mine in that second post of his. Nothing he said amounted to more than "shut up if you're not a programmer yourself". Well, I won't shut up about hacks and scammers who pump out endless DLC without fixing their core game first. I've paid for this, so whether coding AI is hard or not is not the full question.
Now this guy is putting himself on the moral high ground as the great community leader who doesn't really care about games, despite constantly scanning the forums, apparently being a programmer/modder himself, and having more logged game hours than almost anyone, it would seem. Not buying it.
And I can see you're part of the "Talisman Europe League" sect that he runs, so not exactly impartial, are you? Maybe my last comment was better directed at you, hmm?
You call the Talisman Europe League a "sect".
Sorry, but I can't take you seriously.
Well, you haven't offered anything to the issue at hand, so how seriously you take me is not something I will lose any sleep over. I think we're done here.
I check it maybe twice or three times per day probably during my own breaks.
I have a lot of hours in Talisman but recently I have almost no playtime.
Talisman Europe League isn't a sect, it's tournaments that I used to host (and will surely host again in the future once I'm done with my PhD) - I really love the investigation you are doing by the way, are you some kind of FBI agent ? It's unbelievable the lengths at which people will go just to win an argument.
My arguments weren't just "shut up you aren't a programmer". I think I detailed quite a bit and I won't go through it again.
I don't think this thread is going anywhere - those who read have enough information to make their own opinion. I'm opting out for my own sake ; I have better things to do