Silly computer, can we get a fix? — Polls

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/12/2016 8:30 PM UTC #1

Auto-translated

Friends, how would you feel about some fixes to the AI's actions in combat tactics?
For example, I am strongly surprised by modern tactics known as "kiting," where the computer cannot correctly calculate priorities and choose a strategy for attacking player units, splitting its attacking stacks which then chase a single fake unit throughout the entire battle, paying absolutely no attention to shooters, tent saves, or throwing carts at the breach; or in the egregious case of Luna, the computer could avoid splitting units into stacks or wait until Luna runs out of mana and retreats, instead of unnecessary suicide.
I am sure you know plenty of similar examples; it is great that inquisitive players found weaknesses in the algorithms, but nonetheless, this is essentially bug-using. The AI's autobattle on the map is calculated using completely different mathematics, and playing against it becomes totally uninteresting.
Furthermore, such a fix would significantly affect online play for obvious reasons, making it substantially longer; "single-unit" tactics will either stop working or will also be used by the AI, so it would be interesting to hear the opinions of the veterans. It will no longer be possible to resolve some Jebus Cross in an hour or an hour and a half without proper building. Apparently, we'll have to build the Capitol:smile04:
On the other hand, aren't you tired yourselves of looking at ducks and searching for the Griffin Conservatory?

Statistics

Registered: 10/16/2007 6:22 AM UTC

Posts: 3,358

10/13/2016 8:11 AM UTC #2

Auto-translated

A fix is definitely needed, and it's been needed for a while. But who's going to make it? You? I doubt it. This has been discussed countless times, for various reasons, but there are still NO source codes for the game.

Сейчас балуюсь этим https://vcmi.eu/download/
За то что нейтралы не бегают за единичкой, для меня это уже стоит того

In reply to Lider1

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/13/2016 10:00 AM UTC #3

Auto-translated

>This has been discussed a hundred times already
Would it be difficult to provide a link to the discussion?
>you?
A group of enthusiast friends, as always, including myself.

In reply to Doomsday

Statistics

Registered: 10/16/2007 6:22 AM UTC

Posts: 3,358

10/13/2016 3:46 PM UTC #4

Auto-translated

Doomsday

Wouldn't it be easy to provide a link to the discussion?

It will be difficult to provide a specific link, as this happened not yesterday and not in one place. But the fact that the AI is an idiot became known not this morning. (Although for some, this may be a revelation, so congratulations to them).
Mentions of this were in the thread with the ND mod, in the round table, in the offline game section, and reports also periodically mention the AI's stupid actions.

Doomsday

>you?
A group of enthusiast friends, as always, including myself.

Good luck to you. However, I highly doubt that this will go beyond words. Because programming AI is a very complex task.
I'm not even talking about how you (you all) will do this without the source code?

Сейчас балуюсь этим https://vcmi.eu/download/
За то что нейтралы не бегают за единичкой, для меня это уже стоит того

In reply to Lider1

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/14/2016 10:11 AM UTC #5

Auto-translated

>how will you do this without the source code?
Just like a respected group of friends created the HD mod. I wouldn't want to discuss the technical aspect of the problem here because it doesn't present significant difficulties. We have dealt with "pirated" releases, modifications, and localization of several Korean MMO games, for which there were also no source codes or reference materials.
>The AI is idiotic; this has been known for a while.
>Programming AI is a very complex task.
That's why I would like to see some theoretical developments on what the tactics could be, because perhaps there are already brilliant, unrealized ideas here.

Statistics

Registered: 4/29/2007 10:30 AM UTC

Posts: 11,704

10/14/2016 11:11 AM UTC #6

Auto-translated

Doomsday

For example, I am very surprised by the modern tactic called "divorce," where the computer cannot correctly calculate priorities and choose a unit attack tactic, dividing its attacking stacks, which chase after a single decoy throughout the battle, paying absolutely no attention to ranged units.

You have Sandro from the second Heroes game as your avatar. I hope it's not a secret to you how neutral units react to ranged units? They are ready to run 10 turns around to somehow reach the ranged units. That is, from one extreme, we get another. In any case, no matter how much the AI is changed, players will learn to make it a "dumb computer." It also turned out that the computer's role is to be a bystander in a game with people. Therefore, I do not like the idea of making any changes to the gameplay.

However, if the AI's intelligence is brought to the level of chess, it would certainly improve it as an opponent. But firstly, this is a very complex and expensive process. Secondly, playing chess against a computer is boring.

In reply to Dirty_Player

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/14/2016 11:18 AM UTC #7

Auto-translated

>from the second game
Yes, it's worth mentioning that I prefer the second game to the third.
>10 turns to go around, just to somehow reach the archers
The archers need to be "assessed" first; rushing towards 10 gargoyles isn't very wise if the player has a significantly stronger stack(s), against which the damage from the gargoyles will be negligible. Furthermore, the AI evaluates its chances of reaching the archers; if there are none, the AI should inflict maximum damage on the player, rather than trying to attack what it's given.
In my opinion, the main problem is the assessment of the situation. Before the battle begins, the player usually has information about the possible outcome and builds their tactics based on this. At best, the AI can only review the preliminary results of the auto-battle.

Let's take a situation where the player has a stack of units, let's say level 1, which the AI sees as units from 1 to 5, and expert slow: obviously, it's not beneficial for the AI to divide its units into several stacks, but it does so because it doesn't know that there's 1 unit, not 5. As a result, all units are slowed, and the superior forces are killed by a single stack of divided units that is only slightly stronger.
This raises several questions:
1. Should the AI have access to information about the presence of "key" spells for the attacking player? Resurrect, slow, haste, etc.
2. Should a "decoy" of level 1 units be left? In this situation, the AI could distribute its 60 conditional centaurs not into 20/20/20, but into 1/54/1/1/1/1/1.
3. The AI could assess the player's SP, remaining mana, and perform a kind of predictive analysis: "the player can cast slow one more time for 3 turns," go through the player's spellbook options, and defend until the slow wears off, if the damage from the ranged stacks is insignificant and doesn't reduce the number of centaurs to the point where the player's stack can kill them without significant losses, or the remaining centaurs can still reach and inflict some damage on the player's key stack.

>like chess
I see it exactly that way; the AI should model a table of its possible actions and the player's actions for n+ turns until damage is inflicted, and choose the combination that inflicts the greatest strategic damage, adjusting it each turn, depending on changes in the player's actions. The difficulty level of the game will be determined by the depth of that n.

Statistics

Registered: 10/16/2007 6:22 AM UTC

Posts: 3,358

10/14/2016 1:37 PM UTC #8

Auto-translated

As I see, you didn't just come here to theorize.

The developments of vcmi might be useful:
https://github.com/vcmi/vcmi
They are developing an engine with open source code.

Website: http://forum.vcmi.eu/portal.php
Bug tracker: http://bugs.vcmi.eu/my_view_page.php
Unofficial daily builds for Windows: http://vcmi.arseniyshestakov.com/windows/

It is planned to be cross-platform. There are versions for Windows, Linux, and Mac. Someone from the community compiled it for Android, but it hasn't been updated for a very long time and is hopelessly outdated.
The version is still raw, with a large number of bugs, it's basically a deep alpha. And one of the main drawbacks is that the AI is even weaker than in Heroes 3.
I'm helping a bit with testing as a player. Usually, these are minor, obvious bugs. Personally, my goal is to see a competitive AI. Well, maybe not competitive, but at least not as idiotic as in the original. Therefore, personally, I am all for such an idea.

Added 1 hour 51 minutes ago

Dirty_Player

Secondly, playing chess against a computer is boring.

Of course, competing with a tractor in strength or with a racing car in speed, who would be interested in that?

Dirty_Player

In any case, no matter how the AI is changed, players will learn to make it a "dumb computer".

Dirty_Player

If, however, the AI's intelligence is brought to the level of chess, it would certainly raise it as an opponent.

You contradict yourself ^ It's not a fact that it will be "dumb".

Dirty_Player

Also, it turned out that the computer's role is to be a spectator in a game with people. Therefore, I don't like the idea of making any changes to the gameplay.

That's just how it historically happened. And many have convinced themselves that it should be that way. But why?
If there were a real choice in practice, I wouldn't even discuss this issue.
Either wait for the opponent's turn, or rush in with a unit against a stronger one, even if it's not as skilled as a human player.

Dirty_Player

But firstly, this is a very complex and expensive process.

This is the key here. I just don't believe in the implementation of this idea. Although I would really like it.
It's like reaching for space travel; the game is really famous all over the world, and many companies like IBM tried to showcase their power in it. If a group of enthusiasts can improve the AI in Heroes, even as a fix, that would be a great achievement.

Doomsday

1. Should the AI have access to information about the attacking player's "key" spells? Resurrect, slow, haste, etc.

Ideally, no. But as an optional feature for those who like it, it could be left in.

Doomsday

2. Should the "secret" of using single units be left in? In this situation, the AI could distribute its 60 conditional centaurs not as 20/20/20, but as 1/54/1/1/1/1/1.

Of course, the AI should be aware of the method of using single units, removing the counter, and attacking with the main stack.

Doomsday

3. The AI could assess the player's SP, remaining mana, perform a kind of predictive analysis: "the player can cast slow one more time in 3 turns", go through the options of spells from the player's spellbook, and defend until the slow ends, if the damage from ranged stacks is insignificant and does not reduce the number of centaurs so much that the player's stack can kill them without significant losses, or the remaining centaurs can still reach and inflict some damage to the player's key stack.

That's a plus.
It would also be good to take into account that the AI, as soon as it gets its turn, immediately casts a spell, and then it is countered. For example, it gets its turn, casts slow, and in response, it gets haste. And so on.
That is, waiting and not casting immediately is also useful. But for this, it is necessary to correctly assess the situation on the battlefield.
And a spell like expert slow is a must-have, the AI should "know" that it should be there. And strive to get it. (And also think about how to counter it)

Doomsday

I see it this way, the AI should simulate a table of its possible actions and the player's actions for n+ turns until damage is dealt, and choose the combination that causes the greatest strategic damage, adjusting it each turn, depending on changes in the player's actions. The level of difficulty of the game will be determined by the depth of the n.

Good solution.

In general, are we talking about AI as a player, AI for neutral creatures, or AI as a battle simulator in general, one for all? As I understand it, it's for all?

Сейчас балуюсь этим https://vcmi.eu/download/
За то что нейтралы не бегают за единичкой, для меня это уже стоит того

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/14/2016 3:19 PM UTC #9

Auto-translated

>VCMI
Thanks for the tip, I've heard about this product, but I thought it was just another port to Android.

>I just don't believe in the feasibility of this idea
The "idea" is no different from what Chessmaster does; the analysis mechanism has been described in various sources.

>And is striving to achieve it
This is already a strategic map; I don't think we'll be working on it. The developments in tactics might be useful in other projects. We had an idea to collect information from the map about the spells available to the player and the potential increase in units that the player can obtain and buy, and use it to evaluate tactical decisions, but this would greatly complicate the task; there are too many options, and it would be easy to draw an incorrect conclusion.

Added after 41 minutes
>Should the AI have access to information about the attacking player's "key" spells? Resurrect, slow, speed, etc.
>Ideally, no. But as an additional option for those who like it, it could be left in.
I envisioned this on the level of "rumor," where some neutral creatures tell others that Galtran is running around and frying everyone with chain lightning. The implementation would again be at the level of a table, where we record the spell and level used by the player after each battle.

>Secondly, playing chess with a computer is boring.
Aren't you worried about the slow transformation of Heroes into a very slow-paced, turn-based Dota? Players gather on the map, grab what they can, and run to the center to find out who got more angels and ducks from the random number generator.
To some extent, of course, there is a competitive element to this, like in Magic The Gathering; the player with less skill can win against an opponent who is much better in skill and deck quality, simply because the opponent is unlucky with mana. But the box of Homm3 said "strategy game."

In reply to Doomsday

Statistics

Registered: 10/16/2007 6:22 AM UTC

Posts: 3,358

10/15/2016 11:30 AM UTC #10

Auto-translated

Doomsday

>I just don't believe in the implementation of this idea
The "idea" is no different from what Chessmaster does; the analysis mechanism has been described many times in various sources.

I'm still talking about the technical part :D

Doomsday

>And seeks to obtain it
That's already the strategic map; I don't think we'll be dealing with that.

Without competent play on the strategic map, the AI will still be in the toilet.
First of all, chains. With chains, the player will gain a huge advantage by devouring everything more.
Secondly, the imba spell mass slow — under it, everything turns into mere meat.
Perhaps the game will boil down to "got slow and smashed to splinters".

Doomsday

I imagined it at the "rumor" level — some neutrals telling others that Galtran is running around here frying everyone with chain lightning. The implementation would again be at the level of a table, where after each battle we record the spell and level used by the player.

Yeah, I'd be interested in playing against an AI that sees the entire map, knows my whole army, and the spells in my spellbook. Sort of a "cheater" level. But not everyone would be into that :D

Doomsday

But the box of Homm3 said "strategic game".

Many are so used to such an AI that they don't even entertain the thought of improving it — like, what, the exploit with one dragonfly and a hundred zombies won't work anymore?
The offline players have achieved a special art. It truly is an art (of exploitation). I vaguely remember there was some tournament; someone found a sequence where the computer had an army many times stronger and just sat in the castle doing nothing :D Yeah, just sat there watching itself get killed.
Offline players are the best testers :D

Doomsday

The game's difficulty level will be determined by the depth of analysis of that n.

And ideally, the calculations should be done on all cores, not just one. Have an eight-core processor? — then let all 8 work at 100%. Like in distributed computing.

Сейчас балуюсь этим https://vcmi.eu/download/
За то что нейтралы не бегают за единичкой, для меня это уже стоит того

In reply to Lider1

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/16/2016 3:40 AM UTC #11

Auto-translated

>Yes, I would be interested in playing against an AI that sees the entire map, knows all my units, and the spells in my spellbook. It would be like a "cheater" level. But not everyone would like that :D
It won't immediately know the spells. For each hero in the game, a table will be created with the spells they used against the Computer Player and against neutral creatures, which will be analyzed in the next battle, taking into account the remaining MP.
However, it seems that I will have to create 2 different tables: one for the Computer Player and one for neutral creatures, with an option for neutral creatures to share information about the player's spells with each other.

In reply to Lider1

Statistics

Registered: 4/29/2007 10:30 AM UTC

Posts: 11,704

10/16/2016 5:23 AM UTC #12

Auto-translated

Lider1

You're contradicting yourself, aren't you? It's not a given that it will be "dumb."

Perhaps I didn't express myself clearly, but if the AI is slightly modified, removing some obvious flaws, it won't help because other flaws will emerge, which can then be exploited. And there's no point in that, and it's impossible to do it well because:

Lider1

This is the key point here. I simply don't believe in the feasibility of this idea. Although I would really like to.
It's like comparing chess to space travel; the game is truly famous worldwide, and many companies like IBM have tried to demonstrate their power in it. If a group of enthusiasts can make an improvement/fix to the AI in Heroes, that would already be a great achievement.

Lider1

It just happened historically. And many have convinced themselves that it should be this way. But why?
If there were a real choice in practice, I wouldn't even discuss this issue.
Whether to wait for the opponent's moves or to rush in with a large army, even if it's not as experienced as a human player.

All tournaments have been and will be held in a human-versus-human format. If the computer "suddenly" starts playing better than a human (as in chess), the game will turn from "who got luckier" to "who played better than the computer" – that is, with well-known strategies like 8mm6a, it often won't even come to contact with a human, and sometimes it will all end even earlier, in a battle with neutral creatures. For offline tournaments, it would be great! But who will play them if no one is playing them now?

Statistics

Registered: 8/14/2013 9:07 AM UTC

Posts: 215

10/16/2016 4:49 PM UTC #13

Auto-translated

Doomsday

Friends, how would you feel about some fixes to the AI's actions in tactical combat?

very positively, but a lot needs to be considered and reworked...
actions in combat:
- AI as a computer player
- AI as neutrals and object guards
it would be desirable to make the AI smarter, with some deeper analytical capabilities: so that it waits out a wall of fire, uses single-target spells and removes retaliatory strikes, attacks with all stacks on one target, and much more, instead of just blindly charging and getting exposed, as it often does now.
but then you also need to think about how to avoid bugs and glitches in combat between the AI player and AI neutrals.
and also, so that it doesn't happen that the computer simply doesn't attack and waits until it has more troops than the neutrals.
another important aspect is spells; it would be desirable to remove all restrictions on their use and subject their use to more in-depth analysis.
actions on the map:
the AI computer should have clear priorities, and it is also necessary to determine: whether the entire map is known to the AI or whether there is also terra incognita for it; in any case, many separate and complex algorithms will be needed, since the AI is not capable of improvisation, so everything will have to be set in the order of priority goals...
if we talk about chess analysis, then in this regard, it is easier for a computer to analyze moves and options, since everything is in one plane - the pieces move, each along its own trajectory, and one at a time....
in Heroes, it is necessary to analyze and prioritize: the movements of many AI heroes, i.e., in one turn, many heroes can move and many battles can be fought, which requires analysis and comparison, objects on the map - their visit and capture, then maneuvers on the battle map + spells, and even the construction of objects + the purchase of creatures, i.e., the computer will have to calculate all this.
then the computer will need to somehow program: the construction of chains, the dismissal and hiring of heroes, the reasonable division of the army and its combination with one hero, minimizing losses, using spells, and the timely exploration of the map;
when building all these programs and algorithms, there should be no crashes, glitches, freezes, or errors, and at the same time, the AI should be adequate in complexity and quite beatable, i.e., the difficulty levels will also need to be reworked, because with such a restructuring of the AI's actions, the principle of division into difficulty levels will have to be changed, i.e., it will be necessary to put up barriers somewhere and somehow on the depth and power of all this analysis....

therefore, at the moment, to be honest, I can't imagine how and in what form this will turn out, and how much it will cost those who undertake such a colossal task of reworking and testing the improvement of artificial intelligence

Added 5 minutes ago

Dirty_Player

All Tournaments are held and will be held in the game as player versus player. If the computer "suddenly" plays better than a person (as in chess), then the game will turn not as it is now, "who got better loot," but into "who got better loot than the computer" - i.e., based on known templates.

then, most likely, the templates will also have to be reworked, and if the AI system is improved in the behavior of neutrals, then the strength of the neutrals on the templates will also have to be revised

In reply to Doomsday

Statistics

Registered: 10/16/2007 6:22 AM UTC

Posts: 3,358

10/16/2016 6:39 PM UTC #14

Auto-translated

Doomsday

>I would be interested in playing against an AI that sees the entire map, knows all my units, and knows all the spells in my spellbook. It would be like a "cheater" level. But not everyone will like that :D
It won't immediately know the spells. For each hero in the game, a table will be created with the spells that he used against the Computer Player and against neutral creatures, which will be analyzed in the next battle, taking into account the remaining MP.
The thing is, it seems that 2 different tables will have to be made: one for the Computer Player and one for neutral creatures, with an option for neutral creatures to share information about the player's spells with each other.

Perfect
(but even better if there is an additional option without restrictions = the entire map, all information about the opponent player)

Dirty_Player

Maybe I didn't express myself correctly, but if the AI is slightly changed, removing some obvious flaws, it won't help, because other flaws will appear, which can be exploited.

If you mean "we'll fix the old bugs... and create new ones," then the new ones can also be fixed if there is someone to do it.

And if it means that other methods will be used that give gameplay advantages, that's not so bad.
Maybe the computer won't die in the first week, as it usually does, but you'll have to take a castle, a lair, or upgrade your resource income before doing that.
At least a little, but you'll have to be wary of it (take it into account) on the map.

Dirty_Player

All Tournaments were and will be held in player-versus-player mode.

Well, I'm not arguing with that. But controlling 1-2 fairies and a hundred zombies across the entire field is clearly not normal. Although I'm sure that for some (many), this is so ingrained in their minds that it's nostalgic, vanilla, and that's the only way it should be.

Dirty_Player

"If" the computer suddenly plays stronger than a human (like in chess), the game will turn not into what it is now, "who gets the better loot," but into "who gets better loot than the computer."

I'm sure this won't happen ))

Dirty_Player

For offline tournaments, it would be great! But who will play them if no one plays them now?

I think grandmasters with the "take back" option can win against the strongest program. The same applies to playing against the computer using autosave.

But nevertheless, improving the AI would breathe new life into the game. This is what it lacks. Because playing against the AI is absolutely uninteresting. After online play, it's just garbage. Or rather, after understanding how it works.

And in general, imagine an offline tournament with an "intelligence patch"? I would be one of the first to sign up for it

Сейчас балуюсь этим https://vcmi.eu/download/
За то что нейтралы не бегают за единичкой, для меня это уже стоит того

Statistics

Registered: 10/12/2016 8:00 PM UTC

Posts: 7

10/17/2016 10:43 AM UTC #15

Auto-translated

>so, to be honest, I can't imagine how and in what form it will turn out at the moment
For now, the fix will only affect the tactical part; we won't be changing the strategic map. The AI already receives enough bonuses on the strategic map, depending on the difficulty level. Perhaps a patch for the battles will be enough for the time being.
>who got better results than the computer"
I think that if we stop the AI from controlling hundreds of zombies with a single fake hero, teach it to dispel during battles, use protection, not rush in recklessly, and not divide stacks inefficiently, then yes, the player will have to be a little wary of the computer, especially if it "deals more damage."
>i.e., on well-known builds like 8mm6a, it often won't reach the point of contact with the player,
Have you ever seen a surrender in Civ after losing one or two cities? But in Heroes, everything ends with this very first, and often last, contact, in which the one who attacked gains the advantage. And all because there is no background and the gold was spent on abusing the tavern.
>and sometimes it all ends even in a battle with neutrals
That's how it is now: brave players rush to the Guardian of Order in the first week, lose to a bunch of miscellaneous creatures because they messed up the split, or attack a beast lair with a single angel "so as not to lose tempo," not noticing that it's the maximum level, lose the angel to an upgraded stack, and also surrender. Are these really normal practices for playing a strategy game?