The_trouble_with_AI's

SSI's Imperialism and Imperialism II games have a number of features that
aren't as "realistic" as they could be (strategic movement is the obvious
example) but which were modelled a particular way to allow the AI to use
them more effectively. The overall "feel" of these strategic-level games
remains good, though, and they're still among my favorites in their genre.

Also, don't forget the design philosophy behind SSG's original Battlefront
(not the new Matrix computerised boardgame). SSG knew the AI could never
perform as well as a human player (especially back in those 8-bit, 64k RAM,
days) so they "crippled" the player by forcing him to play like the
computer. The player used the same movement and combat routines as the AI
and players were restricted to a number of decision points identical to
those evaluated by the AI.

A side effect of this was that you got an interface with a good "commander's
eye view" feel to it (no opportunity to micro-manage) but the fundamental
aim of the design was to place the AI and a human player on equal terms.

Andy
 >> Stay informed about: The_trouble_with_AI's 

Over on the Matrix forum several months ago I started a poll thread to
see how people were playing "Forge of Freedom," and found that the
overwhelming majority of our customers (80%-85% range, if I remember
correctly) were playing solely against the AI. Some people did play
against both AI and a PBEM opponent, and a small number were just into
PBEM. The most surprising number was that just 2% or 3% were playing
TCP/IP. My guess is that if one of you were to start a similar poll
thread over at Wargamer you would get similar results, and pretty much
put this debate to rest.

From a developer's point of view, it actually makes no sense to add
TCP/IP, based on the enormous amount of resources implementing this
feature requires -- all for the benefit of at best 3% of one's
customers. And a similar argument might even be made for PBEM, if
just 10-15% of customers are using that option. But if a developer
were to release a game that is AI-only, the developer and game would
be crucified by reviewers and subject to an enormous outcry here and
in the various forums. So, multiplayer always needs to be a feature
-- and a well-executed feature, with PBEM playback, etc. -- but the AI
is bread-and-butter.
 >> Stay informed about: The_trouble_with_AI's 

On Fri, 21 Dec 2007 17:36:25 +0100, "Vincenzo Beretta" wrote:

>IIRC, Brad Wardell, the creator of Galactic Civilization I & II, always
>followed this principle: there are a lot of "features" asked by the fans
>since the first GalCiv that he never implemented because "the AI would not
>be able to manage them in an adequate way". Before including *anything* in a
>patch or in a new version of the game, Wardell always makes sure that it is
>something that the AI can understand. The result is an artificial opponent
>that really kicks a** (maybe the best AI ever in a videogame). And, in my
>opinion, a side effect is that the design of his games is always very clean
>without being shallow. GalCiv II + Dark Avatar is probably the best 4x space
>game I ever played.


Not to contradict any of what you said, but IIRC GalCiv is
single-player only, with no multiplayer component at all. Of course
they had to optimize it for play against the AI!

Was that changed in GalCiv II?
 >> Stay informed about: The_trouble_with_AI's 

> IMO, wargames, especially those for more than two people, should be
> designed strictly with human-to-human play (PBEM, TCP/IP, hot seat) in
> mind, with the AI added as an afterthought"

Absolutely disagree with him.
Multiplayer is in the minority. Most gamers play even wargames in single
player only.

That sort of thinking damages otherwise good wargames - or any games.
For example Commander Europe at War would be much better game, if the AI
had some capability and thinking sense. But now, a nice game just
doesn't get any more gameplay from me, as it is just silly and boring.

Good AI should always, always be the first thing in the design....
 >> Stay informed about: The_trouble_with_AI's 

Thatks amazing, Eddy!

You've managed to push my buttons borth on customers' contradictory AI
demands and unused multiplayer features in a single post !




PAT PROCTOR
President, ProSIM Company
http://www.prosimco.com/writing
 >> Stay informed about: The_trouble_with_AI's 

pproctor69 DeleteThis @yahoo.com wrote:
> Thatks amazing, Eddy!
>
> You've managed to push my buttons borth on customers' contradictory AI
> demands and unused multiplayer features in a single post !
>
Personally I don't look for multi-player when I go looking for a game. I
suspect that the target audience of all those game sites and magazines is a
fairly small percentage of the actual buying and playing customer base.

--
Because of heavy computing requirements we are currently using some of
your unallocated brain capacity for backup processing. Please ignore
any hallucinations, voices, or unusual dreams you may experience.
Please avoid concentration intensive tasks until further notice. Thank
you.
 >> Stay informed about: The_trouble_with_AI's 

eddysterckx.RemoveThis@hotmail.com ha scritto:
[CUT]

I heard that (an Example) Bethesda implemented a good AI in Oblivion.
Now, the story was plotted, scripted... but the NPCs began to interact
with each other. Chaos was the response of a good AI. Yes, not a
storyline to follow but a world living without much control. But alive
and with it's owon life. So they did implemente only a "controlled
amount" of AI. I still prefer the living world!!


Bushi Solitario


P.S: maybe it's only a P.R. story but I still prefer to interact in an RPG!
 >> Stay informed about: The_trouble_with_AI's 

> I heard that (an Example) Bethesda implemented a good AI in Oblivion.
> Now, the story was plotted, scripted... but the NPCs began to interact
> with each other. Chaos was the response of a good AI. Yes, not a storyline
> to follow but a world living without much control. But alive and with it's
> owon life. So they did implemente only a "controlled amount" of AI. I
> still prefer the living world!!
>
>
> Bushi Solitario
>
>
> P.S: maybe it's only a P.R. story but I still prefer to interact in an
> RPG!

No, it is true: I saw the "real" Radiant AI used in Oblivion in an early
presentation of the game in London, and it was amazing. Then the final
version was crippled. I remember going again to London for the official
European presentation of the game, and the AI was much worse. Then, at
lunch, one of the developers told funny stories about what happened in the
game with the full AI: people attacking enemies with arrows and hitting the
wrong target - with retailations quickly escalating into full urban warfare;
guys with "duplicator wands" who duplicated both allied characters and
weapons - so that the allied characters picked up weapons *by themselves*
and marched together as an army; key chatacters being killed in creative
ways and so on. Basically, it was impossibible to maintain a coherent
storyline in the game.

Now "The Witcher" shows a much better example of "independent acting" by
NPCs, better than the one in Oblivion, but it is still not a "true
independent AI" - probably for the same reasons.
 >> Stay informed about: The_trouble_with_AI's 

Vincenzo Beretta wrote:
>> I heard that (an Example) Bethesda implemented a good AI in Oblivion.
>> Now, the story was plotted, scripted... but the NPCs began to
>> interact with each other. Chaos was the response of a good AI. Yes,
>> not a storyline to follow but a world living without much control.
>> But alive and with it's owon life. So they did implemente only a
>> "controlled amount" of AI. I still prefer the living world!!
>>
>>
>> Bushi Solitario
>>
>>
>> P.S: maybe it's only a P.R. story but I still prefer to interact in
>> an RPG!
>
> No, it is true: I saw the "real" Radiant AI used in Oblivion in an
> early presentation of the game in London, and it was amazing. Then
> the final version was crippled. I remember going again to London for
> the official European presentation of the game, and the AI was much
> worse. Then, at lunch, one of the developers told funny stories about
> what happened in the game with the full AI: people attacking enemies
> with arrows and hitting the wrong target - with retailations quickly
> escalating into full urban warfare; guys with "duplicator wands" who
> duplicated both allied characters and weapons - so that the allied
> characters picked up weapons *by themselves* and marched together as
> an army; key chatacters being killed in creative ways and so on.
> Basically, it was impossibible to maintain a coherent storyline in
> the game.
Even with the "crippled" AI I've seen a few strange interactions in
Oblivion. At the stables outside Skingrad I once watched one NPC try to
pick the pocket of another and fail, which drew in the City Guard nearby.
Last I ever saw of the would-be thief he was running over a hill with a
lynch mob in hot pursuit.

--
Because of heavy computing requirements we are currently using some of
your unallocated brain capacity for backup processing. Please ignore
any hallucinations, voices, or unusual dreams you may experience.
Please avoid concentration intensive tasks until further notice. Thank
you.
 >> Stay informed about: The_trouble_with_AI's 

On 25 dec, 00:13, pprocto....RemoveThis@yahoo.com wrote:
> Thatks amazing, Eddy!
>
> You've managed to push my buttons borth on customers' contradictory AI
> demands and unused multiplayer features in a single post !
>
>

Well, I'm not the author, so don't blame me for pushing the wrong
buttons

Ok, here are my thought about it :

I think that with all the evidence presented to us it's safe to assume
that overall multiplayer gamers represent a tiny (but vocal) minority
and that most games are played single-player versus the AI. This would
lead to the inescapable conclusion that as far a resource allocation
goes, MP support should be ditched in favour of making a better AI,
resulting in a better game for the majority of the customers.

However, there are a few niggling issues and an analogy with
boardgames is maybe in order here. Most boardgames are designed to be
played MP, yet all surveys done indicate that a majority of people
buying boardgames don't have opponents and actually play solitaire.
Reviewers have recognized this for decades and "solitaire playability"
has always been a rating criteria. Of course there have been some
solitaire board designs, some of which were even quite successful
(Ambush !) but overall it's the MP designs which have become hits, not
the solitaire ones. The virtual promise of MP games is what makes
solitaire gamers buy MP games. It's not logical, but that's the way it
is.

Then there's another thing : not many gamers play MP wargames - maybe
that's because of a chicken & egg situation where most pc wargames are
not really designed to be played MP. And by designed I mean MP being
the design philosophy from the start. It's the MP community which has
kept games like the original Combat Mission series, Close Combat,
Panzer Campaigns and other TOAW's alive, so games which lend
themselves to MP play attract MP players (duh).

Given the above I now tend to think that a developer should look long
and hard at his design and decide if the MP players out there are
going to bother with his design or not - if not, a bold move could be
to do away with MP altogether and focus on a good AI. Conversely if
you have a design that's just made for MP play you could drop the AI,
or put just a rudimentary, just-for-training-purposes AI in there.

There's a MP-only wargame, with no AI, being developed right now
(Computer War in Europe) and it will be very interesting to see how
gamers react to it.

Greetz,

Eddy Sterckx
 >> Stay informed about: The_trouble_with_AI's 

On 26 dec, 12:27, "RobP" <robertapoll....RemoveThis@hotmail.com> wrote:
>
> Lol, So I wasn't the only one playing Panzer Blitz and Squad Leader
> alone....

Nope, the original attraction of computer wargames was that it gave
you an AI opponent - this was about the only advantage they had over
boardgames.

That and the fact you couldn't get into an argument with yourself over
the exact interpretation of rule #34.5 bis

Greetz,

Eddy Sterckx
 >> Stay informed about: The_trouble_with_AI's 

<eddysterckx.RemoveThis@hotmail.com> wrote in message
news:35e66870-9814-4db4-8a47-23f533d97fb9@d4g2000prg.googlegroups.com...
> On 25 dec, 00:13, pprocto....RemoveThis@yahoo.com wrote:
>
> Well, I'm not the author, so don't blame me for pushing the wrong
> buttons
>
> Ok, here are my thought about it :
>
> I think that with all the evidence presented to us it's safe to assume
> that overall multiplayer gamers represent a tiny (but vocal) minority
> and that most games are played single-player versus the AI. This would
> lead to the inescapable conclusion that as far a resource allocation
> goes, MP support should be ditched in favour of making a better AI,
> resulting in a better game for the majority of the customers.
>
> However, there are a few niggling issues and an analogy with
> boardgames is maybe in order here. Most boardgames are designed to be
> played MP, yet all surveys done indicate that a majority of people
> buying boardgames don't have opponents and actually play solitaire.
> Reviewers have recognized this for decades and "solitaire playability"
> has always been a rating criteria. Of course there have been some
> solitaire board designs, some of which were even quite successful
> (Ambush !) but overall it's the MP designs which have become hits, not
> the solitaire ones. The virtual promise of MP games is what makes
> solitaire gamers buy MP games. It's not logical, but that's the way it
> is.

Lol, So I wasn't the only one playing Panzer Blitz and Squad Leader
alone....

>
> Then there's another thing : not many gamers play MP wargames - maybe
> that's because of a chicken & egg situation where most pc wargames are
> not really designed to be played MP. And by designed I mean MP being
> the design philosophy from the start. It's the MP community which has
> kept games like the original Combat Mission series, Close Combat,
> Panzer Campaigns and other TOAW's alive, so games which lend
> themselves to MP play attract MP players (duh).
>
> Given the above I now tend to think that a developer should look long
> and hard at his design and decide if the MP players out there are
> going to bother with his design or not - if not, a bold move could be
> to do away with MP altogether and focus on a good AI. Conversely if
> you have a design that's just made for MP play you could drop the AI,
> or put just a rudimentary, just-for-training-purposes AI in there.
>
> There's a MP-only wargame, with no AI, being developed right now
> (Computer War in Europe) and it will be very interesting to see how
> gamers react to it.

The experiences I have had with TALACOSI seem to reflect what you have
written. This game is multiplayer only with no single player game. Finding
opponents is a right royal pain, especially given that potential opponents
can download the free client and connect it to a registered owner's server.

If I'm perfectly honest apart from the odd TALACOSI game the nearest I get
to multiplayer is PBEM, the majority of my games are vs the AI.

Regards
RobP
 >> Stay informed about: The_trouble_with_AI's 

eddysterckx.DeleteThis@hotmail.com <eddysterckx.DeleteThis@hotmail.com> wrote:
>
> "The worst part of it is, that developers might sometimes not
> implement certain features of a game because it might be too difficult
> (or impossible) to program the AI for it. In other words, an aspect of
> a game that might work very well in regard to PBEM play might be
> compromised, or even left out altogether, because AI would not be able
> to adequately follow the routine.
> IMO, wargames, especially those for more than two people, should be
> designed strictly with human-to-human play (PBEM, TCP/IP, hot seat) in
> mind, with the AI added as an afterthought and mainly as a teaching/
> practicing aid and nothing more."

I'm inclined to agree, although mostly because good AI in a multiplayer
game is a lost cause anyway. In a game featuring both humans and AIs,
both groups will at least have a very different groupthink, and most
likely the humans will have much more sophisticated diplomacy and strategy
in such a game.

I would love to see really good AI in such a game, but I have no illusions
about that happening anytime soon.

On the other hand, there's a game like Galactic Civilizations that has
one human player and multiple AI players, and that works quite well.
Since there's only one human, the AIs determine groupthink and the
strategy and diplomacy remains reasonably coherent. But mixing will
never work well.


mcv.
--
Science is not the be-all and end-all of human existence. It's a tool.
A very powerful tool, but not the only tool. And if only that which
could be verified scientifically was considered real, then nearly all
of human experience would be not-real. -- Zachriel
 >> Stay informed about: The_trouble_with_AI's