A Lesson for Matrixgames. ;)

[ravinhood]

http://www.gamespot.com/news/6139939.html

I particularly liked this line:

"There is a blend of ways to get products out to consumers. The retail market is definitely the dominant way in for people to buy games currently, and that's going to continue for many, many years. It's really important, and we're not going to lose sight of that. "

Maybe Matrixgames can talk to this ELAVATION group, they got big bucks to burn. $300 Million dollars holy moly. Bioware/Pandemic gonna be sitting pretty for years.

[rhondabrwn]

ORIGINAL: ravinhood

http://www.gamespot.com/news/6139939.html

I particularly liked this line:

"There is a blend of ways to get products out to consumers. The retail market is definitely the dominant way in for people to buy games currently, and that's going to continue for many, many years. It's really important, and we're not going to lose sight of that. "

Well, if you are talking console games, they are definitely correct, but as someone commented in another thread, their local EB store is pulling PC games off the shelves for good. Ultimately, if you are any kind of PC gamer, you'll be getting your games online in some fashion (boxed by mail or direct download). I don't have a problem with that scenario.

For what it's worth, I buy nearly everything online these days except groceries and gasoline. This has bee true for at least the last 5 years and I've never had a problem with a compromised credit or debit card. I'm more concerned about someone intercepting a credit card offer and setting up an account in my name than I am concerned that buying a Matrix download would get me in trouble.

Identity theft is what scares me and EVERYONE should be lobbying Congress to put restrictions on accumulation and distribution of our personal data all over the Internet. That is what we should all be worrying about, not about whether you have to order games online with Matrix because they don't support retail channels.

[ravinhood]

Heh I never would have guessed you'd be first to condemn this Rhonda, but, I likes you anyways.

It was also more about that $300 million dollar backing that Bioware and partners got. Mano Mano if Matrixgames could get that kinda backing imagine the wargames we'd get then.

[pasternakski]

Ravinhood, old pal, I can't agree that more money brings better games. More money comes from sources that are studied by marketing experts and exploited by publishers. Wargamers aren't a group that will bring in the "big, big money." Never have been.

Besides, the primary obstacle I see between computer wargamers and better games is the lack of effort being put into development of more sophisticated and effective approaches to AI. Don't forget that the promise back in the 1980s was that computer games would bring you a satisfying game when you didn't have a human opponent. So far, this promise has been broken by the business, with no sign that anyone will invest the resources needed to make an improvement. AI is a hot area, but not among computer wargaming designers. Why, I don't know, as this is where whatever money there is to be made in computer wargaming will be made. TCP/IP and PBEM capabilities are great, for those who want TCP/IP and PBEM capability (me included). Unfortunately, unless the sun starts coming up in the northwest every day and Tom Cruise learns how to act, nothing will change, and computer wargaming will die again, perhaps this time never to re-emerge from the crypt and say, "Hey! I'm Lazarus! I stink, but I'm alive!"

Note carefully that surveys dating back to the old SPI and AH paper-and-cardboard days indicate that 70 percent plus of wargames of all kinds are played solitaire. You lose that significant a segment of the market, and you may as well kiss computer wargames goodbye one more time, and for the last time. We have recently seen a mild resurgence in this business, with Matrix playing a crucial part, but, if it dies again, it's hello "Age of Empires" for life.

The problem is a dearth of effort being expended in the right area. And, as Dearth Effort could tell you, "The Force has two sides, and the Sid Meier side is the stronger."

[Shannon V. OKeets]

ORIGINAL: pasternakski

...

Besides, the primary obstacle I see between computer wargamers and better games is the lack of effort being put into development of more sophisticated and effective approaches to AI. Don't forget that the promise back in the 1980s was that computer games would bring you a satisfying game when you didn't have a human opponent. So far, this promise has been broken by the business, with no sign that anyone will invest the resources needed to make an improvement. AI is a hot area, but not among computer wargaming designers. Why, I don't know, as this is where whatever money there is to be made in computer wargaming will be made. TCP/IP and PBEM capabilities are great, for those who want TCP/IP and PBEM capability (me included). Unfortunately, unless the sun starts coming up in the northwest every day and Tom Cruise learns how to act, nothing will change, and computer wargaming will die again, perhaps this time never to re-emerge from the crypt and say, "Hey! I'm Lazarus! I stink, but I'm alive!"

Note carefully that surveys dating back to the old SPI and AH paper-and-cardboard days indicate that 70 percent plus of wargames of all kinds are played solitaire. You lose that significant a segment of the market, and you may as well kiss computer wargames goodbye one more time, and for the last time. We have recently seen a mild resurgence in this business, with Matrix playing a crucial part, but, if it dies again, it's hello "Age of Empires" for life.
...

Age of Empires for life would shorten my life immediately.

AI is hard to write. The programming courses in the universities aren't teaching it. Indeed, they teach a mishmash of programming techniques without much understanding of what it really takes to write functional code. Here I am referring to their focus on teaching how to write parsers and compilers with recursion front and foremost. This is similar to back in the 1980's when the AI they were teaching was Prolog and Lisp. Stretching my imagination for an analogy, I come up with teaching English by focusing on words that contain Q because Q is such an interesting letter. Words that end in 'at' or 'ate' are too boring to discuss, so they are never included in the curriculum (sp?).

My design document for the AI opponent for MWIF is up to 78 pages and I expect it to be close 200 pages by the time I am done. And that is before writing a single line of code to implement the design. There will be a parser and mini-compiler in the code but it will be less than 1% of the total and I could even implement the design without those if push came to shove.

One of the hardest things about creating an AIO is that the computer opponent is essentially blind. While we can look at a position and draw conclusions instantly, the AIO plays by braille. It has to 'touch' every unit and finger each hex between the unit's current hex and where it wants it to go, to see if the way is clear. If its finger gets wet, then it knows it is crossing a river. To get a feel for what I mean, try playing your next war game blindfolded. Better yet, start a new wargame that you have never played before and try playing that without looking at any of the pieces or map - ever. I can play chess without looking at the board (poorly), but even the simplest board game would be way beyond my ability to visualize what is going on during a game.

If you are going to put a lot of money into developing the AI (and please do, send me a bucketful), then the staffing would have to be people who can play the game extremely well plus people who can write functional code extremely well. In the AI community this is known as developing expert systems (or at least it use to be - jargon changes frequently) where the smallest team consisted of an Expert in the subject matter and a AI Researcher/programmer.

Anyway that's my 2 cents worth (all I can afford).

[Veldor]

ORIGINAL: pasternakski
Note carefully that surveys dating back to the old SPI and AH paper-and-cardboard days indicate that 70 percent plus of wargames of all kinds are played solitaire.

My own simple survey of 450+ wargamers shows that number might even be much higher now (haven't actually processed all responses yet as I'm waiting for at least 500) but it is among a few questions that only require a casual look through the responses to note the trend is definitley overwhelmingly "Solitaire first followed by either PBEM or Network Play".

AI is a hot area, but not among computer wargaming designers.

Can't speak for other designers/developers, but my companies Name, Logo, and Slogan were all chosen around an A.I. theme. If that doesn't stress that A.I. is the priority, then I don't know what would. I personally prefer solitaire only play, but am constantly forced into network or PBEM because the AI is too easy. But, often in my case, online or PBEM is too hard This is mostly because I play too many different games to ever get as good as someone who plays one day and night. Really I just want an AI in games that is just good enough to be challenging, or can grow with me as I get better. And I've always felt a large number of other wargamers were in the same boat.

I'm wondering if the A.I. isn't one of the last things looked at with many wargames. Some games would appear as if they made the entire game minus AI. Worked out most of the bugs, and then started full force on the AI implementation. IMO that is one of the first and easiest mistakes to correct. I believe planning and integrating the AI much much earlier on helps achieve better results.

There are also other models for A.I. then what has traditionally been used. Some of the more interesting ones would involve Server-Side processing (a bit similiar to some Business AI Models and lets face it.. many wargames have a lot in common with business apps. lots of data, lots of statistical processing by the AI, etc.) but the main problem would quite possibly be some buyers reluctance to be forced to be connected to the internet in order to play.

It happens to beautifully solve piracy issues (along the lines of whats achieved with MMPORGs), but if you could have a kick-ass AI but had to be connected to the internet to get it, would you?

I'm also wondering, should say 475 out of 500 people have chosen solitaire as their first preference, if it wouldnt be prudent for some developers out there to simply forget about PBEM and Network Play altogether and just simply release a single-player only game with an AWESOME AI.

I think the problem there is that most reviewers would just trash the game for lack of what is considered standard feature (Network or PBEM).

Theres really no great answer. I think there is just this acceptance by wargame developers that the AI is not going to be all that good... and its better to spend extra effort on the multiplayer aspects which is, survey says erronously, currently viewed by companies as the ultimate and preferred experience for the wargamer....

[ravinhood]

That's a very good explanation Shannon, thanks for that. Enjoyed reading it.

[ravinhood]

Having played various solo games of various types, over the years I've come to find AI's of different quality. What I don't understand is why can't some developer take the "best" of the best AI's and put them into ONE?

For instance Slitherines Spartan AI does a fantastic job of building units and making a beeline for the human player usually by 4 different factions. But, it's flaw is it makes a beeline for one particular city without deviation, banging its head along the way on anything that gets in its path and by the time it gets to the city it's decimated or the city is so well defended it doesn't stand a hogs in hot water chance of surviving. While on the other hand the ole Warlords III AI would get to it's destination and if it saw it was outmatched, it would move onward to another city less protected, but, it's flaw, it never protected it's leaders, treated them just like another unit and didn't build quality stacks of various units and left itself very weak on defense. The flaw in practically all of them is lack of any defensive building within their structures. It's all out war or nothing for the most part. Thus, makes it easy for a player to storm their rears or middle and wipe them out piecemeal. There needs to be a double AI, an offensive one an a defensive one and a weighting system imposed where one becomes more important than the other as the situation arises.


So, I'm just asking anyone that designs AI's why can't you incorporate these types of AI's together. One for building and knowing the human player is the objective and the other with some common sense to know not to keep banging it's head on the same brick wall? And then another that knows the leader/hero is the most important unit and will do it's best to protect him/her/it in the best stack possible? And then another that prepares a defense that is sound and not just militia units that are easily overrun?

From playing all these different games, the best AI's have most often been in these provincial/city structured games. Hardly any hex based hex by hex games. I can definitely see the complications there. But, when it's provincial or cities only as most important, then the best AI's have shined. Even CIV IV's AI is much improved as far as that goes. I see so many games that come close, but, just never finish the AI equation. They have one or two of the four main aspects and at least make the game challenging. But, I'd really like to see someone take the chance and develop and AI that is programmed to WIN, not just be a roadblock on the path to victory for the human player.

Heh I have to laugh somewhat though here. Slitherine had to make a patch to "WEAKEN" the AI of Spartan because of the whines they got that the AI was TOO HARD! hahaha, finally get a game with a hard AI and people whined. I guess serving the masses of consumers is the most important thing. And most are pretty bad at playing strategy games it looks like. So, those of us who enjoy a challenge don't see it too often. Games are made so people WIN and win easily for the most part, so they can rush to the forums and brag how they beat it on Deity the very first try. Yeah that already happened with CIV IV lol (played on a teenie tiny map against 2 computer AI's that only got 3 cities each. lol)

What we need again is a good ole fashioned "basement" programmer and developer who's more interested in making a quality game than $$. I think those old days of the early 80's are long gone.

[Zap]

If you are going to put a lot of money into developing the AI (and please do, send me a bucketful), then the staffing would have to be people who can play the game extremely well plus people who can write functional code extremely well

Well, if I was a very rich man, I think that I would be interested in funding you to realize your vision. Your presentation conviced me Shannon.[;)]

[Shannon V. OKeets]

ORIGINAL: ravinhood
Having played various solo games of various types, over the years I've come to find AI's of different quality. What I don't understand is why can't some developer take the "best" of the best AI's and put them into ONE?

The fatal flaws you see in AI opponents are their weakest points. It is like a tire with a pin point hole in it. The tire goes flat. Humans beat on the AIO until they discover a fatal flaw and discount all the work that went into the parts of the game the AIO plays well: "Bad AI, they didn't try." Still, I agree that you can't sell somebody a tire that always loses air. Or at least you shouldn't be able to.

For complicated (i.e., strategic) wargames there are three levels that have to be played well: tactical, operational, and strategic. I have a very good friend who crushes me tactically (he is a grand master at chess) but with board war games I can beat him [roughly] 51% of the time by playing a better game operationally. Strategically, we are about even. If you do not frame the problem to include those 3 levels, the AIO is more or less doomed to be weak.

[sol_invictus]

Ravinhood, I'm sure that you realize that the reason that people complained about the "brilliant" AI in Spartan was that people soon realized the computer started with a mountain of material and several factions just piled on the human with never ending waves of armies. Difficult yes; fun no. I like Slitherine games but I'm afraid that they have not yet solved the great AI dilema.

[ravinhood]

Funny, I didn't have any real problems with those issues playing Spartan and I played on much higher difficulties than "easy". Must just be people were not very good strategists. Just more whinners! The win was all in the tactical setup, so, if they suked at setting up tactically that's not my fault or Slitherines fault, it's their fault for being stupid. I learned to overcome, sure it was hard "at first", but, being persistant makes perfect. That's what I meant earlier by most games built so people can WIN, people don't want to LEARN a game, they just want to open the box and WIN WIN WIN. PFFFT on that. Let em suffer I say.

[Captain Cruft]

The problem with most wargame "AIs" is that they're not, as far as I can tell. Generally you just have a bunch of code cobbled together to produce some hopefully relevant activity. Most wargame developers seem to be amateur and/or self-taught programmers and do not have the computer science background to even attempt to do the job properly. Not that there is anything wrong with that.

"Shannon" here is an exception, as is obvious if you read the "blind me with science" posts on the WiF forum. That is recommended reading for anyone BTW, even if you don't understand half of it, like me

Even where the programmer(s) may have the background and skills necessary to attempt a real AI, it generally isn't going to happen for a consumer game. Reward < Effort required. This is a very small market with virtually no competition, where is the incentive?

[ravinhood]

Reward < Effort required. This is a very small market with virtually no competition, where is the incentive?

Because when word gets around that a game "does in fact have a good/great AI" it makes a difference in sales and the following of that company. Take Mad Minute for example. An unheard of little independent of a couple of guys and some testers. But, now nearly everywhere I go I hear something about Civil War: Bull Run and Mad Minute.

I'm standing pockets wide open waiting on 2nd Manasses (due sometime in December). It's going to be around $45, and I'm gonna break my rule on "waiting for 6 monhts" because I've seen these guys work. I've seen their AI and I've also voiced it around the web and to all my aquaintances and friends.

These guys have proven you can take time to make a "challenging" AI, maybe not BLUE or whatever it's called, but, certainly 10x or even 100x more challenging than most of the crap we see all the time released with pretty pictures and sorry AI's.

But, with Matrixgames releases I sit with "I told you so" whispering on my lips as each new release comes out and I start reading about the gameplay of their games and their AI's. Tin Soldiers Caesar after "patching" has a pretty decent AI. Have to give them credit where credit is due.

When you have companies like Slitherine and now Mad Minute that can in fact make a pretty darn efficient AI, you start to expect that from everyone. I do.

BTW if you haven't played Mad Minutes Civil War: Bull Run Take Command game, I suggest you try it out. It was like $20 bucks brand new, so, probably less than that now. I think Gogamer has it, you can use my link.

http://www.gogamer.com/cgi-bin/GoGamer. ... ID/Catalog

[Veldor]

ORIGINAL: Captain Cruft
Most wargame developers seem to be amateur and/or self-taught programmers and do not have the computer science background to even attempt to do the job properly.

Hmm, I'm assuming that statements based on opinion and not known facts. I'd actually come to somewhat of the opposite conclusion, though perhaps with one important caveat. Most wargame developers I've known or come across are very much professional programmers with a lot of industry experience. The caveat is that its usually not in the Games Industry but rather in Business. Now I, perhaps wrongly so, believe Wargames do in fact have more in common with many Business Apps than they do with more traditional Games. The problem, as I see it, is that the A.I. has been dealt with from a traditional Game standpoint, and not as you would for business A.I. etc.

But yet, even then, I see developer after developer talking about next gen solutions and A.I. models. Even ones whose current games aren't considered to have great A.I.'s. It's not purely a question of skill then really, its more a question of time, priority, and resourcefulness to some extent. It's quicker to make a traditional A.I. because there are already many roadmaps to follow for this.

Check out this link by one current wargame developer on his thoughts towards using a more Data Mining approach to wargame A.I.:

http://www.itsc.uah.edu/till/index.html

[Captain Cruft]

Yes of course it is all opinion. I am just on the outside looking on in ignorance ...

Tiller is one of those I would have suspected of being a professional programmer. The AIs for most of his consumer games still suck though TFB (the latest game) is supposed to be somewhat better.

What it boils down to is that I personally can't see the satisfaction in playing against the lump of metal that is a computer, however good or bad it is. Paul Sinatra (Steel God) over at the Blitz has a very good analogy which I can't really quote here, it's rude but absolutely on the money.

[Erik Rutins]

ORIGINAL: Captain Cruft
Most wargame developers seem to be amateur and/or self-taught programmers and do not have the computer science background to even attempt to do the job properly. Not that there is anything wrong with that.

I can say that at least where our developers are concerned, this could hardly be further from the truth. Did you have any basis for this?

Even where the programmer(s) may have the background and skills necessary to attempt a real AI, it generally isn't going to happen for a consumer game. Reward < Effort required. This is a very small market with virtually no competition, where is the incentive?

Some developers are more concerned with AI than others, some are better at it than others. People often also underestimate just how closely AI complexity is tied to design and game complexity. All games are not alike in terms of the challenge of giving them a great AI and feature lists and customer wants often create trade-offs at the design stage in terms of AI competency. Looking around at wargames on the market, I'd put a number of our titles up against anyone else's.

Ravinhood, if I understand correctly, you think that only TS: Julius Caesar among our games has a good AI? Wow, I really don't know what to say to that. Yes, TS:JC has a great AI but so do many others. I would say (in my opinion) that on average our developers write better AIs than anyone else. We have some of the best wargame AI around in a number of our titles and I'll gladly put them up against anyone else's games.

Regards,

- Erik

[ravinhood]

Ok put them up against Mad Minutes AI and Slitherines Spartan AI. I'll acept the challenge. Send me all your games and I'll test out the AI.

[Erik Rutins]

I've played both and (again, in my opinion) those two rank among the best. Not taking anything away from them when I say that we have titles with AI that is just as good. I encourage you to do your own testing as I've already done it here. [;)]

As a side note, we do publish Spartan, so in holding it up you've agreed that one of our published titles has great AI. [8D]

Regards,

- Erik

[ravinhood]

That title was published long before you picked it up Erik and you know it. It was out nearly a year if not a year. You didn't initially publish it so why don't you fess up the truth like a good man.

Also where's Legion Arena? You gonna publish Legion II on initial release?

Also, I'd really be interested in knowing, you're own personal opinion of course what Matrixgames titles you feel match Mad Minutes & Slitherines Spartan AI? And don't say ALL of them cause that will give me a less opinion of your "honor".