Why are the computer players cheating?

[Eagleclaw]

I am a fast builder (using Anno 1701 A.D. version 1.01) and was surprised that the computer players overtake my development very quikly when the starting capital is 10.000. The computer players even take a pause in the beginning before their ships are moving and they start building rather late. Even so they very quikly have much more buildings than I have. By ending the game at different steps I found out that the computer players only pay 1/10 of the price for buldings. Then it is not surprising that they can build so many buildings in a game.

Another interesting cheat by the computer players are seen if you start without the free trader and without guests of honour. Each player have only 40 t of tools and can only get more by producing it or by buying it from other players. All three computer players start by buliding so many market buildings, lumber jackets and sheep farms that they cannot bulid ore mine, ore melter and tool smith. Anyway the rather quikly get ressources to build a lot of buildings and they finally add a complete line of tool fabrication.

Many of us are winning the single player games at the hard level anyway but you cannot compare this type of games to multiplayer games where all the players are following the same rules.

I suppose that the computer player cheats have been introduced into the games because the developers have been unable too write a sufficiently good program to control the computer players. That is sad in a project were so much money have been spent on silly graphics.

Does anyone have another explanation to this?

Let us hope that we get some better artificial intelligence in future versions.

[zippyriver]

Many of us are winning the single player games at the hard level anyway but you cannot compare this type of games to multiplayer games where all the players are following the same rules.

And therein lies the answer. You can't expect a game AI to give you any kind of a challange that even comes close to the level of competition you can get when playing against other humans. When you decide to play against a "hard" AI you have to expect you will make concessions.

I know how the AI [/i]feels[/i]. I used to play my uncle chess all the time (and I am not too shoddy of a player), but this guy could decimate me at any given point of time in the game. Sometimes he would spot me his whole king or queen side, yet could still mop up the board with my arse. On a level playing field, I did not stand a chance against him. To make it more challanging for him (and to help me learn) he would increase the difficulty with spots. I certainly was not cheating because I started out with more pieces, as he chose those "game settings"

If you choose to play an AI at hard difficulty levels (ANY GAME), you have to understand your giving the AI spots. Programs can't take time to ponder alternate and devious paths of action like a human can. They are very good at math, but actual strategy is beyond the current capabilities of home use hardware and software.

If you want to truely evaluate an AI's programming as it stands up against the human thought process, play at a level that every one starts out and plays on equal footing (normal).

[Betty]

P.S. Welcome to the board, Eagleclaw! :halloha:

[BaldJim]

... you cannot compare this type of games to multiplayer games where all the players are following the same rules.

I suppose that the computer player cheats have been introduced into the games because the developers have been unable too write a sufficiently good program to control the computer players. That is sad in a project were so much money have been spent on silly graphics. ...
Let us hope that we get some better artificial intelligence in future versions.

You are absolutely correct that the "Single Player Mode" with computer players can not be directly compared to the "Multiplayer Mode" with two or more human players. Why should it be? It is a home computer game, not on-line participation in game resident on a mainframe computer.

You certainly expect super-dooper artificial intelligence. Now tell us, are you a professional programmer? Many of us suspect that the developers were constrained by the limits of the hardware of the home computers owned by the market population they needed to buy the product, rather than their skill in programming AI. Is it really a trade-off between "silly graphics" and super-robust artificial intelligence? Tell us what you know about the technology, please.

:rolleyes: :cool:

[Eagleclaw]

BladJim, I do not believe the hardware to be the real problem here. Several chess programs can beat a top 100 chess grand master on a standard computer. This is not a fair comparison because chess programs has been developed over many decades. In 1977 I could beat the computer chess world champion program running on a mainframe even when I played blind folded. Today I do not stand a chance.

Zippyriver. Yes computers are good at math and that is why they should have a natural advange planning the city layout and the production processes. This is a perfect task for computers, just take a look of some of the analys made on the German Community Board here. Computer analysis of production prices, sales prices and taxes could be done knowing the exact values used in AD 1701.

Your comment "but actual strategy is beyond the current capabilities of home use hardware and software" is clearly incorrect. Just look at computer chess. Computers are better at tactics than strategy in chess but the situation is different in a game like 1701 where pure math should give the computer player a tremendous advantage without cheating.

[Rnett]

Lately it's been hard to find a game that doesn't cheat.

[BaldJim]

... Several chess programs can beat a top 100 chess grand master on a standard computer. This is not a fair comparison because chess programs has been developed over many decades. In 1977, I could beat the computer chess world champion program running on a mainframe even when I played blind folded. Today I do not stand a chance.

... computers are good at math and that is why they should have a natural advange planning the city layout and the production processes. This is a perfect task for computers, ... Computer analysis of production prices, sales prices and taxes could be done knowing the exact values used in AD 1701.

... Just look at computer chess. Computers are better at tactics than strategy in chess but the situation is different in a game like 1701 where pure math should give the computer player a tremendous advantage without cheating.

Did you in 1977 beat the world champion computer chess program playing blindfolded? For that matter, how did you do with human opponents?

In 1977, there was hardly anything other than "main frame" computers; and they really did not have time to fool with chess. Game players had to punch a fist-full of cards to input a single move.

Put simply, I do not accept your falling back on discussions of chess as a test of computer hardware as compared with games which have a serious visual component. Oh, pardon me, "where so much money have been spent on silly graphics".

Just record me as one who finds chess and ANNO 1701 to exist in two different computer worlds.

:rolleyes: :sad:

[Eagleclaw]

Well maybe you should find the facts before you comment on them. We did not punch cards in 1977 but had nice keyboards and text terminals. The programmers did not fool with chess but found the development of programs to be an important research area for artificial intelligence. You cannot be taken seriously stating: "In 1977, there was hardly anything other than main frame computers". Have you ever heard about Apple computers or Microsoft? They were founded in 1975. Other personal computers in 1977 were TRS and Commodore PET. You obviously do not know your subject.

This is not relevant anyway. The point is that it is disappointing that the computer players are cheating. It has some important influences on the single player game. You cannot make a blockade and prevent the computer player getting tools. The computer player cheats and creates the tools from nothing. I find it even more disappointing because it should be fairly easy to write a good program as explained earlier.

[achkarou]

I'm not a computer wiz but i belive that in a game like ANNO the largest requirements for computer hardware comes from the graphical aspects of the game (and i don't think they are silly because i woudln't play the game if it was an ugly monochromatic 2D version). For example if you turn down the graphics options to lowest the game performs much better than on high settings (at least for my pc).

Therefore, i don't think that enhancing the "brain" of the AI's woudl put a significant strain on the performance since the processing of "numerical" logistic information is far less demanding than graphical enhancements. So the developers were probably a bit lazy to create smarter AI's and the company was probably in a rush to get the game tested and released! And that is why there are many obvious drawbacks to the game.

In one of my threads about production/consumption values there are more obvious downfalls to the realism of the game and all could be EASILY resolved by a group of knowledgable game software programmers and beta testers without beefing up the PC hardware requirements, including more different goods and services and realistic values for things.

[Rnett]

This is not relevant anyway. The point is that it is disappointing that the computer players are cheating. It has some important influences on the single player game. You cannot make a blockade and prevent the computer player getting tools. The computer player cheats and creates the tools from nothing. I find it even more disappointing because it should be fairly easy to write a good program as explained earlier.

It seems to be the industry standards, they have 18 months to turn around a product and most of the emphasis is spent in graphics.

[zippyriver]

Just to clarify, my comment regarding playing chess with my Uncle had to do with accepting concessions for a greater challange, and NOT a comparison between AI chess and 1701.

The two are not even in the same realm when it comes to programming (and no, I am not a programmer. If you want to disregard my comments based on that, be my guest). Chess, really, is not that hard. If Anno were chess, the map you play on would be microscopic compared to what it is now. You also would not have to worry about multiple opponents, nor railbirds assisting your opponent (colonial factions). You would not have to create a support system to pay for your various units and tiles you occupy. Units would not be able to have freedom of movement to defend, patrol or attack in any direction, nor would you have to worry about the same. I could go on, but I hope it makes the point. 1701 has many more variables than chess, and every single dot and tittle requires very specific lines of code. To have an AI in a game like this be made well enough to give even a basic player a challange requires a lot of code. But no where near the amount of code required when you start talking about advanced AI algorithms that try to predict what a human player might be planning, and to determine if that is a ruse or an actual attack, and to act accordingly.

I don't think people want that anyway. What fun would a game be if the AI could deftly beat you every time. When an AI makes a mistake, or shows "poor planning" in it's development, it is creating an a situation the human player can exploit and progress towards victory. It also has to be able to determine when it should exploit mistakes the human player makes, yet still maintain a level of fun for the player.

So to compensate for the lack "intelligent" code, the AI at the most difficult levels gets big bonuses. Selecting a hard level is more along the lines of saying "I want to make it easier for the AI", than "I want to make the game harder for me".

That being said, I agree with the point you state that you would like to see a better AI. But I believe obtaining that goal is LOT harder than most people think.

I am not completly sure what your biggest complaint is though. In your OP, you focused mainly on the bonuses (cheats) the AI gets. But in your reply you talk about city lay-out and production chains (and the shortcommings the AI has in those areas). Indeed, the code required for sane city layouts and production chains would not be that difficult. The two areas of topic are quite a bit different.

On the points in your reply, I offer the following thought. Yes, the AI seems like it would have a hard time folding an envelope and I would love to see some improvement in that area. But I don't want the AI to be completly flawless in it's designing as there would be less areas to exploit, and it would totally kill trade. They don't buy tobacco if there is not a shortage of it. The other perspective is, take a look at the number of posts and activity trying to determine the most efficent chains and lay-outs. The players don't stop playing the game even when they shut it down and get online. The entire spectrum of those determinations is extensive, and a hell of a lot of fun. If the AI made perfect placements for all to see, all that investigational whoopla and fun would be removed.

Yes, I want to see a (little) smarter AI, but I also want the game to maintain it's fun and re-playability. Just my $0.02

[Reka]

The two are not even in the same realm when it comes to programming (and no, I am not a programmer. If you want to disregard my comments based on that, be my guest). Chess, really, is not that hard. If Anno were chess, the map you play on would be microscopic compared to what it is now. You also would not have to worry about multiple opponents, nor railbirds assisting your opponent (colonial factions). You would not have to create a support system to pay for your various units and tiles you occupy. Units would not be able to have freedom of movement to defend, patrol or attack in any direction, nor would you have to worry about the same.
It really depends on which part the game we're talking about. The actual battle is very hard to master for a computer program, I agree; how to optimally move war ships & troops around is not easy to see. But even during the simple building part the NPCs don't behave very smart, when taking decisions of the kind "when to built which part of which production chain". As with chess, algorithms for such things are not that easy to figure out, but as in the case of chess, this has been done (e.g. as some variant of integral linear programming in this case). Such algorithms have been developed for decades. Still the NPCs do some very poor business decisions; I described two very clear ones of them in these threads:
http://forum.sunflowers.de/showthread.php?t=16873
(NPCs selling whale & buying ambergris)
http://forum.sunflowers.de/showthread.php?t=16727
(NPCs not building palaces on their main island)
The threads are in german, I'm afraid, sorry for that...google translator is your friend.

1.0.2 is supposed to improve the NPCs behavior, let's see about the changes it brings.

[Eagleclaw]

Chess, really, is not that hard. If Anno were chess, the map you play on would be microscopic compared to what it is now.

I would say Anno is really not that hard. Any kid can play a rather good game after playing it a month or two.

The games are obviously different but in a way that shuld favourite the computer in the playing phase. The similarity between chess and Anno is that the posibilities are so many that the computer cannot evaluate the outcome of all actions. Even the solutions with only 8 pieces on the chessboard has not been calculated and put into tablebases yet. It can take 24 hours to calculate a solution for 7 pieces.

All the things you mention as difficult are relly things that is easier for a computer to figure out fast.

Calculating building layouts, production chains, trade routes and a trading strategy should be very easy for the computer. Also the computer has no trouble following all opponents military movements and building actions.

I do not agree that better computer play would destroy the fun. I think it would be more fun. Now you know that you have won when you get copper plates on your ships.

I am not completly sure what your biggest complaint is though. In your OP, you focused mainly on the bonuses (cheats) the AI gets. But in your reply you talk about city lay-out and production chains (and the shortcommings the AI has in those areas). Indeed, the code required for sane city layouts and production chains would not be that difficult. The two areas of topic are quite a bit different.

You ask which "bonus" I find most annoying. The worst thing (so far) is that the computer player can generate tools without producing it or trading it. Maybe that "maigic" is also used on other goods meaning that you cannot us blockade to harm a computer player.

I fail to see how that makes the game more fun or replayable.

My comment about the city layout, production chains and so on is just a reply to the statement that the reason for the cheats is the difficulty in writing a good program controling the computer players.

[BaldJim]

:offtopic:

Well maybe you should find the facts before you comment on them. We did not punch cards in 1977 but had nice keyboards and text terminals. ... You cannot be taken seriously stating: "In 1977, there was hardly anything other than main frame computers". Have you ever heard about Apple computers or Microsoft? They were founded in 1975. Other personal computers in 1977 were TRS and Commodore PET. You obviously do not know your subject. ...

I'm sorry you felt it necessary to ridicule me. I was brusque in my comments about your complaint because I have been listening to that kind of complaint for several years now. They typically come from folks who regard the computer as a magic "black-box."

Anyone can read about the history of the development of the personal computer here:
http://en.wikipedia.org/wiki/Personal_computer

My personal experience was that punch cards were the most efficient input device available in my work environment. The "dumb" terminals with keyboard and video were for the operators. We were also using remote data capture terminals which used tape cassettes and telephone dial-up. Other outfits had been for years using paper tape to capture data to be transferred to a central facility by mail. A keyboard and video display did not a desktop computer make.

I was a subject matter expert, not a computer "professional." I still suggest that the personal/home computer was the province of the super geek or the super rich until the early 1980s. It appears that we had different work experiences. Let's leave it at that.