Software with heart – THE WORLD

The game Man vs. Machine is decided before it has begun. At least in chess, where the computer can calculate all possible combinations, trains and their consequences at lightning speed, where he quickly checked in databases, are considered as similar games. So why still want to play against computer chess? The frustration seems programmed.

Just before this problem are the software developers. You have to feed these powerful machines without frustrate the human buyer. The solution to the dilemma is artificial intelligence, AI. One would, so the idea of ​​developing games that can be adjusted to the respective level of the player. The software would have during the game, the player’s performance constantly check. In the best case, the idea of ​​his players, a game then even work out a little so that it is better and gradually reaches a higher level. In some cases this is already done. But in many games is lagging behind even with the AI. They support most of the players – but face far from intelligent to him

The new, 15th version of the chess program “Fritz”, the on in October. the market is, after all, has a so-called “adaptive intelligence”. The aim is to motivate the players and keep on board. “Fritz 15″ has been geared towards the needs of “normal” chess player: For more than seven million games the world’s largest with more than 45,000 tasks tactics database including solution clues was developed using artificial intelligence

In principle it comes to offer the user the greatest possible challenge with minimal frustration. He should be able to solve the tasks assigned to it, without the failure is a relevant factor in the game. To play does not mean to deny, but to learn. The special thing about “Fritz 15 ‘is that the difficulty of each task, so for example, the challenge of” mate in two moves “is specified, in a so-called Elo rating, as well as the tactical skill level of the player. This Elo rating is – similar to the golf handicap – the strength of a player on. If a problem is solved, the player wins Elo points and the position will lose points. He solves the problem only with references by the program, the dot gain is less. The failure of the players, so the position is gaining added points. Thousands of users around the world work with this tactic database and have the strength values ​​of each task order “calibrated”. This means that each user gets exactly the tasks over time, corresponding to his skill level.

The current game fair “Gamescom”, the next week in Cologne takes place, shows that many manufacturers to support the player. In the new version of Siedelungsspiels “Anno 2205″, which appears in November, city planners, for example, virtually invent the future and break this time even into space. Initially have small settlements to cities and then metropolitan As in the previous games. Commodities have reduced, driven controlling and citizens are held at the same time happy – a balancing act that demands a lot of players

Also in “Anno 2205″, the question arises. how to prevent frustration. In the “Anno” series it is always possible that economic cycles do not work and suddenly the money runs out. In the new edition of the programmers have therefore a financial injection in the game stage “simple” and “normal” installed. It supports players when they are close to bankruptcy. If the player needs help, he gets it. In addition, there are virtual assistants who warn when needs of virtual inhabitants are unfulfilled.

Even the upcoming Assassin’s Creed adventure “Syndicate” adapts to the players on. You must complete various missions in the historic London . But take a mission that does not match their level of experience or their equipment, the software alerts. She recommends retrofit initially or once to take lighter missions.

These are all supportive measures, but no real AI. This would be to implement the request of Professor Klaus Peter Jantke the Fraunhofer Institute for Digital Media Technology IDMT in Erfurt: “So far we have only loops, making the queries.” Adaptive assess the user during the game – so far, the developers have not yet. But there are the beginnings, about the racing game “Mario Kart”. Here the virtual drivers drive worse if the player drives worse and vice versa

But with real AI should the software can adapt to different approaches. It would have to be recorded if a player hires awkwardly, plays inaccurate or slow to respond. Then the game would have to be able to react in different ways. “Technically, it’s already possible today,” said Jantke. “But for manufacturers, this means a very high programming effort.”

There’s a fine line, on the move, the developers. “Raises the software tasks too obvious as well as the frustration can stir up,” says Linda Breitlauch, professor of Intermedia Game Design at the University of Trier. That means an adaptive difficulty adjustment is therefore to use rather sparse. “The player may not realize that it something is taken.” As a rule of thumb applies: “The game should always be a little more challenging and demand from anything else you feel challenged in the long run then the game is no longer fun…” The cash injection in “Anno 2205″ Breitlauch therefore looks more than one to be assigned resources wisely.

But that support may work well, prove chess programs for years. And anyone who still loses, it should be easy to keep as the Dutch chess master Jan Hein Donner, the to the question “How to Beat Schachcomputer best?” replied, “The best with the hammer.”