Unfair Practices
Designing good AI in a computer game is hard. Apart from a handful of cases where a simple optimal strategy is concealed beneath the game’s surface complexities, computers are notorious for poor performance in strategy games. They overreact to short-term challenges and opportunities, or overlook them. They fail to combine tactics in ways that magnify the results. They overlook exploitable loopholes. They blunder forward with a chosen strategy without fine-tuning it to their opponents’ behavior. They fall for the same tricks over and over.
Easy victories make for a bad game. But because good AI is hard, designers often decide to cheat. Computer players get free units, or cheaper goods. They peek at the map. Random numbers are skewed in their favor. Sometimes, they operate under entirely different rules. This preserves the challenge, but often at a sharp cost: it’s possible to catch the computer cheating.
Simple cheating, like lower costs or skewed battle results, is so easy to design that the option is often placed right in the startup menu. Choosing a high difficulty setting is an open invitation to give the computer opponent(s) a handicap, just as easy settings give the same artificial bonus to humans. That’s okay; players expect it, and can choose to play with or without.
Other kinds of cheating are decidedly not okay. Computer opponents in the original Warcraft, for example, did not gather and spend resources on new units, as the human player did. They went through the motions of collecting resources, but in fact simply received new units on a regular basis, independent of whether they had the gold to pay for it (or indeed, if they had collected excess resources). Destroying a computer opponent's economy was pointless. Colonization placed settlers in a race to declare independence from their respective mother countries, and make that declaration stick. A player needed at least 50% of his colonists to favor rebellion even to try; computers, possibly due to a programming error, needed only 50 rebels, a huge advantage in colonies upwards of 300 people, which could only be overcome by annexing awkward, mismanaged, unwanted colonies. Civilization would, among many, many other cheats, allow computer opponents to change production in a city under attack during an attacking player’s turn, to permit an emergency garrison to appear before the next turn’s renewed assault, something the human could not do on the defense. 1830 formed coalitions against the human(s), sabotaging human stock prices and line development even at the expense of losing—badly—while blatantly ignoring one another’s lucrative lines.
Catching a computer cheating takes a lot of the fun out of a game. If you lose to a computer, you may not be losing because you’re playing badly; you may be losing because the computer literally holds all the cards. And the dice. And the unexplored sections of the mapboard. In extreme cases, it may not just peek at the cards ahead of time, but actually draw whichever card it wants. As a corollary, when you’re winning, you must suspect it’s because the computer is letting you win. A hard-fought factory is the most satisfying outcome, and a cagey designer could program the game to give you just that, every time. Cheating makes losses frustrating, victories hollow.
Still, computer cheating is a fact of life, because smart computer opponents are so very difficult to create, and I can sympathize with designers who settle for cheating while seeking do better. Unfortunately, the fact that cheating is endemic to computer games has an unpleasant side effect, which is my thought for the day:
Even when I don’t catch a computer game cheating, I suspect it is.
That hurts the fun. Games are good mental exercise, at least during the learning curve between the point of learning to play and learning to play well. Ideally, trouble winning should be the mark of a good game, suggesting hidden depths of strategy and tactics to explore, and should inspire a flurry of interest. Instead, because cheating is so common, I first start looking for the ways computer opponents are taking unfair advantage. If I find them, I often stop playing entirely, because there’s small satisfaction in developing a strategy that should, by all rights, succeed, and there’s not much point in learning to get better at a rigged contest. And that’s a shame.