Logically, simulators only need to include cards that might get bought. There are sometimes cards outside a strategy that are relevant due to specific actions (swindler, procession) but in those circumstances the user can find ways to add them to the simulated kingdom. More often the cards that get bought are the only cards relevant to a strategy. There is no need to fill the kingdom with any other cards.
The validity of the results gets worse when the kingdom gets more complex, essentially.
- Card play heuristics generally support simple strategies and cannot cannot see a 'bigger picture', such as when choosing cards to put on the deck after a ghost ship attack.
- Some Dominion concepts, such as end game control, are difficult to realize in one script let alone in competing scripts.
- Scripts get overly long for complex strategies and are prone to human errors.
- If the script doesn't cover enough decisions then it will be inaccurate - it's the old computer phrase of "garbage in, garbage out".
If you stick to simple kingdoms the simulator can be fine and it produced many interesting results in the past. Experienced players are often unimpressed as they can work out simple kingdoms without a simulator. Also with so many cards now available for Dominion, a kingdom you simulate is far less likely to reappear in a new game. At the time the simulator was released you could simulate fishing village and wharf and expect to play lots of games with those two cards. If you want to look back at a game you lost and ask "how did this happen" then the simulator might be able to answer.
