83 — Game: Mastermind, guessing a secret code

The guessing games so far answered only "too big" or "too small". Mastermind gives two clues at once: how many colors are right and in the right spot (black), and how many colors are right but misplaced (white). Turning those clues into correct feedback is a small piece of real algorithm thinking.

#Steps

1. A secret code is four picks from the color list — the random_pick from guide 10, called four times:

   use random latest
   
   colors = ["red", "blue", "green", "yellow"]
   
   secret = []
   for i in range(4):
       secret.append(random_pick(colors))
   

   copyrun it →

   append grows the list one color at a time, the same list-building loop that made the playlist in guide 44.

2. Black counts exact matches — same color, same position. Compare each guess color with the secret color:

   black = 0
   for i in range(4):
       if parts[i] == secret[i]:
           black = black + 1
   

   copyrun it →

3. White is trickier: right colors that are in the wrong place. The honest way counts every color that appears in both lists, then removes the black ones:

   white = 0
   for color in colors:
       white = white + min(parts.count(color), secret.count(color))
   white = white - black
   

   copyrun it →

   parts.count("red") counts one list, secret.count("red") the other; min takes the smaller — a color guessed three times cannot match a secret that has it once. Subtracting black leaves only the misplaced ones. This is the standard Mastermind feedback algorithm.

4. The full game asks for guesses, validates them, and stops on four blacks. Save mastermind.nme:

   # mastermind.nme — guess a 4-color secret code.
   # Run: nme r mastermind
   
   use random latest
   
   colors = ["red", "blue", "green", "yellow"]
   
   secret = []
   for i in range(4):
       secret.append(random_pick(colors))
   
   turns = 0
   while True:
       ask guess, "guess 4 colors (red blue green yellow): "
       parts = guess.split()
       if len(parts) != 4:
           show "please type exactly 4 colors"
           continue
       turns = turns + 1
       black = 0
       for i in range(4):
           if parts[i] == secret[i]:
               black = black + 1
       white = 0
       for color in colors:
           white = white + min(parts.count(color), secret.count(color))
       white = white - black
       show f"black: {black}  white: {white}"
       if black == 4:
           show f"solved in {turns} turns"
           break
   

   copyrun it →

5. Run it. A scripted round (the three guesses are piped in, then the game ends) against a secret of ["yellow", "red", "blue", "red"]:

   printf 'red blue green white\nblue yellow red white\nyellow red blue red\n' | nme r mastermind
   

   copy

   guess 4 colors (red blue green yellow): black: 0  white: 2
   guess 4 colors (red blue green yellow): black: 0  white: 3
   guess 4 colors (red blue green yellow): black: 4  white: 0
   solved in 3 turns
   

   copy

   The first two guesses place yellow and red somewhere wrong; the third shows every position matching. Play against your own secret by running the program without the pipe — every run hides a new code.

#Try it yourself

Count invalid guesses (a wrong color word) as a miss instead of re-asking, or add a tries limit and a "you lost" message when it runs out. Add two more colors and make the game ask for 5 — the feedback algorithm does not change at all.

#What you learned

• random_pick inside a loop builds a random secret of any length.
• Black feedback is a position-by-position comparison.
• min(count, count) per color is the honest way to count shared colors.
• Subtracting black from white leaves exactly the misplaced colors.
• A continue guard keeps invalid input out of the scoring loop.