Revisiting Words With Fiends

I have a lot of free time on my (washed) hands whilst staying inside due to COVID-19. I noticed my boyfriend got back into playing the popular Zynga game Words with Friends.

Years ago while on my commute to work, I started dabbling with mitmproxy and monitoring the network traffic and API, but never got to the meat of solving the game.

Thinking of looking back into it now, perhaps as a sequence of interview-type problems.

Part One: Make a Move

Given a set of words, a board, and a set of N tiles, where 1 <= N <= 7 and tiles represent one character [A-Z], determine the best possible move.

public Move makeMove(String[] words, char[][] board, char[] chars)

Not necessarily a trivial problem, especially since best possible move is loosely-defined. For the sake of this exercise, we’ll assume an offensive play for now and disregard. Since I wanna get this up and running today, I will do a simple brute force solution.

Just gonna throw a quick outline of what I’d do:

• normalize and index words
• represent and relate words on board as Word objects
• find all places a Word can be placed
• return the move with highest score

Guess I’ll get started!

Part Two: Consider Letter and Word Multipliers

Given a board configuration with DL, TL, DW, TW multipliers, adjust the makeMove method to take the configuration into account.

Part Three: Connect to WWF API

Most of this is done here

Part Four: Scan Board Screenshot (Computer Vision)

Part Five: Defensive Play AI

Part Six: Will I Even Get This Far

TBD?!?