This is about the exercise fib1_normal, which is a Randomizer exercise. In it, I have authored the following conditions: <condition xref="fib1_normal_good1"><diagnosis><achievement value="1.0"/></diagnosis><syn_eq>$M$</syn_eq></condition> <condition xref="fib1_normal_good1"><diagnosis><achievement value="1.0"/></diagnosis><syn_eq>$neg(L)$</syn_eq></condition> <condition xref="fib1_normal_near1"><diagnosis><misconception xref="LeAM_calculus_misconceptions/misconception_8"/><!-- buggy rule: expression not simplified --><achievement value="0.8"/></diagnosis><syn_eq>$lambda(z,M-z)$$0$</syn_eq></condition> <condition xref="fib1_normal_good1"><diagnosis><achievement value="1.0"/></diagnosis><sem_eval context="arith1">$M$</sem_eval></condition> where the variables are defined as following: $A$ $1/4$ $1/2$ $2$ $3$ $4$ $5$</parameter> $B$ $neg(3)$ $neg(2)$ $neg(1)$ $1$ $2$ $3$</parameter> $L$ $1/(2*A*B)$</parameter> $M$ $neg(L)$</parameter> As you see, I have even written two "equivalent" conditions, one comparing to M and one comparing to -L (which is equal to M) because I know of difficulties with unary minus. Still neither of the first two conditions applied in the case that you can see in the screenshot. Instead, condition 3 applied telling the user to "simplify" 1/12... This bug is critical since it needs to be solved for the upcoming release (evaluations).