Meet Judge Bob, the AI judge
Somebody has to call it, and it is not going to be a popularity vote. Judge Bob looks at both drawings the moment the clock stops, scores each one on five things, and names a winner with a line explaining himself. He sees the drawings, not the handles: no ratings, no names, no idea who has been on a streak. The same rubric every duel, no hard feelings, just happy little losses.
What Judge Bob scores
Four craft axes are worth up to 100 each and add up to a score out of 400. Accuracy then multiplies that total, so drawing the right thing beats drawing a pretty wrong thing.
Accuracy
Did you draw the thing? This one multiplies everything else, so a gorgeous cat is worth almost nothing when the prompt was "lighthouse". Writing the word instead of drawing it lands here too, and lands badly.
Clarity
Would a stranger know what it is at a glance? Clean, readable shapes beat a detailed mess.
Creativity
Your angle on it. When two drawings are both plainly correct, this is usually what separates them.
Effort
How much deliberate care fits into 90 seconds. Frantic scribbling is not effort, and an empty canvas scores zero.
Beauty
Composition, proportion, line quality: does it just look good? A crude or deliberately stupid style still counts, as long as it is well made.
Is that fair, and can I argue?
Both drawings go to Judge Bob together, in an order that says nothing about either of you, labelled only as two seats. He never learns who drew which, what either of you is rated, or who has been on a tear this week, so there is no seat to prefer and nothing to be impressed by except the picture.
The verdict is final, which is rather the point of handing it to someone with no stake in it. What you get instead of an appeal is the per-axis breakdown and his one-line reason, which between them tell you exactly what to fix. He is software, and software is wrong sometimes. Rematch him.