By having people post in the topic, you're letting your participants be influenced by subconscious suggestion and therefore messing up the controls on your experiment.
Also, you're sampling from a community that is almost 90% males. It's a well known fact that the eyesight of women and men differ, and age is also a factor too. The Blockland Forums is not a representative population to sample from.
Your images are also different sizes, which could possibly impact which ones participants guess correctly.
You may want to completely scrap your experiment and find a better place to sample from.