Putting three instances of names as layers on top of each other with transparency effect already uncovers a large part of the most common name there, so I definitely disagree.

If you know the exact typeface and font size used, you can just try to put letters top of the image to guess them as well (so you try to match parts of the letter with one exact character to be precise). And I am pretty sure there are some tools that make this even easier (or at least made internally by some U.S. security agencies :) ).