Wow. I'm actually kind of surprised that the noise generator has that little difference at the highest level across similar numbers. I mean, I guess it's not normally a problem when there are billions of billions of seeds, but I would have expected them to do something to separate out the numbers a bit more.
Like multiplying the x-value by some number to make the jumps bigger.
Or better yet, using the seed to run the RNG a certain number of times and then taking one number as the index to the x-axis and the next number as the index to the y-axis. If the random number is sufficiently large, you'll have some really different results from one seed to the next.