Obviously Patrick met the requirements of the loosely worded challenge.
Though I wonder if the inverse would be acceptable. Take 100 files all the same size, but smaller than the filesystem block size. These files are n100 bytes, but take up blocksize100 bytes on the disk. Then, using tar (which itself does not compress, just tacks files end to end with metadata in between), 'compress' them into a single file that takes up fewer disk blocks than before.
Though I wonder if the inverse would be acceptable. Take 100 files all the same size, but smaller than the filesystem block size. These files are n100 bytes, but take up blocksize100 bytes on the disk. Then, using tar (which itself does not compress, just tacks files end to end with metadata in between), 'compress' them into a single file that takes up fewer disk blocks than before.