I have seen files with say 200kb have like 1MB of size on this. Wouldn't a file with 200kb on a 4kb per block disk have, at most, 204kb of size on disk?
it is as mentioned by the others, one cluster, only one file can be there. With 200KB of files (plural is to be noted), it could take 1 MB quite easily. Each of the files have a bit of extra space, not just a bit of extra space after all the files have been counted.
taking the approach of multiple small files of 1 byte, and a cluster size of 4K (so one cluster, one file), you could have a total amount of actual data of 1MB (1 million files) and end up with the amount of space being used of about 4 TB.
ie: remove those million small files and you can place 4TB of data in it's place.
There have been a few methods over the years to address this issue, and NTFS' 4K cluster size is one of the better ones. with earlier ones like FAT and FAT32, you could set the size to what ever you wanted, up to 64KByte cluster sizes IIRC (needed a hack, otherwise it was limited to 16 or 32KBytes per cluster).
Another way was to run drive compression, but while it worked for making use of that cluster space, it required overhead and data corruption could have resulted in most of the data on the drive being lost instead of just a few files.