For instance, here's my grandmother and her family in 1901:
|10||81||Woodard Angus M||M||Head||M||Jan 25||1865|
|11||81||Woodard Sarah E||F||Wife||M||Sep 6||1869|
|12||81||Woodard Sarah E||F||Daughter||S||Aug 20||1894|
|13||81||Woodard Lola P||F||Daughter||S||Jan 14||1894|
|14||81||Woodard Charles W||M||Son||S||Aug 13||1897|
|15||81||Woodard Lydia M||F||Daughter||S||May 18||1900|
And here they are in 1911:
|7||Woodard Sarah E||F||Wife||M||Sep||1869||41||Links|
|7||Woodard Sarah C||F||Daughter||S||Apr||1892||18||Links|
|7||Woodard Lolla P||F||Daughter||S||Jan||1894||17||Links|
|7||Woodard Charles W||M||Son||S||Apr||1897||13||Links|
|7||Woodard Lula G||F||Daughter||S||Aug||1905||5|
|7||Woodard Iva B||F||Daughter||S||May||1908||3|
I think this illustrates perfectly why this can't be done automatically. Sure, Angus and Sarah E are consistent names, but can the computer recognize why Lola became Lolla? And why has Angus's birth month changed from January to February, and his daughter Sarah's from August 1894 to April 1892?
In another census year Lola and Charles are given as Pearl and Wellington, the names I knew them by, but actually their middle names. Additionally, I knew the daughter Sarah as Great-Aunt Sadie. That makes it difficult for searchers to match people up by name alone.
It becomes a little treasure hunt, finding someone related to you in one census, and then trying to track them down in the other. I'm about to do a little more hunting...